Findings From the WildlandFirefighters Human FactorsWorkshopImproving Wildland Firefighter PerformanceUnder Stressful, Risky Conditions: TowardBetter Decisions on the Fireline and MoreResilient Organizations

The views expressed in each paper are those of the authors and participants and notnecessarily those of the sponsoring organizations of the USDA Forest Service.

The Forest Service, United States Department of Agriculture, has developed this informationfor the guidance of its employees, its contractors, and its cooperating Federal and Stateagencies, and is not responsible for the interpretation or use of this information by anyoneexcept its own employees. The use of trade, firm, or corporation names in this publication isfor the information and convenience of the reader, and does not constitute an endorsement bythe Department of any product or service to the exclusion of others that may be suitable.

The United States Department of Agriculture (USDA) prohibits discrimination in its programson the basis of race, color, national origin, sex, religion, age, disability, political beliefs, andmarital or familial status. (Not all prohibited bases apply to all programs.) Persons withdisabilities who require alternative means of communication of program information (braille,large print, audiotape, etc.) should contact the USDA Office of Communications at (202) 720-2791. To file a complaint, write the Secretary of Agriculture, U.S. Department of Agriculture,Washington, D.C. 20250, or call (202) 720-7327 (voice) or (202) 720-1127 (TDD). USDA is anequal employment opportunity employer.

Ted PutnamProject Leader

Workshop sponsored by USDA Forest Service,Fire and Aviation ManagementJune 12–16, 1995, Missoula, Montana

Technology & Development ProgramMissoula, Montana

TE02P16–Fire/Aviation/Residues–Tech Services

November 1995Updated July 1996

United StatesDepartment ofAgriculture

Forest Service

Technology &DevelopmentProgram

5100–F&AMNovember 19959551-2855-MTDCUpdated July 1996

2

Part 1 of 4

Part 1 of 4Background _____________________________________ 3

Part 2 of 4Workshop Overview ______________________________ 4

The Start _____________________________________________ 4High Reliability Organizations: A Vision for Fire Reorganization ___________________________________ 5Using the Crew Resource Management Model in Fire __________ 6Assessment and Feedback _______________________________ 7

Workshop Findings ______________________________ 8Fire Organization and Culture _____________________________ 8Fire Management, Incident Management Teams, and Fire Crews in a Crew Resource Management Context _______ 10Assessment and Feedback ______________________________ 16

Part 3 of 4Discussion _____________________________________ 18

Recommendations ______________________________ 19

A Final Note ____________________________________ 21

References _____________________________________ 22

Part 4 of 4Appendix A—Overview __________________________ 23

Appendix B—Agenda ____________________________ 24

Appendix C—Participants ________________________ 26

Appendix D—Keynote Presentations _______________ 30Addressing the Common Behavioral Element in Accidents and Incidents _______________________________ 30Naturalistic Decision Making and Wildland Firefighting ________ 33Cultural Attitudes and Change in High-Stress, High-Speed Teams ___________________________________ 36South Canyon Revisited: Lessons from High Reliability Organizations _______________________________________ 42

Appendix E—Related Reports _____________________ 54The Collapse of Decisionmaking and Organizational Structure on Storm King Mountain _______________________ 54The Collapse of Decisionmaking and Organizations: The Mann Gulch Disaster ______________________________ 59

Contents

3

Part 1 of 4

II t has become increasingly clearthat wildland firefighters areexperiencing collapses in

decisionmaking and organizationalstructure when conditions on the firelinebecome life-threatening. Since 1990wildland fire agencies have lost 23

people who might have survived hadthey simply dropped their tools andequipment for greater speed escapingfires. We are averaging more than 30entrapments each year now. And duringthe 1994 fire season, 34 people died,14 on the South Canyon Fire alone.

These facts tell us that firefightingorganizations, crews, and individualsneed to be much more proficient atdecisionmaking under stressful, riskyconditions. Improving proficiency willrequire new training and attitudechanges. And this in turn requires athorough examination of the humandimensions of wildland firefighting. Thisexamination is not limited to firefightingcrews and teams (i.e., smokechasers,engines, helitack, incident management,type I, and type II) but extends to firemanagement officers, dispatchers, firesupport, managers with fire and resourceresponsibilities, up to Agency heads.These people encompass a firecommunity. Fire community implies anawareness that we are interconnectedand interdependent and shouldapproach firefighting from the point ofview that we are all in this together.

To begin to address some of the humanfactors questions, experts in psychology,sociology, organizations, fire safety, andwildland firefighting attended a 5-dayworkshop in June 1995 to discuss waysof improving firefighter safety.

Workshop participants exploredfirefighter psychology, interactionsamong firefighters and among fire crews,and better ways to organize. Afterseveral days of discussions, theydeveloped a series of recommendationsfor beginning to implement changes thatwould improve the fire organization andfirefighter safety.

This paper outlines the workshop’sfindings and recommendations. Theworkshop represents a first step in whatwill be a long journey toward a betterunderstanding of the human side ofwildland firefighting.

Ted PutnamWorkshop Organizer

The main entrapment site at the South Canyon Fire where 12 firefighters lost their lives on July 6, 1994.

Background

-End Part 1-

4

Part 2 of 4

The Start

FFrom the beginning, workshopemphasis was on people, notfire. It was about peopling fires,

not fire suppression. With the former,we organize trained people to performa task safely and efficiently, and therelevant task is fire suppression. In thelatter, we suppress fires using people.Historically, this has led to overempha-sizing the fire and de-emphasizing anddevaluing the firefighter. We have spentmillions on fire research but little onfirefighter research. We have many fireresearchers. We have no firefighterresearchers.

On July 6, 1994, we lost 14 firefighterson Storm King Mountain. Theinvestigation of these fatalities clearlyshowed both psychological andorganizational failures. How did thesefailures come about? What can bedone to bring the primary focus back tovaluing people? Trees regrow, housescan be rebuilt, but the loss of a life isforever. What has unfolded in theaftermath is a reaffirmation that peopleare first. All else is secondary inwildland firefighting.

The 1994 fire season in which 34 peopledied was the catalyst that broughttogether firefighters, safety managers,psychologists, and sociologists for theworkshop. Together we discussed thehuman side of fighting fires. Weexamined firefighters, firefighter crews,fire management, fire culture, and firecommunities with the goal of enhancingthe firefighter amid a more highlyresilient organization.

The workshop began with four keynotespeakers who discussed new conceptsto give firefighters a look into ways toimprove themselves, their interactions,and the entire wildland fire community.Kurt Braun discussed the role humanbehavior plays in safety and injury, withemphasis on risky behaviors common

Mark Linane (left), Bill Bradshaw, and Buck Latapie discuss the Mann Gulch strategies from a “human factors” standpoint.

Workshop Overview

5

Part 2 of 4

High ReliabilityOrganizations: A Visionfor Fire Reorganization

The wildland fire community shouldreorganize using High ReliabilityOrganizations (HRO’s) as a model.Examples of HRO’s are nuclear powerplants and aircraft carriers.

Characteristics of HRO’s include(Rochlin 1993):

❖ The activity or service is inherentlycomplex in that tasks are numerous,differentiated and interdependent.

❖ The activity or service meets certainsocial demands that requireperformance at the highest level ofservice obtainable within presentsafety requirements, with both adesire for an even higher level ofactivity and a penalty (explicit orimplicit) if service slackens.

❖ The activity or service containsinherent technological hazards incase of error or failure that aremanifold, varied, highly consequential,and relatively time-urgent, requiringconstant, flexible, technology-intrusivemanagement to provide anacceptable level of safety to operators,other personnel, and/or the public.

“As stipulated at the outset, theorganization must not only meetservice and safety goalssimultaneously, but also must beperceived to have done so.”

Although fighting wildfires is not astechnologically complex as classicHRO activities, the management issuesare similar, particularly in the urbaninterface and prescribed fire arenas.

The yardsticks to determine a wildlandfire HRO’s reliability and effectivenesscould include the following (Creed andothers 1993):

in the wildland fire environment and howto change to reduce those risks. GaryKlein showed how experiencedfirefighters used recognition-primeddecision (RPD) strategies and howexperience is crucial for quick, effectivedecisions in a fast-changing, riskyenvironment. David Hart discussedcultural attitudes that can enhance orhinder firefighter safety andeffectiveness and how training canmake individuals and crews moreresilient to failures. Finally, Karl Weickintroduced insights from high reliabilityorganizations that help improvecommunication, leadership, groupstructure, and sensemaking, which inturn decrease stress and the chance ofcatastrophic errors.

That afternoon and the following day, theworkshop experts discussed firefighters,firefighting, and the fireground, andexplored the interconnections,emphasizing what was working or whatwas not. Possible solutions werediscussed. The third day participantstook the discussion into the field with atrip to Mann Gulch. The fire scenariowas reviewed where it happened,including how people interacted witheach other, the decisions that weremade, and how events unfolded in anincreasingly risky, changing environment.Insights not found in original reportswere put forth to explain how and why13 firefighters died on the Mann GulchFire. These new insights from apsychological perspective show thatanalysis and conclusions depend uponthe experiential bias of the investigator.The Mann Gulch experienceinvigorated the participants. The finaltwo days were spent exploringsolutions and developing both long-term and short-term recommendations.

The goal of the workshop was not tocome up with quick solutions. Rather itwas to explore the human issues ofwildland firefighting and recommend to

fire management corrective actions thatwould have lasting effects. As with allexplorations of human behaviors, thecomplexity and variety of issues wasapparent. But it became clear that agreat deal of relevant knowledgealready exists that other organizationshave institutionalized to reduce risk andimprove safety. Before we can use thisknowledge in the wildfire context, wemust establish baselines for relevantbehaviors. Without such benchmarks, wewould have no precise way to measurechange once corrective changes areimplemented.

It was quickly apparent in ourdiscussions that fire agencies are notroutinely collecting and analyzing datathat would give us a good idea aboutthe current behaviors of wildlandfirefighters. We don’t even collectcrucial near-miss information on the widevariety of risks inherent in firefighting.We only do a good job of recordingfatalities, Ensely (1995), but thisstrongly biases our view of normal,routine behaviors. Such a narrow focusprecludes warning trends that wouldbecome apparent in an analysis ofnear-miss situations. Therefore,workshop output depended on theexperience level and ability to recallrelevant information gathered inworkshop discussions, but for purposesof future discussion and correctiveactions, the information is grouped intothe following three main areas.

❖ A broad vision of how to reorganizewildland firefighting based on insightsfrom High Reliability Organizations(HRO’s).

❖ A specific reorganization of IncidentManagement Teams and fire crewsalong crew resource management(CRM) lines.

❖ Better assessment and feedback forall wildland firefighting activities.

6

Part 2 of 4❖ From whose perspective is effective-

ness or ineffectiveness judged?• Management • OSHA• Firefighter • Public• Politicians • Media

❖ On what domain of activity is theanalysis focused?• Safety • Training• Acres burned • Cause and effect• Houses saved • Decisionmaking• Accidents • Sensemaking• Near misses • Attitudes

❖ What level of analysis is being used?• Individual behavior • Baseline• Crew behavior • Culture• Longitudinal

❖ What is the purpose for assessingeffectiveness?• Error reduction• Promoting safety• Determining causal relationships

❖ What time frame is being employed?• Short term • Long term

❖ What types of data are being used forevaluating effectiveness?• Error rates • Compliance• Incidents • Safety checks• Accidents

❖ What is the referent against whicheffectiveness is being judged?• Agency standards• OSHA standards• Similar organizations

In analyzing the safety culture inHRO’s, the factors and theircontributory weights were (Koch 1993):

While HRO’s depend more ontechnological controls than wildland fireagencies, the process of looking at theirorganizational structure is relevant.

Using the CrewResource ManagementModel in Fire

Crew resource management (CRM)focuses on behaviors of crews.Adoption of CRM training and culturalchanges has dramatically reduced nearmisses and accidents in the airlineindustry. Most of the organizational andinteractive behaviors that are part ofCRM are relevant to the entire wildlandfire community.

CRM focuses on honing seven skills:situational awareness, mission analysis,decisionmaking, communication,leadership, adaptability, andassertiveness (Prince and others 1993;Frantz and others 1990).

These seven skills can be dividedinto taskwork skills and teamwork skills.Taskwork skills include: situationalawareness, mission analysis, anddecisionmaking.

❖ Situational awareness is theperception of what the fire is doingand what you are doing in relation tothe fire and your goals. It involves anawareness of fire behavior andterrain and the ability to predict wherethe fire and you will be in the future.This skill depends both on individualperception and sharing it with the restof the team.

❖ Mission analysis involves organizingand planning. It involves breaking themission down into subtasks, assigningpriorities to these subtasks, andmonitoring completion until themission is over. It begins with anorganized briefing and clarifiesimportant issues related to the mission.

❖ Decisionmaking involves decidingwhich decision model is mostappropriate for firefighters, such asRecognition-Primed Decisionmaking.It also involves training firefighters indecisionmaking and using it undersimulated stressful conditions.Decisionmaking includes collecting,integrating, and implementinginformation for the most effective taskperformance.

Teamwork skills include: communication,leadership, adaptability, andassertiveness. Communication andleadership involve at least two people,

❖ Effective communication primarilydepends upon the clarity, quality, andtimeliness of the message.Miscommunication has been acausal factor in many accidents.

❖ Leadership skills include delegatingtasks, providing feedback, promotingcrew motivation and cohesion—all inan atmosphere that fosters opennessby allowing crew members to presentalternative views without fear ofcriticism. The most effective leaderstake an active role in involving theentire crew in a team effort, discussinginteractions required for the tasks,and clarifying norms and roles.

❖ Adaptability refers to the ability tochange behaviors during a fire toreact to changing conditions and toother crew members. It refers totrying new behaviors when oldbehaviors are no longer effective.

❖ Assertiveness is necessary to helpindividuals who may feel intimidatedby a person’s position or fireexperience. It assures thateveryone’s special knowledge willbecome group knowledge.

Communication and leadership involveat least two people, whereasadaptability and assertiveness aremore individual characteristics.

Percent Factor explained

by factor

Accountability/Responsibility 23.2Adaptiveness/Responsiveness 16.3Openness/Cooperation 15.4Hazard awareness 14.2Inquisitiveness/Search for detail 13.2Role clarity 9.7Maturity 8.0

100.0

7

Part 2 of 4

Components of the CRMTaskwork Skills as TheyRelate to Fire (Prince andSalas 1993)

Situational Awareness—• Identify problems/potential problems• Recognize the need for action• Attempt to determine why

discrepancies exist with informationbefore proceeding

• Provide information in advance• Demonstrate ongoing awareness of

fire assignment status• Demonstrate awareness of your own

task performance• Note deviations

Mission Analysis—• Define tasks based on fire assignment• Structure strategies, tactics, and

objectives• Identify potential impact of unplanned

events on a fire• Critique existing plans• Devise contingency plans• Question/seek information, data, and

ideas related to fire plan

Decisionmaking—• Cross-check information sources• Anticipate consequences of decisions• Use data to generate alternatives• Gather pertinent data before making a

decision• Evaluate information and assess

resources• Identify alternatives and contingencies• Provide rationale for decision

Components of the CRMTeamwork Skills as TheyRelate to Fire

Communication—• Use standard terminology• Provide information as required• Provide information when asked• Ask for clarification of a communication

Assessment andFeedback

Assessment and feedback are essentialfor effective individual, team, and agencysuccess. That is why assessment andfeedback are such an important part ofboth HRO’s and CRM. But within theFederal wildland fire establishment,assessment and feedback are used soseldom that the workshop singled themout as the third area of major concern.

Throughout the workshop it was evidentfirefighters are being sent conflictingmessages from a variety of sources:political oversight, the agency, thepublic, and the fire organization. Mostfirefighters feel the task of putting outthe fire is primary and concern for theirsafety is secondary. Despite claims tothe contrary, safety is not yet thenumber one priority.

Firefighters want to be safe and avoidinjury, but there are times when thedemands of the job obscure safepractices. To deal with these instances,firefighters need to be equipped withbetter situational awareness anddecisionmaking skills. And they needfeedback about how they are performingthese tasks. Individuals and crewsseldom receive feedback. But without it,there is no way to measure performanceimprovements. Assessment is neededat all levels of the fire organization toestablish a baseline for policy, attitudes,and behavior. As changes areimplemented, measurements candetermine results. Feedback at all levelsis crucial for achieving positive changes.

• Make no response (negative)• Acknowledge communication (okay)• Repeat information• Reply with a question or comment• Use nonverbal communication

appropriately

Leadership—• Determine tasks to be assigned• Establish procedures to monitor and

assess the crew• Inform the crew members of fire

assignment progress• Verbalize plans• Discuss ways to improve performance• Ask for input; discuss problems• Tell crew members what to do• Reallocate work in a dynamic situation• Focus crew attention to task• Provide a legitimate avenue for dissent• Provide feedback to crew on

performance

Adaptability/flexibility—• Alter fire plans to meet situation

demands• Alter behavior to meet situation

demands• Accept constructive criticism and help• Step in and help other crew members• Be receptive to others’ ideas

Assertiveness—• Advocate a specific course of action• State opinions on decisions and

procedures even to higher-rankingcrew member

• Ask questions when uncertain• Make suggestions• Raise questions about procedures

This enumeration of examples undereach of the seven CRM skills clearlyshows the similarity in requirements forsuccess between the cockpit and thefireline. Both place a premium onindividuals operating as close-knitteams. Because of this similarity, CRMresearch data and training courses canbe readily tailored to wildlandfirefighting.

8

Part 2 of 4 Workshop Findings

TT he first three days focusedprimarily on determining wherethe fire community is organiza-

tionally, where it should be going, andhow the needs of both the firefightersand the fire organization could bebrought into closer alignment, with safetythe first priority. Workshop participantinputs were organized into three areas:(1) reorganization strategies for fireagencies based on HRO’s; (2) fire man-agement Incident Management Team(IMT) and fire crew reorganization usingCRM as a model; and (3) better assess-ment and feedback. The fourth dayfocused on future organizational studies,changes, and training that would movesafety to the forefront and improvefirefighter attitudes and effectiveness.

Fire Organization andCulture

❖ The wildland fire agencies shouldcompare themselves with HRO’s anduse research results to improve theagencies.

❖ Fire crews should be organized usingrelevant CRM concepts for improvedsafety and effectiveness.

❖ There is a need to clarify managementand public expectations of firefighters.Management and the public need tobe more realistic in their expectationsof the fire community. We should notfeel pressured to do more whileresources continue to dwindle. Wecannot always do a good job withwhat we have now, and the situationis getting worse. There are too manyconflicting messages about safetyfirst versus getting the job done.• Maximizing forest growth means

more severe fires in the future.• Often politicians and the public

exert pressure to go all out to savehomes in the interface.

• Unqualified personnel are makingfirefighting unsafe. This includesinexperienced EEO, downsizinglaterals, and others who have notworked their way up in the fireorganization with a combination oftraining and experience.

• Lack of financial and positionincentives to keep experiencedfirefighters in the organization.

• We taught the public we should andcan control all fires. Now they expectus to fight all fires with people,planes, helicopters, and retardant.This has led to higher cost fires andmore risk taking in the sky and onthe ground. There is a real need tore-educate the public about all theissues of fire management. Weneed to return to a more naturalview that all fires are not stoppablein the same sense that we cannotstop hurricanes, earthquakes,floods, and other natural events.

• Management needs to redefine“success” and “failure” in firefighting,together with priorities andconsequences. Evaluate allmessages against agency goalsespecially the goal of safety first.Eliminate miscues.

❖ It is easier to modify behavior thanattitudes. Changing attitudes occursafter a 3- to 5-year effort. Attitudesneed to be exemplified in behaviors.

Dave Thomas points toward Wag Dodge’s escape fire while Ted Putnam and Dave Turner consider his analysis.

9

Part 2 of 4❖ Agencies are not well organized to

handle extended initial attack andtransition fires, where most fatalitiesoccur.

❖ The current fire culture does not fosterrespectful interaction. If a fire is goingto blow up, is it culturally acceptablefor anyone to voice an opinion? Doall firefighters have the courage toraise this point? All firefightersshould be allowed to verbalize theirfears. Firefighters should be givensituational assessments in a respectfulcontext. When the situation is unsafe,they should also be allowed to pullback without loss of respect or threats.There is a need for organizationalclarity on factors involved in notengaging or disengaging, and whenthese factors align to result in actionto pull back. Some crews and crewsupervisors have a good, experience-based comfort level for when to pullout, whereas others do not.

❖ Firefighters need to be responsiblefor their own destinies and help workthrough the “more with less” period. Itwill be worth it in the long run.

❖ We need to identify constantsbetween firefighters’ attitudes andmanagement, and identify firefighterrituals, norms, values, etc.

❖ What is controlling fires about? Whatare we trying to do? Who is makingsense out of all this—firemanagement? Firefighters? Are weon the same mission? Whose vision?

❖ Are rational models of fireorganizations synchronized with theinformal work culture?

❖ Worsening organizational strainsinclude mixing personnel, decliningexperience levels, uncertainty ofexperience, under-funded training,downsizing, then placing laterals andEEO personnel with little or noexperience in high responsibility firepositions.

❖ There is a critical organizational needto rebuild a sense of community fromthe top down and the bottom up,because it seems to be disintegratingnow. If it takes up to six weeks forcrew cohesion and trust to develop,are people and crews reallyinterchangeable as managerspresume? Are there better ways toaccelerate cohesion and trust?Continually emphasize the fact thatthe humanity of the fire community isfar more precious than any otherresource. Remove barriers andinconsistencies between culturalexpectations and actual practices.Promote better cohesion.

❖ Cultural differences between groupsof firefighters:• The public and firefighters promote

group images that pressure “elite”groups to “aim to please” and “liveup to expectations.”

• The group culture affects risk takingand decisionmaking.

• More respectful interactions areneeded to bring expectations intoline with capabilities, for a bettersense of community.

• Management and IMT’s need totake group differences into account.

• Elite crews need to feel that theyare allowed to back down fromrisky, unsafe actions without anyloss of respect.

• Crews of different racial mixes haveunique cultural concerns.

❖ Too many red-carded personnel donot have the expertise indicated bytheir cards and positions on fires. Asa result a mistrust of all individuals isgrowing, and this in turn is a mistrustof the organization. There is an “us”versus “them” attitude betweenfirefighters on crews versus IMT’s orFMO’s and dispatchers. Most of thetraining opportunities, hence higherred card ratings, go to PFT’s asopposed to seasonals who haveconsiderably more fire experience.Filling fire vacancies with engineers,

foresters, and EEO candidates ratherthan seasonals further underminesthe experience base, and it is gettingworse. Creating unsafe managersthrough hiring practices flies in theface of upper level managementpronouncements about safety first.These “new” fire managers who donot see the big fire picture often areoverzealous micromanagers. Theagency needs to take a hard look atqualifications of FMO’s anddispatchers. They need CRM typetraining to better size up situations,make good decisions, andcommunicate the outcomes in anopen, two-way atmosphere. Thereare too many incompetent people onthe fireline. The red card system isfailing, which puts more firefighters atrisk. New evaluation processes suchas “hot-seat” simulations, panelreviews, etc., are needed for keydecisionmaking fire positions, toeliminate the possibility of one personbeing able to sign off another in a“buddy system” because of perceivedpressure or because the organizationneeds them. The rating system mustbe consistent throughout the nationand between agencies.

❖ Management needs to stop talkingand promote actions that foster realchanges in the organization.Policymakers could usedecisionmaking and situationalawareness training.

❖ Working safely is a natural outgrowthof clear, effective management andleadership. It is the result of actions,not words.

❖ Most of the fireline firefighters areseasonal employees. What is the bestway to organize, train, and acculturatethem for the future benefit of boththem and the fire community? Theremay be real benefits to bringing themon two weeks before the start of thefire season to foster safety trainingand cohesion. Currently, most

10

Part 2 of 4recognition goes to permanentemployees. More recognition ofseasonal employees and their valueto the program is needed. Betterincentives for seasonal workerswould promote safety and learning.

❖ Organizational defensive behaviorsare leading to unsafe practices. Wheninvestigation teams or managerscover up the causes of accidents andnear misses, no learning takes placefor the individuals or the organization.There is a need for forthrightinformation and open discussion at alllevels of the fire community.

❖ Psychologically, there is morepressure on firefighters to put the fireout than to do it safely.

❖ There appears to be too many fireorders and watchouts. A formalcontent analysis study may be ableto reduce these guidelines to a fewkey ones such as LCES (Lookouts,Communications, Escape safetyzones) that then should be prioritized.If some should never be violated, nomatter what the circumstances, thenthey should be identified. Some fireorders and many watchouts areroutinely disregarded. This isnecessary at times to accomplishsome fireline tasks and can lead toviolating orders that are not justguidelines. When an order is violatedand it works out okay, this can leadto more future violations. There is ageneral feeling that you must violatesome, but that can get you in troublewhen you string them together. Needto look at all the orders, watchouts,LCES and reorganize them formaximal clarity, minimum rules withclear direction from management, thenenforce them routinely. Since attitudesand rules do not always predictbehaviors, who is responsible foroversight and ensuring compliance?

❖ The agency should reorganize tosupport the firefighters and maximizetheir potential. The firefighters wantto perform at a high level and need

the organizational support to achievethat level. The agency has made firesuppression number one, and thisneeds to be changed so people arenumber one.

❖ There is an agency failure to followup to see if objectives, training, etc.,actually accomplish their goals. Oftenmanagement sets things in motionwithout any idea what effect itproduces in the field. Withoutfeedback the organization does notlearn.

❖ Fire managers, IMT’s, and fire crewsshould periodically shut down theirentire operation for a day, especiallyafter near misses or accidents. Stopdoing normal routines and reassesslarger goals. Groups need to focuson what is going right and what isgoing wrong. What is the worst thatcan happen, and what can be doneabout it. Organizational shut downscan be valuable learning experiences.

❖ Agencies should encourage more jobswapping for one year or one fireseason. Examples would be hotshot/smokejumper or FMO/hotshot swaps.We could also have a safety officer,FMO, or dispatcher shadow a hotshotcrew or be shadowed by a hotshot.This would help bring down barriersand create a true community feeling.

❖ The long-range forecast is for a periodof cooler fire seasons. This is comingat a time of accelerated skill erosionof fire personnel, fewer FTE’s anddeclining training dollars. Asprescribed burning increases tenfold,the “classroom” should be moved tothe burn site. OJT needs to beincorporated into the prescribed burnprocess.

❖ There is a need for more FTE’s andcareer tracks for key firefightersupervisory personnel in order topromote better experience levels andprovide a more professional nucleusfor supervising seasonal hires. A shiftshould be made to more tenured

firefighters as opposed to moreFMO’s and managers.Overdependence on firefighting as acollateral duty has diluted theprofessional firefighter base.

❖ Type I crews should have commonphysical fitness requirements.Current standards are too low, andthe poorer fitness levels of a few arecompromising the safety of the rest ofthe crew. This problem is especiallydisturbing when supervisors are lessfit than their crews.

Fire Management,Incident ManagementTeams, and Fire Crewsin a Crew ResourceManagement Context

Situational Awareness(Size-up)

❖ Basic situational awareness is highlydependent upon good information,skill, and experience. It is one of themost difficult skills to master and is aweakness in the fire community.

❖ Although basic subskills are taught invarious classroom courses, little isdone to see if the overall skill hastransferred to the fireground.

❖ Most firefighters possessing situationalawareness demonstrate decliningperformance as the fire accelerates.This indicates a need for simulationtraining in faster paceddecisionmaking, to facilitate quicksize-ups that keep pace with the fires.

❖ With lower tempo fire situations, wehave better recall and use rationalprocesses for assessing our situation.With high tempos, rational processesare too slow. We need recognition-primed decision (RPD) skills that comeprimarily through years of experience.

11

Part 2 of 4❖ The focus here is more on

sensemaking than decisionmaking.Sensemaking (Weick 1995) isobserving or creating patterns as weexperience reality. These conceptualexpectations form the basis tocomprehend, explain, attribute, andpredict events. It is experience drivenrather than a logical decision process.When expectations are disconfirmedan ongoing activity is disrupted andthen sensemaking is the process ofcoping with interruptions andsurprises. It is the process of makingthings sensible.

❖ During OJT, situational awarenessneeds to be an expected, formalaction and made public to others orwritten down. Then feedback shouldbe used to compare predicted versusactual results to improve predictiveskills. Otherwise, we tend to reviseour past predictions to fit what actuallyhappened. This latter process actuallymakes us worse at predicting futureevents. Later, under high-tempoconditions, this skill will be fluid andrapid.

❖ Part of the process of understandingsituational awareness is to ask whatare the adverse effects of incorrectsize-ups.

❖ Does the local FMO or dispatcheraccept your size-up? Do they give youall the resources you order? Are theresources timely? How does yoursituational awareness compare totheirs? Do they advise you of resourcestatus, recommend alternatives, andassess consequences?

❖ Situational awareness is critical formaking decisions on whether or notto fight the fire and later on whetherto stay engaged or disengage fromthe fire.

❖ The higher the tempo the more oftenyou need to perform anothersituational check.

❖ When any significant event changes,then another situational check mustbe made. When situational checksbecome too frequent, this is a cue toconsider disengaging.

❖ Whenever you become unsure of yoursituational assessment and vacillateover various inputs then “safety first”directs you to assume the worstbecause people tend to underestimatethe severity of situations. For example,if you are vacillating between whetherthe situation is severe enough to orderretardant, then order the retardant.

❖ Part of situational awareness is tohave a clear understanding when youare getting in over your head, whenthe situation no longer makes sense.Then it is time to call for moreresources or to pull out.

❖ There should be a requirement tocommunicate revised size-ups amongcrews, FMO’s, and dispatch every “x”hours, depending on fire danger andtime of day.

❖ Identify situations requiring heightenedawareness such as extended initialattack, transitions, interims untilresources arrive, urban interface, theactual arrival of the resource, andinterims after accidents or nearmisses.

❖ May need a checklist of factors toconsider when sizing up a situationso no factor is missed. As a minimum,LCES should be included. Discussemergencies, what are the earlywarning signs and what to do if theyoccur.

❖ Part of situational awareness shouldinclude giving good briefings anddebriefings that communicate all theessential facts. This becomes thebasis for the situational awareness ofother firefighters. There should bestandard briefing practices that aregiven and expected. Briefings shouldbe face-to-face whenever possible.Ask questions to see if the essential

content of the briefing has beenunderstood.

❖ A pre-accident situational awarenesswould be to run through all known andsuspected risks associated with a fire.This initial information becomes achecklist to consider once you get tothe fire.

❖ Need a good sense of time. How longdo certain actions take, how long untilresources arrive, and how long toshadow during transitions?

❖ Situational awareness cannot bemandated. We need people to bethinking, discussing, and observingconstantly for most effective use ofthis skill.

❖ Consider using the Campbell dangerrating system or one like it forformalizing situational awarenessand the language to communicate itto others. Need a system that teachesinexperienced firefighters to size upfires the way experts do. The samesystem should be used by the IncidentCommander (IC), FMO, and dispatchfor maximal information transfer.

❖ Part of situational awareness isknowledge of safety and deploymentzones, escape routes, and escapetime. This must be planned andcommunicated to all firefighters.Emergency actions must be wellpracticed and understood for them tobe available and effective whenneeded.

❖ Situational awareness should includethe fire, other people and resources,and a periodic internal check, andhow all these interrelate over time.

❖ What are situational awareness redflags?• Change—large, unexpected, faster

rate• Expectations not met—resource

changes, times• People not communicating• Stress—various stresses are additive

12

Part 2 of 4❖ FMO’s at the district, forest, or area

level must develop clear criteria fordetermining when they are in severeor extreme fire danger. Then theymust warn against business as usualand function in a high-tempo mode.They must communicate the situationto local and nonlocal fire personnel.

❖ Dispatchers, FMO’s, coordination, andresource allocation centers mustdevelop clear criteria for determiningwhen they are in over their heads andthen call for help. The process andcriteria must be in place before theneed, then reviewed weekly or dailyas the fire season progresses.

❖ It is useful to project a likely situationand a worst plausible situation, thenbuild a plan that can survive the worstplausible situation and can also workeffectively for the likely situation.

❖ Judgment of safety margins, patternsof cues that signal that risk is too high,must be carefully trained before theassignment is accepted or crewsdeployed. It is easier to avoid than getout of a bad situation. The judgmentcan be refined to reflect changingconditions to determine when thesafety margin has been graduallyreduced to a point where it isunacceptable. Gradual reductions areparticularly difficult to observe.

❖ Training should ensure breakpointsare overlearned for improved safety.Breakpoints involve the rapidrecognition that the situation hasbecome untenable and, rather thancope and adjust, it is time to radicallychange the game plan; survival hasbecome the number one priority. Thisincludes learning to abandon firelinesthat were built at considerable cost ofeffort.

Mission Analysis

❖ Mission analysis begins with overallfire strategies and tactics, situationalawareness with size-ups and

briefings. Then the larger tactics arebroken down into specific tasks, taskassignments are made, tasks aremonitored, then tactics reassessed.

❖ Mission analysis tends to work wellexcept for extended initial attack andtransitions, and during interims beforeresources arrive, etc. In thesesituations environmental changes areoccurring faster than strategies,tactics, and tasks can be changed totry to keep the mission on track.

❖ Mission analysis also includesawareness and knowledge of whenthe mission can no longer beaccomplished safely. Do not start, ordisengage as appropriate.

❖ It is crucial for overall mission successto explain the mission to the crew,explain their individual parts, thenallow them a chance to ask questionsand clarify the mission. It involvesboth briefings and debriefings. End ofmission debriefings are importantlearning processes for transferringknowledge and learning.

❖ Mission analysis must take intoaccount LCES and be ready toimplement alternate plans whencurrent plans fail. Complications occurwith mixed resources, indefiniteresource arrival times, andunexpected fire behavior.

❖ Each team member must haveappropriate training and knowledgeto accomplish a specific task. Missionanalysis must clarify roles and ensureeach person performs a role, yetinteracts well with people or crewsthey border. When the missionchanges, the people may need tomake role changes quickly. The morerisk or faster the tempo, the moresupervisors must pay less attentionto specific work tasks and moreattention to the big picture andoversight supervision. At some pointeveryone must switch to emergencyroles where escape becomes para-mount and all individuals stop ordinaryactions and focus on supervisory orders.

Decisionmaking

❖ Different decisions necessitatedifferent models. A rational modellooks at strategic decisions andtherefore prescribes best tactics.Naturalistic models look at decisionsunder stress with minimal responsetimes and focus on making sense ofthe situation and taking rapid actionto alleviate problems. RPD is anaturalistic model. Firefighters needtraining with both models andguidelines that help determine wheneach model is better.

❖ Current firefighters receive little or notraining on decisionmaking skills.Firefighters need to recognize a needfor balance between individualdecisionmaking and groupdecisionmaking. They need trainingon how situations, stress, other people,and groups affect their decisions, andon aids for clear decisions. They needto discriminate between sensemakingand decisionmaking.

❖ Training needs to be specific to thejob. Firefighters need to make tacticaldecisions, and managers need tomake organizational decisions.

❖ Factors in decisionmaking:• Decision point or branch• Errors• Does person have prerequisite

skills?• Biases• Cultural differences• Intelligence differences• Reliability of the information

❖ Need to study decisionmaking increws, operations, and IMT’s. Trainingshould be group specific. Somepositions, such as division and crewsuperintendents, may need morethan one type of training due tovariable roles.

❖ Currently there is no clear sense ofwhat is expected of firefighters.Institutional messages are conflicting,so decisions are not alwaysconsistent with management

13

Part 2 of 4expectations. Firefighters are askedto take risks, fight fire aggressivelybut safely. Where is the boundarybetween risk and safety. Whodecides on where the boundary is:management, IMT, crew supervisor,or individual firefighters?

❖ There is a need to do a factor analysison all the decision aids currently invogue:• 10 Standard Fire Orders• 18 Watch-out Situations• 5 Common Denominators• 4 LCES• 10 Downhill/Indirect Line

Construction Guidelines• 9 Urban/Wildland Watch-outs 56 total

A factor analysis would reduce these toa bare minimum. They should begrouped into never violate, transgresswith extreme caution, and watch outs toavoid. If all these aids are onlyguidelines, then we should not criticizefirefighters who do not follow themperfectly and accept that they made thebest decision given their experience,training, and awareness level. Puttingthem in order of priority would help.

If we adopt a rule “safety first,” then itmust be reflected in all decision aids orat least be the top priority.

❖ Internal Watch-outs• Physical fatigue• Mental stress• Fear/Anxiety• Tight stomach muscles• Action tunneling• Want to speak out but don’t• Overconfidence, confidence

increases• Decisions made without feedback• Situation ambiguous or doesn’t

make sense• Microsleeping• Changing belief to match action• Accepting increased risk• Recent family problem• Organization or individual distrust

❖ Intrapersonal/Crew Watch-outs• Two inexperienced persons in direct

line of command• Other person/crew is tired or

stressed out and is making crucialdecisions

• Person won’t talk or is hostile• Cocky, overconfident individuals• Group polarization• Declining communication and

feedback; supervisors are reluctantto ask for help

• It is unclear who is in charge of the“big picture”

• Group consensus without sufficientinformation

❖ Management Watch-outs• You don’t receive resources or the

dispatchers argue about whatresources you need

• Resources will be late arriving• Politicians are in the area• Multiple agencies are involved• Dispatchers/FMO’s keep track of

things in their heads rather than onpaper

• Norms for radio discipline are loose• Agency is reluctant to ask for help• Administrators are getting on-the-

job training• Administrators say keep it simple• When overheads are unknown or

tough to find• Dispatchers are more concerned

with homes than firefighters• News media are in the area• Tensions and conflicts exist before

the fire season

❖ Stresses that interfere with gooddecisionmaking include:• Anxiety • Sleep loss• Frustration • Vibration• Noise • Hunger• Alcohol • Cold• Heat • Time pressure• Fluid loss • Time of day• Drugs • Incentives• Fear • Punishments• Anger • Personal problems

Stresses are additive!

❖ Stress affects decisionmaking by:• Lowering awareness• Lowering concentration or ability to

focus• Making it harder to access long-term

memory• Locking us into repetitive, habituated

behaviors• Focusing more on task, working

harder, and ignoring environment

❖ There is a crucial need to studyfactors involved in deciding whetherto engage or disengage a fire. Thisincludes initial attack and standardfireline duty. This whole area is vagueto firefighters.• What objective factors are involved?• What subjective factors are

involved?• What is official agency policy?

Rules take pressure off individuals.• What rewards and punishments

affect the decision?• Where is the boundary between

safety and normal, risky,aggressive firefighting? How narrowis the margin of safety?

• After difficult engagement decisionsare acted upon, we need to followup with good feedback anddebriefing, then use the incident toimprove decision factors.

• Must use a common language so itcan be discussed more accurately.

❖ Should agencies enforce the use ofLCES at all levels? Needs to be topto bottom, bottom to top. Ifinstitutionalized, LCES would be partof every briefing on the fireline, aswell as for the IMT, FMO’s, anddispatchers.

❖ Can LCES be an absolute, neverviolated? What are safety zones if aspot fire is in the middle of a 5-square-mile brush field? Do you need alookout? Or does the procedure thatsays to discuss fire in relation toLCES become the basis forsituational awareness on which tomake the decision to engage?

14

Part 2 of 4• Better between similar crews (i.e.,

hotshots)• Better between people who know

and trust each other• Hard during transitions; need

guidelines• Need more skill training on

maximizing information with fewestwords

• Need to foster a cultural attitude ofrespectful interaction to promotetrust

• Temporary employees have a hardtime communicating upward

• Need nonthreatening method tocommunicate personal experiencelevel. Try to communicate face-to-face as soon as possible.

• Need for more dialogue when peoplefirst meet, even if on radio, as thisreduces the number of wordsneeded for effective communicationlater as the people betterunderstand each other’s point ofview.

❖ Story telling is an effective method forcommunicating agency values andlessons learned.

❖ Essential to have a commonlanguage (English), common terms,and common expectations (size-upand LCES) to convey moreinformation in less time.

❖ Need for training, especiallysupervisors, IMT, FMO’s, anddispatchers on interacting moreeffectively and removing mistrust andcommunication barriers. Needlanguage and training to resolvedifferences of opinion as opposed toavoidance or going around someonewe have difficulty with.

❖ Everyone in the fire community needsto talk and interact more with theircounterparts both during the fireseason and off season. This will re-establish a feeling of fire communityand trust and improvecommunications when the tempoincreases in severe fire seasons.

taking. When we talk about savingsomething we are more conservativein taking risks. When we talk aboutlosing something we will take greaterrisks.

❖ Information occurring close in timetends to be automatically linkedtogether even when it is unrelated. Beaware of this when making decisions.When unsure of information, requestclarification. Also be careful about howyou put information together to briefothers.

❖ Factors affecting whether to engageor disengage:• Fire resources committed• Fire resource timing• Risk assessment• Fire behavior—actual and expected• Urban interface• Public pressure• Political pressure• Value of resource you are protecting• Recognized options• Clear management guidelines

Communication

❖ Functions and Problems (Kanki andPalmer 1993)• Functions

-Provides information-Establishes interpersonalrelationships

-Establishes predictable behavior-Maintains attention to task andmonitoring

-Is a management tool

• Problems-Lack or misinformation-Interpersonal strain-Non-standard, unpredictablebehavior patterns

-Loss of vigilance, situationalawareness

-Lack of or misdirected leadership

❖ Communication on the fireline• Good within a crew but not between

crews

❖ Making decisions without feedbackshouts watch out. The tendency is tobe overconfident when feedback isweak. No learning without feedback.Should give feedback to others andexpect it from them.

❖ Explore types of decisions and whenthey are made. When are most crucialdecisions made? Do we make themin an active or reactive state? If muchinformation is being processed, is theinformation reliable, timely, andnecessary? Are inputs assumed oris a checklist used?

❖ Consider adoption of the Campbelldanger rating system or one like it tofoster better decisions.

❖ Currently, there is no training to teachyou when you’re in over your head.Usually, by the time it sinks in, yoursafety has been compromised.Tendency is to hang on too longbecause it is admitting defeat if youdo not. There needs to be moreagency direction here to take pressureoff the individual. Need training torecognize cues and early warnings topull out or to ask for more resourcesbefore the situation becomesdesperate. FMO’s, dispatchers, andothers need to monitor fire activity andassume a more active role in thesedecisions from a position of mutualrespect with the IC.

❖ When there is a difference betweenexpectations/beliefs versus action, wechange our expectations and beliefsto fit our actions. If we are trying tofoster new expectations such as“safety first,” then we need to useincentives to reinforce theexpectation and use feedback tocorrect inappropriate actions.

❖ When a group of risk takers is puttogether, the group will take more risksthan any individual would take alone.This and other factors associated withrisk taking need to be incorporatedinto the decision process. Even theway you think about risk affects risk

15

Part 2 of 4❖ Greater information flow up, down,

and across improves everyone’sexperience and competency. Thisprocess takes years to develop. Weshould start now, stay enthused, andexpect change over a longer periodof time.

❖ Need for open dialogue whenproblems occur. Discuss and manageproblems while they are small andless emotional. If you’re thinking it,express it out loud.

❖ Firstline supervisors set the tone forcommunications. Agency must sendclear signals to supervisorsconcerning their responsibility topromote open, two-way, respectfulinteraction. Supervisors should leadcrews to avoid emotional-laden topicsuntil mutual respect and crewcohesion have formed. Supervisorsshould clearly communicate expectednorms of behavior, then useincentives and feedback to ensurecompliance. Crews and individualswant cohesion and trust if it’s allowedto develop naturally.

❖ Need a common tactical languagesuch as the Campbell danger ratingsystem to foster clearer communica-tion of fire behavior, expectations,briefings, and feedback.

Leadership and Cohesion

❖ Leadership is a crucial skill forimproving firefighter safety. An open,democratic leader promotes crewspirit, cohesion, and maximum crewgrowth. This occurs through an activeteacher/mentor role to foster crewknowledge. A cohesive,knowledgeable, open crew is a safecrew.

❖ After a size-up, a good leader sharesthe information with the crew.Individual crew members are

encouraged to do their own size-up,determine the outcome, and askquestions about why their size-up orthe leader’s size-up was on or offtarget. The leader should quiz crewmembers, who in turn should quiz theleader.

❖ A good leader provides maximalfeedback to the crew to foster crewlearning. The leader sharesexperience, training, and knowledgewith the crew.

❖ In times of declining budgets andtraining dollars, a crew leader musttake the classroom to the field on thejob.

❖ On initial attack and transition fires, itis not always clear who is in charge.When authority is delegated, the chainof command should be clear to allfirefighters. Official transfers shouldbe face to face and signed in diaries.If a leadership change occurs on thefireline, the change should be relayedto dispatch and recorded.

❖ All leaders must have leadership andsupervisory training, even if theirofficial jobs do not require that skill.To be a leader on the fireline, youmust be trained. Too often untrainedleaders regress to being regularfirefighters when conditions becomestressful.

❖ Leadership training for firefighters ispoor. Being an office supervisor doesnot equate to being a leader on thefireline. We need to determine whatskills a fireline leader needs, thentrain people in those skills. Manyproblems occur on the fireline due toassuming office rank equates to firerank.

❖ There is no good system in place topromote individuals who excel infireground leadership. More FTE’sshould be set aside to create acareer track for people who exhibit

fireline leadership. They are thenucleus of the fire crews, and theirexperience is essential for safety onthe fireline.

❖ It is essential for crew leaders todebrief their crews after each incident.Leaders should insist on a debriefingfrom the IMT or IC and give their owndebriefing to the crew. This feedbackis essential for learning to occur.Leaders should give orders, thenexplain them as much as possible.

❖ Crew supervisory job descriptionsshould be revised to reflect the needfor people who are open and honest,and who can act as teachers andmentors as well as being skilled inleadership and knowledgeable aboutfire behavior.

❖ All incident leaders need to fostermore intermixing between people andcrews to create an open atmospherefor sharing experiences andknowledge. This should be expectedbehavior among all firefighters.

❖ Identify skills needed for effectivefireground leadership, including:• Command and control practice• Time and space relationships• Quick, bullet-type communication• Stress awareness• Experience• Situational awareness and

assessment• Criteria on when to engage or

disengage• “Hot-seat” decisionmaking under

stress for quicker decisions—RPDtype decisions

• Task assignment• Mission awareness

❖ Leadership training courses shouldbe mandatory for all IC’s and divisionsuperintendents. Courses should beMarana style (upper level) withsimulations under stress.

16

Part 2 of 4❖ There is definite skill erosion during

light fire years. Leaders should beheavily involved in prescribed fire tohone skills.

❖ When leadership changes on thefireground it should be formal:• Face to face• Declared to dispatch and entered in

the dispatch log• Both IC’s should sign diary with time

and date of exchange• Consider other positions for sign off

(in addition to IC’s)

❖ Leadership, crew cohesion, andsafety are strongly correlated. Openleadership style fosters bettercohesion and safety.

❖ Good crew supervisors do not focuson safety but rather on goodsupervision, crew cohesion, and workethics. Safety is the result.Supervisors who constantly talk aboutsafety have more accidents thanthose who focus on workingrelationships.

❖ A lot is known about crew leadership,cohesion, and trust, which takes 6 to8 weeks to develop. It may developquicker for fire crews. Is there a wayto study this and accelerate the effect?

❖ When people off districts, forests, etc.,are brought together to form a crew,they are much more effective andsafer if they spend a day togethergetting to know each other beforegoing on the fireline. This techniqueshould be further investigated as amethod to speed up group cohesion.

❖ There used to be a better sense offire community among firefightersand managers. Has this sense beenlost or has the fire family becomedysfunctional?

❖ Leaders need to work with crewmembers and promote respectfulinteractions; encourage their input sothey feel part of the crew. Onceleaders get input, crew members

Assessment andFeedback

❖ The current system for reportingentrapments is working, but not veryeffectively. Some entrapments arereported only after long delays, andsome aren’t reported formally untilsomeone follows up on rumors andpressures a person or crew to fill outthe forms. This system should be re-examined and made more effective.Firefighters should not have theoption to fail to report entrapmentswithout penalty. They should not bepenalized when they do reportentrapments in a timely manner.

❖ A new system must be implemented torecord and track near-miss situationsfor all wildland fire operations. Itshould include all accidents andincidents, even minor ones. Thisbaseline information is necessary todetermine where we currently haveproblems and if management ortraining changes decrease near-misses, accidents, and incidents. Thissystem should be modeled after theairline industry where there is nopenalty for calling in an accident ornear miss when reported at theearliest opportunity. An open,nonthreatening system will promotemore frequent and more accuratereporting, therefore greater safety.

❖ It would be useful to have trainedindividuals or teams go out on thefireline each fire season to observecrews and individuals in action. Theinformation gathered would showwhether training or managementobjectives have transferred to thefireline. IMT and crew memberscould be quizzed or interviewed todetermine skills and knowledge.

❖ The agencies should require thatleaders reassess their situation every15 to 60 minutes, depending upon firedanger. Taking time out to reassessallows you to determine if new actionsare required. There should be aformal checklist like LCES.

should expect leaders to makedecisions and lead them toaccomplish goals.

Adaptability/Flexibility

❖ Adaptability skills need to beaddressed. How flexible are wildlandfirefighters to quickly change tacticsas environmental conditions change?Do our crews stay too long at the taskat hand when a new approach iscalled for?

❖ Need flexibility to keep reassessingthe situation on a routine basis.

Assertiveness

❖ Assertiveness is natural for somefirefighters. But for others, it is a skillthat must be learned, then practiced.

❖ Leaders of teams and crews arepivotal in creating a climate thatencourages all firefighters to speak up.

❖ Firefighters have a tendency tointernalize what’s bothering themrather than speak up about it. Weneed to emphasize more externaldialogue.

❖ We also need more assertivenessbetween leaders to communicate theirsize-ups to others and to discuss theirexperience level with others. We needthis exchange so both leaders perceivethe same external environment as abasis for future decisions and knowwhat to expect from the other personbased on their past experience.

❖ Assertiveness is also necessary torequest fire and weather information,briefings, debriefings, etc., when theyare not given. This includes askingquestions or requesting that someonerepeat information you did notunderstand.

17

Part 2 of 4❖ Every person in the fire community

and on the fireground needs toincrease communication andfeedback up and down the chain ofcommand to maximize learning.Everyone needs to become moreexpert at both giving and receivingfeedback.

❖ Attitudes don’t always predictbehavior. So it is important todetermine what behavior isencouraged or discouraged in theactual work environment. What arethe real consequences for followingvarious orders. Stories, games, andvideos are three methods ofcommunicating expectations andconsequences.

❖ Once entrapments and close-calldata are analyzed, the facts must getto individual firefighters for learning totake place. This feedback heightenssituational awareness and the abilityto recall the information if needed.The individual and crew names canbe removed as long as the key factsare well communicated.

❖ Firefighters need quality briefingswhen they first arrive on a fire. If theystart out behind, they will rememberand process less information incritical situations.

❖ Individuals must practice behaviorbefore it happens automatically.

❖ Consider a 1-800 hotline to collectsafety data. It should be anongovernmental agency to ensurehigher reporting rate and anonymity.

❖ Try to teach in the field as much aspossible. It promotes better learningand recall because that’s where it willbe needed in a critical situation.Prescribed burns are a greatclassroom setting.

❖ An agency protocol is needed forbriefing each other on our currentfirefighting qualifications. The redcard ratings are deceptive and thereneeds to be more face-to-facediscussion of qualifications to size upindividuals or crews you will beworking with. That is part of theoverall situational awareness. Whatis agency protocol if you feel theother person isn’t qualified?

❖ There is a need to explore alternativetraining and feedback methods:• Interactive investigative books• CD games• Hot-seat simulations

❖ MTDC should publish a quarterlyhuman factors newsletter similar toHealth Hazards of Smoke . Targetall fire safety personnel and firefightingcrews in addition to normal region/forest/district distribution.

❖ Are extended initial attack, transitions,urban interface, helicopter downwash,etc., really our most risky, hazardoussituations or is this rumor? What arethe trends and how significant arethey? What are the situations thatcause the most firefighter injuries?

❖ Start using computers to move peopleto and from fires and while on fires toeliminate all the waiting time. Figureout ways to use down time for training.

❖ Small individual AM receiving radiosare a dollar or two. If each firefighterwore one, it would be a means forbroadcasting weather, fire behavior,news, and other general information.

❖ Situation checks should be requiredwithin a crew and among crews as adouble check that everyone agreeswith the situational analysis. Thecheck could follow LCES. Respectfullydiscuss differences. When a situationgets critical, ask the recipient to repeatthe analysis back to you.

❖ Fire safety officers should do spotchecks on safety equipment andpractices. They can determine whattraining has been given and iffirefighters know the basics. Theycan ask firefighters to give them asituation size-up based on LCES andhazards in the immediate area.

❖ Need better, consistent post-firedebriefings for individuals, crews, andIMT’s. The process should encouragefeedback both up and down the chainof command.

❖ Need a long-range look at what weare about and what we do. Needlongitudinal field studies to accomplishthis task. This would make it clearwhether management objectives getincorporated into behaviors in thefield.

❖ Greatest safety factor on the firelineis clear thinking. Look for clues,analyze the input, and predict. If youcan’t predict, then stand back andwatch what’s happening until you canpredict. Then take action based onclear thinking.

-End Part 2-

18

Part 3 of 4 Discussion

HH igh Reliability Organizations andCrew Resource Managementmodels were used to organize

workshop discussion topics. This

appeared to be a successful strategybecause most workshop topics fit thecharacteristics cited for HRO’s andCRM. This in turn suggests these areas

are worthy models to pursue as we lookat ways to better organize wildland fireagencies and fire crews.

Joseph B. Sylvia’s marker cross on a steep Mann Gulch slope where 13 firefighters lost their lives on August 5, 1949.

19

Part 3 of 4

Workshop participants eat lunch on top of the ridge where Bob Sallee and Walt Rumsey escaped the Mann Gulch Fire.

❖ Contract to have organizationalexperts evaluate Fire and AviationManagement (F&AM) and proposeways to reorganize it into a highreliability organization able tofunction at a high tempo during fireseason. Evaluate F&AM using theseven factors presented earlier foreffective HRO’s. Consider workshopinput for organizational change whenevaluating F&AM.

❖ Contract to have CRM coursematerials adapted to wildland firecrews and teams. Determine if otherskills are necessary that are uniqueto the fireground environment.Consider workshop input whenmodifying CRM for firefighters.Change name to Fire Crew Dynamicsand Fire Team Dynamics if coursematerial would be different forfirefighting crews and IMT’s.

❖ Study current loss of experiencedfirefighters, crew supervisors, FMO’setc., to determine how to reversethese trends. Consider more FTE’s,higher pay, and other incentives.What is the effect of combiningpositions and collateral duties on theorganization?

❖ Offer more incentives for seasonalfirefighters to return and be bettertrained. Consider:• Bring them on earlier for extra

training• Increasing bonus system for those

returning for a third, fourth year, etc.• Pay for training costs incurred by

firefighters in the off season whenit is relevant, and they are returninganother year, or give bonus in lieuof all costs.

❖ Contract to examine all the fire orders,situations, etc., to determine if theycan be simplified and prioritized. Areany of them absolutes? Can what’sleft be followed and still put out fires?Add management, crew, and internalwatch-outs as needed.

Recommendations

❖ Red Card Qualification System doesnot work effectively. Contract to deter-mine what the system is supposed todo and how to make it work.

❖ Study and formalize guidelines forengaging and disengaging from fireassignments. Study real crews anduse content analysis and interviewtechniques such as Cognitive TaskAnalysis.

❖ Initially develop decisionmakingexamples suitable for wildland fire-fighters. Use firefighting examples todemonstrate how stress and otherenvironmental and psychologicalfactors affect decisions. To beeffective, decisionmaking must beincorporated in all other trainingprograms rather than, offered as astand-alone course.

❖ Develop a situational awareness classand determine critical cues and howto accelerate training of inexperiencedfirefighters. Study the RPD model of

rapid awareness and decisionmakingby studying firefighters in their naturalenvironment.

❖ Adopt common protocol andlanguage for all firefightingcommunications. Consider Campbelldanger rating system forcommunicating vital fire informationquickly and accurately.

❖ Develop leadership course(s) for allIC’s and crew supervisors. Determinetype of leadership needed on firelineand train people accordingly.

❖ Develop a family of “hot-seat” stylefire simulators to train and evaluateCRM skills while in a high-temposituation. A good developmentalproject could use the same inputs thatallow computer modeling of an actualfire when relevant data is input. It canbe used for training when inputs arechosen and firefighters must make aresponse.

20

Part 3 of 4❖ Conduct longitudinal study on fire

crews to identify relevant behaviorsthat increase cohesion, safety, andproductivity. Study whether manage-ment attitudes become crew attitudes.What factors and activities speed upthe learning process? Are there sexual,racial, or age factors involved?

❖ Require all prescribed burn plans toadopt a “classroom” element so theburn is fully utilized as a trainingexercise.

❖ Publish a human factors in wildlandfirefighting newsletter similar to HealthHazards of Smoke .

❖ Hire professional instruction systemdesigner to determine best format forimplementing training, i.e., video,printed materials, computer simulation,etc., to maximize training transfer.Need to consider more hands-on,interactive training. Need more field-based training, which improves mem-

ory transfer and learning, and is lessboring than classroom-based training.

❖ Organize more national, regional, andlocal rendezvous where there is moremixing of type I, type II, engine, andhelitack crews, FMO’s, IMT’s, anddispatchers so they can shareknowledge and discuss problems.

❖ Implement all assessment and feed-back proposals from the workshop.Without strong institutionalizedbaseline measurements and incidentreporting, there is little chance tolearn.

❖ Develop methods to speed up crewcohesion and work practices beforefireline assignments.

❖ The bulk of training occurs throughOJT, but little preparation and careare given to make OJT work efficiently.Contract to study the best way toboost skills in a relatively short timewith little cost through improved OJT.

❖ Contract to have professionals provideguidance in setting up procedures forcollecting and disseminating lessonslearned from fireline duties andentrapment that will be interesting andused by firefighters and managers.

The above recommendations are not inany order of priority. The participants didnot discuss priorities. In addition, evenmore recommendations could be pulledfrom the workshop notes. We felt thatpriority setting should have greaterconsensus than our group. In the mid-1980’s, a number of conflicting equip-ment needs surfaced. To set priorities,more than 2,300 questionnaires weresent out to Forest Service and InteriorDepartment offices and State agencies.Something similar is recommendedhere. We would like to encouragereaders of these notes to suggest otherrecommendations, and we in turn willask that NWCG consider surveying andprioritizing projects through input fromthe entire wildland fire community.

21

Part 3 of 4

TT he workshop findings andrecommendations are not meantto be an end product but rather

the beginning of a continuingassessment. It would be a mistake tothink that a one-time effort to developnew training or a new organizationalstructure is enough.

We hope to set in motion a process thatwill lead to a fire organization thatreinvents itself as a high reliabilityorganization where:

❖ The capacity to learn and adapt arevalued and encouraged for theorganization and the individual.

❖ The people are committed to theprinciples of CRM on the fireground.

❖ Safety and firefighters are numberone.

A Final Note

Suggested reading, see page 23.

22

Part 3 of 4 References

Creed, Douglas; Stout, Suzanne;Roberts, Kathlene. 1993. Organiza-tional effectiveness as a theoreticalfoundation for research on reliability-enhancing organizations. In: Roberts,Kathlene H., ed. New challenges tounderstanding organizations. NewYork: Macmillan Publishing Co.: 63-70.

Ensely, Pam. 1995. Historical wildlandfire-fighter fatalities 1910-1993.NFES #1849. Boise, ID: NationalWildfire Coordinating Group, Safetyand Health Working Team, NationalInteragency Fire Center.

Frantz, T.M. [and others]. 1990. Theidentification of aircrew coordinationskills. In: Proceedings of the 12th

annual Department of Defensesymposium. Colorado Springs, CO:U.S. Air Force Academy: 97-101.

Kanki, Barbara; Palmer, Mark. Commun-ication and crew resource manage-ment. 1993. In: Wiener, Earl; Kanki,Barbara; Helmreich, Robert, ed.,Cockpit resource management. SanDiego: Academic Press, Inc.: 99-136.

Koch, Barbl A. 1993. Differentiatingreliability seeking organizations fromother organizations: development andvalidation of an assessment device.In: Roberts, Kathlene H., ed. Newchallenges to understanding organiza-tions. New York: Macmillan PublishingCo.: 81.

Prince, Carolyn; Salas, Eduardo. 1993.Training and research for teamworkin the military aircrew. In: Wiener, Earl;Kanki, Barbara; Helmreich, Robert,ed., Cockpit resource management.San Diego: Academic Press, Inc.:337-366.

Rochlin, Gene I. 1993. Defining ‘highreliability’ organizations in practice: ataxonomic prologue. In: Roberts,Kathlene H., ed. New challenges tounderstanding organizations. NewYork: Macmillan Publishing Co.: 15.

Weick, Karl E. 1995. Sensemaking inorganizations. Thousand Oaks, CA:Sage Publications.

-End Part 3-

23

Part 4 of 4 Appendix A—Overview

Decision Workshop

Improving Wildland Firefighter Perform-ance Under Stressful, Risky Conditions:Toward Better Decisions on the Firelineand More Resilient Organizations

June 12–16, 1995Village Red Lion Inn100 Madison StreetMissoula, Montana

Overview—

It has become increasingly clear since1990 that wildland firefighters are exper-iencing collapses in decisionmakingand organizational structure. Wildlandfire agencies have lost 23 people since1990 who might have survived had theysimply dropped their tools and equip-

ment for greater speed escaping fires.More than 30 people are entrapped eachyear. Our crews are not as proficient atescape, fire shelter deployment, anddecisionmaking under stressful, riskyconditions as they could or should be.Partly, this reflects attitudes, and partlyit is a lack of knowledge.

This workshop will explore firefighterpsychology, interactions among fire-fighters and among fire crews, and betterways to organize. To do this, we havebrought together experts in psychology,sociology, organizations, fire safety, andwildland firefighting. We will be lookingat the current situation on the fireline andways to make the often-dangerous jobof wildland firefighting safer. Finally, wewill develop a series of recommenda-tions for implementing the changesneeded to improve firefighter safety.

The session begins with four presenta-tions that outline the psychological andorganizational aspects of wildlandfirefighting. These talks will set the tonefor the session as we focus on theindividual firefighter, fire crews, andorganizational structures on the fireline.We hope the unique mix of professionalswill create a synergism that leads tomeaningful change and a safer firefight-ing environment.

Ted PutnamWorkshop Organizer

Dave ThomasWorkshop Moderator

Jerry MeyerWorkshop Facilitator

24

Part 4 of 4 Appendix B—Agenda

MONDAY, JUNE 12 0800–1200

Keynote Presentations(open to the public)

0800–0900Behavioral Bases of Accidents andIncidents: Identifying the CommonElement in Accidents and IncidentsCurt Braun, Ph.D., University of Idaho

Human behavior plays the largest rolein firefighter safety. Faced with theknown and essentially constant risks ofa wildland fire, human behavior is theonly factor that can greatly increase ordecrease the risk of injury. Despite itslarge role in safety, however, behavioris frequently overlooked during accidentinvestigations. Endeavoring to discoverthe cause of a workplace injury orfatality, investigators often focus on thespecial environmental circumstancesand not on the behaviors that precededthat accident. This overemphasis oncircumstances fails to consider the factthat the vast majority of accidents resultnot from the environment but fromknown risky behavior that is part ofcommon work practices.

To address safety challenges, specificconsideration must be given to theelement common to all accidents:human behavior. A behaviorally basedsafety program can reduce the risksassociated with wildland firesuppression by: (1) identifyingantecedent behaviors that lead toaccidents; (2) determining thefrequency of these behaviors; (3)evaluating the training programs andmanagement systems that eitherdirectly or indirectly support thebehaviors; and (4) developing a trainingand management remediation programaimed at changing behavior.

0900–1000Recognition Primed DecisionStrategiesGary Klein, Ph.D., Klein Associates

Studies of firefighters show that theyrely primarily on Recognition-PrimedDecision (RPD) strategies as opposedto sifting through alternatives andcomparing how they rate on differentevaluation dimensions. The RPD modelexplains how people can makedecisions under conditions of timepressure and ambiguity and shiftingconditions. Specifically, the modelexplains how experienceddecisionmakers can generate areasonable course of action withouthaving to contrast alternatives, and howthey can evaluate a course of actionwithout comparing it to alternatives.

We can also use the RPD model tounderstand some of the errors that canarise in naturalistic settings. Theseprimarily stem from inadequateexperience bases. In turn, these errortypes suggest some strategies fordecision-centered training.

1000–1100The “Cultural Inertia” Impacts ofTeam DecisionmakingDavid Hart, TIG, Inc.

Cultural attitudes permeate thedecisionmaking of teams working withinthe organization. “Anytime, Anywhere,”“Can Do,” and “Make It Happen” areexamples of adopted cultural attitudesthat have both assisted and (in somecases) inhibited crew effectiveness.This discussion investigates theimpacts of cultural-based attitudes asbarriers to individual and crewdecisionmaking processes. It includesdiscussion of lessons learned fromother high-risk/high-threatenvironments in establishing a non-attribution/non-retribution environment,and overturning cultural attitudebarriers within the individual and theorganization as a whole.

1100–1200South Canyon Revisited: Lessonsfrom High ReliabilityOrganizationsKarl Weick, Ph.D., University ofMichigan

There is an emerging body of work thathas begun to describe howorganizations that face the possibility ofcatastrophic error every day, cope withthis prospect. These organizations,referred to as high reliabilityorganizations, include nuclear powerplants, air traffic control systems,aircraft carriers, flight crews, andchemical plants. Several issues that arediscussed in studies of theseorganizations are similar to issues thathave surfaced in discussions of theSouth Canyon Fire incident on July 6,1994. The purpose of this presentationwill be to discuss some of thesesimilarities with special attention beingfocused on issues of communication,group structure, stress, mindsets,leadership, and sense making.

1200–1300: LunchClosed Workshop Session Begins(limited to invited participants)

1300–1700: Focus will be on thedynamics of individual decisionmakingand individual firefighter experience

1830–: No-host bar and dinner (if thereis sufficient interest)

TUESDAY, JUNE 13 0800–1200: Focus on interaction between crew members

1200–1300: Lunch

1300–1700: Focus will be oninteractions between crews andorganizational structure

25

Part 4 of 4

WEDNESDAY, JUNE 14 Trip To Mann Gulch

0545: Assemble at Village Red Lion Innand board bus

0600: Depart Missoula

0600–0800: Workshop discussionsenroute

0830–1800: Mann Gulch guided tour

On August 5, 1949, a wildfire overran16 firefighters at Mann Gulch. Therewere only three survivors. Significantcontroversy has surrounded this fire,including firefighter decisions andactions as well as the ensuingentrapment investigation. On July 6,1994, a wildfire overran 49 firefightersat South Canyon. There were 35survivors and 14 fatalities. Manyinvestigators believe the two events areconnected and ask how much have welearned in the intervening 45 years. Atrip to Mann Gulch has been planned toexplore that connection and fosterfurther workshop dialogue in a “real life”setting.

Mann Gulch is about 150 miles east ofMissoula and 25 miles north of Helena,Montana. A bus will leave the VillageRed Lion Inn parking lot promptly at0600. It is about a 2-1/2 hour drive tothe Gates of the Mountain Marina andabout a 30-minute boat ride down theMissouri River to Mann Gulch. We arepaying for the bus, but each participantwill need to pay about $10 for the boat

ride (round-trip). We will return to themarina around 1730-1800 and will beback in Missoula by about 2000. Eachparticipant should bring a sack lunchand a canteen of water as well assnacks for the return trip. In addition,bring clothing appropriate for theweather, as well as other items desiredsuch as cameras, daypacks, and soforth.

We plan to have two guidesknowledgeable about Mann Gulchavailable to retrace the events. An EMTwith a first aid kit and radio will alsoaccompany us.

The walk up Mann Gulch is about 1-1/2miles over grassy, rocky ground. Sturdywork or hiking boots (well broken in)are strongly recommended. The slopeis steep, but people of varied fitnesslevels have tackled it successfullywalking at their own pace. Snakes andfooting are the only other hazards,though snakes are rare. The trip will becanceled if rain or strong winds areforecast. More information will beprovided at the workshop.

1800–2000: Return trip to Missoula

THURSDAY, JUNE 15 0800–1200: Focus on future research, investigations, and training

1200–1300: Lunch

1300–1700: Workshoprecommendations

FRIDAY, JUNE 16 0800–1700: Workshop volunteers finish write-ups on findings and recommendations

Suggested Reading:

Young Men and Fire. Norman Maclean.1992. University of Chicago Press.Chicago, IL.

Fireline: Summer Battles of the West.Michael Thoele. 1995. FulcrumPublishing, Golden, CO.

Mann Gulch Fire: A Race That Couldn’tbe Won. Richard Rothermel. 1993. INT-GTR -299. U.S. Department of Agricul-ture, Forest Service, IntermountainResearch Station, Ogden, UT.

*Mann Gulch Fire: A Race That Couldn’tBe Won. Richard C. Rothermel.

*Available on the worldwide web at:http://www.xmission.com/~int/pubs.html

For a printed copy, contact:

Tom Cook, Program AssistantFire Behavior ProjectIntermountain Fire Sciences Laboratory5775 Highway 10 WestMissoula, MT 59802Phone: (406) 329-4820Fax: (406) 329-4825DG: T.Cook:S22L01A

26

Part 4 of 4 Appendix C—Participants

Participants

Dave Aldrich— branch chief forground operations safety, Forest ServiceWashington Office, Fire and AviationManagement. Dave began his ForestService career as a seasonal employeein 1958 on the Powell District in R-1. Hehas worked in R-1 and R-3 in fire man-agement jobs as well as the NationalOffice and at the Intermountain FireSciences Laboratory in Missoula. He hasbeen a fire behavior analyst on nationalfire teams and has been involved withseveral national fire training courses.Dave chairs the NWCG Safety andHealth Working Team. He has a BS inforestry from the University of Montana.

Bill Bradshaw— works for the ForestService Washington Office in fiscal andaccounting, specializing in incidentadministration and claims. Bill has beenactive with decision analysis projects inthe past and is currently involved withefforts to enhance wildland firefightersafety through improved attitudes,leadership, and responsibility.

Curt Braun, Ph.D.— is an assistantprofessor of psychology at the Universityof Idaho. As an ex-firefighter for theSawtooth National Forest, he evaluatesfire suppression from the firefighter’sperspective. He recently coauthoredHuman Decisionmaking in the FireEnvironment, which will appear in anupcoming special issue of Fire Manage-ment Notes. He holds a Ph.D. in humanfactors psychology with an emphasis onhuman performance from the Universityof Central Florida.

Jim Douglas— is Director of the Officeof Hazard and Fire Programs Coordina-tion for the U.S. Department of theInterior. Before coming to that post, heserved as the Interior Department’s fireprogram coordinator. His career withInterior began in 1979 in the Office ofPolicy Analysis. He was also in theDepartment’s Office of Budget for 7years. He was on the Interagency

Management Review Team followingthe South Canyon Fire and serves onthe Federal Wildland Fire Policy andProgram Review. He has an undergrad-uate degree in political science fromGrinnell College and a master’s degreein public policy from the University ofMichigan.

Jon Driessen, Ph.D.— is a professorof sociology at the University of Mon-tana. He also holds a faculty affiliateappointment at the Missoula Technologyand Development Center. He specializesin the sociology of work and for the past12 years has studied the culture of workin Forest Service field crews. His latestproject for the Forest Service is a 48-minute video, Making a Crew . Jon hasa doctorate in sociology from theUniversity of Colorado.

Kelly Esterbrook— is currently asmokejumper squadleader with theForest Service, in Redmond, OR. Kellystarted her Forest Service career on theRogue River National Forest in 1978.She spent 2 years on Rogue Riverengine crews and 2 years as a RogueRiver Hotshot crewmember. She thenspent four seasons on the DeschutesNational Forest as an engine foremanand one season with the RedmondHotshots. She began jumping in 1986.She was detailed to the Union Hotshotsin 1992 as superintendent. In 1994 shecompleted Technical Fire ManagementTraining.

Paul Gleason— is currently fire ecolo-gist for the Roosevelt and ArapahoNational Forests in Northern Colorado.His emphasis is the restoration and useof fire as a natural process to achieveland management goals in the centralRocky Mountain ecosystems. From 1991to 1994, he was fire management officerfor the Estes-Poudre and RedfeatherRanger Districts in Northern Colorado.Prior to that time Paul spent 23 fireseasons with the Interagency HotshotCrew programs on the Angeles, Mt.Hood, Okanogan, Pike and San Isabel

National Forests. Because of hisextensive wildland fire suppressionexperience throughout the U.S., Paulhas been active in fire suppression/firebehavior course development andpresentation. He has an undergraduatedegree in mathematics from ColoradoState University and is pursuing gradu-ate studies in fire ecology and effects atColorado State University.

Dave O. Hart— has extensive experi-ence as in instructor and facilitator in thedelivery of crew resource managementtraining. He is a former Air Force B-52instructor navigator, and instructor atthe Air Force’s Undergraduate NavigatorTraining. He served as lead facilitatorfor Hernandez Engineering, the crewresource management contractor for theAir Force Air Mobility and Air CombatCommands. He is an Air Force ReserveC-130 navigator for the 731st AirliftSquadron, which provides wildlandfirefighting support through the MAFFS(Modular Airborne Fire Fighting System)program. Dave cofounded TIG Inc.,where he works as a training consultantand facilitator. He is currently responsiblefor assessment, design, development,and delivery tasks associated with thenew Army National Guard SpecialForces Decision Training Program. Hereceived his bachelor’s degree in aircraftmaintenance engineering from ParksCollege and is pursuing a master’sdegree in aerospace studies fromEmbry-Riddle Aeronautical University.

Jerry Jeffries— has spent his entireForest Service career in fire and safety.He recently was named safety projectleader at the Missoula Technology andDevelopment Center. From 1990 to1995 he served as safety and healthmanager for aviation and fire manage-ment in R-1. Before that time, he wasfor many years safety manager on theBitterroot National Forest. He has held avariety of fire positions during his career,including interregional hotshot crewsupervisor, division supervisor, line boss,air attack boss, and air support groupsupervisor. In 1992 he received the

27

Part 4 of 4Government Employees Insurance Co.(GEICO) public service award for fireprevention and safety from a group ofover 200 nominees worldwide.

Jim Kautz— is a videographer/photog-rapher at the Missoula Technology andDevelopment Center. He began his fire-fighting career at Darby Ranger Districtand was a smokejumper for 3 years inR-1. For the past few years one of Jim’sprimary responsibilities has been toprovide photo documentation as part ofwildland fire entrapment investigations.Jim holds a degree in film and televisionfrom Montana State University.

Gary A. Klein, Ph.D.— is chairmanand chief scientist of Klein Associates.He has performed research on natural-istic decisionmaking in a wide variety oftask domains and settings, and hasdeveloped significant new models ofproficient decisionmaking. His researchinterests include the study of individualand team decisionmaking under condi-tions of stress, time pressure, anduncertainty. He has furthered the devel-opment and application of a decision-centered approach to system design andtraining programs. He has also studiedapplications of case-based reasoning fordomains such as the cost/benefit evalu-ation of training devices and developingmarketing projections for new products.He holds a doctorate in experimentalpsychology from the University ofPittsburgh.

Buck Latapie— is currently the firetraining and safety officer for R-6. Hehas served continuously on incidentmanagement teams since 1983 in plans,operations, and as an incident com-mander. Early in his Forest Servicecareer he served as a hotshot foremanand engine foreman. In 1978 he washotshot superintendent on the BitterrootNational Forest. He later worked as asilviculturist on the Fremont NationalForest, district fire management officeron the Deschutes National Forest, andas a forest aviation and fire management

officer on the Ochoco National Forest. Heholds a bachelor’s degree in forestry/firemanagement from the University ofMontana.

Mark Linane— is the Los Padres Hot-shot superintendent on the Los PadresNational Forest in Region 5. The crewis located at the Santa Barbara RangerDistrict office compound 10 miles northof Santa Barbara, CA. Mark has 30years of wildland fire experience, the last23 as superintendent of the Los PadresHotshots. He is considered a leadingspokesperson for the hotshot commun-ity. He has been involved with safety andtraining issues for years, most recentlyworking on the revision of the StrikeTeam/Task Force Leader trainingcourse.

Lark S. McDonald— has performedassessment, development, and designwork in human factors training programsfor a wide variety of aviation-basedapplications, including aeronauticaldecisionmaking and cockpit manage-ment for civilian pilots. He has servedas designer and developer for crewresource management programs for theNavy T-45, Air Force T-1, and commer-cial MD-80 for McDonnell Douglas Corp.He has worked as a developmentprogram manager for United Airlines andMartin Marietta, and as the lead instruc-tional designer for HernandezEngineering, the crew resource manage-ment contractor for the Air Force AirMobility and Air Combat Commands. Hisrecent work has included assessmentand adaptation of CRM training for usewith Air Force test pilots and theirground-based engineer and logisticcounterparts. In a further extension ofmoving CRM-type training into high-risk,high-stress environments, he recentlycofounded TIG Inc. with David Hart,which currently provides decisional train-ing and leadership programs for teamswith the National Guard Special Forces.He received his education in aviationmanagement and psychology fromMetropolitan State College, Denver, CO.

Robert J. Martin— is the ForestService national aviation safety andtraining manager at the National Intera-gency Fire Center, Boise, ID. Bob’saviation experience covers the fields ofmaintenance, accident investigation,piloting, and program management. Forthe past 30 years he has been employedin the military, commercial, and publicsectors of aviation. His Forest Servicecareer began in 1977 in R-3. Since thattime, he has served at national firecenter and R-6, Portland, OR. During1987-1988, he worked with U.S.Customs air interdiction program andreturned to the Forest Service in 1989.Bob received his BA in aviation manage-ment from Boise State University andhis MA in aviation management fromEmbry-Riddle Aeronautical University.

Jerry Meyer— has worked for theForest Service since 1971 in a numberof capacities, primarily in timber man-agement. He has also worked as afirefighter, wilderness guard, writer/editor,and historian. His most advanced red-card qualification is division/groupsupervisor, but he most often takes fieldobserver assignments. Jerry will facili-tate the workshop discussions. He holdsa BA in history/political science from theUniversity of Montana.

Dave Pierce— is currently the smoke-jumper project leader at the MissoulaTechnology and Development Center,a position he has held since 1980. HisForest Service career began in 1964 asa “smokechaser” on Red River RangerDistrict in Idaho. From 1965 through1968, he was a smokejumper in R-6 andR-1. Between 1969 and 1971, he workedin private industry as a commercial pilot.In 1971 he returned to firefighting as asmokejumper with the BLM. With 30years of experience working with boththe Forest Service and BLM smokejump-ing programs, Dave has accumulatedsome “street smarts” about initial attackfirefighters. During his years with theBLM Alaska smokejumping program, hewas responsible for organizing smoke-jumper crews for safety and effective-

28

Part 4 of 4ness. At MTDC, he has finished severalprojects related to safety in smoke-jumping/aviation operations where theobjective was to develop materialsdesigned to change institutional attitudes.

Ted Putnam— is a fire and safetyequipment specialist at the MissoulaTechnology and Development Center.He started working for the ForestService in 1963 and spent 3 years ondistrict fire crews, 8 years as an R-1smokejumper, and 3 years as a super-visory smokejumper. In 1976 he came toMTDC. He is responsible for developingfirefighter’s protective clothing and fireshelters, including training materials. Heis a member of two National Fire Protec-tion Association standards-settingcommittees for protective clothing andequipment. Ted holds a Ph.D. in experi-mental psychology from the Universityof Montana.

Jim Saveland— is a fire ecologist forthe Forest Fire and AtmosphericSciences Research Staff, Washington,D.C. He began his Forest Service careerin 1978 on a district fire crew in Elk City,ID. Jim spent 4 years as a smokejumperin R-6 and R-1. In 1984 he became firemanagement officer on the Moose CreekRanger District. In the incident commandsystem, Jim was a division/group super-visor and a fire behavior analyst. In 1988he transferred to the Southern Fire Labin Macon, GA. Jim became projectleader in 1991. In 1994, he moved to his

present position in the WashingtonOffice. He has taught several classes onvarious aspects of fire and risk manage-ment at the University of Idaho and atthe district, forest, regional, and nationallevels of the Forest Service. He is theunit leader for the Risk Management andDecision Analysis unit of the NationalInteragency Prescribed Fire BehaviorAnalyst course taught at the NationalAdvanced Resource Technology Centerin Marana, AZ. The Interagency Man-agement Review Team for the SouthCanyon Fire asked Jim to lead a team todevelop a report on the collection, distri-bution, and utility of live fuel moistureinformation. Jim has a BS in mathe-matics from Auburn University, and anMS in forest resources and a Ph.D. inforestry, wildlife, and range sciencesfrom the University of Idaho. His Ph.D.work concentrated on the application ofartificial intelligence, decision science,and cognitive psychology to firemanagement.

Lyle Shook— is currently safety andhealth manager for R-5. He has 21 yearsof experience in Forest Service wildlandfire operations in Regions 3, 5, and 6.His experience ranges from hotshot andhelitack crews to acting Regional firecoordinator. In the incident managementsystem he is a type I operations chief,plans chief, and safety officer. He hasbeen a type II incident commander for 3years. He has been in his currentposition since 1988.

David A. Thomas— is fire manage-ment officer on the Superior RangerDistrict, Lolo National Forest. He startedhis Forest Service career as an emer-gency firefighter in 1967. He has beena member of fire crews on the Kootenaiand Clearwater National Forests. Later,he was helicopter foreman of an 18-person crew. Dave has been a memberof numerous type I and type II incidentmanagement teams. He was a firebehavior analyst on the 1988 fires inYellowstone National Park. As aprescribed fire manager, Dave has devel-oped and implemented many prescribedburns ignited for various silvicultural andfuels management objectives. Daveholds a BA in geography from theUniversity of Montana.

Karl E. Weick, Ph.D.— is the RensisLikert Collegiate Professor of Organiza-tional Behavior and Psychology at theUniversity of Michigan. He is also theformer editor of Administrative ScienceQuarterly, the leading research journalin the field of organizational studies. Hestudies such topics as how people makesense of confusing events, the socialpsychology of improvisation, high relia-bility systems, the effects of stress onthinking and imagination, indeterminacyin social systems, social commitment,small wins as the embodiment ofwisdom, and linkages between theoryand practice. His writing about thesetopics is collected in four books, one ofwhich—The Social Psychology ofOrganizing —is cited as furnishing

29

Part 4 of 4significant background for Peter’s andWaterman’s In Search of Excellence.Karl has consulted with a variety oforganizations, including Corning Glass,Narco, Cole Products, Dalton Foundries,Southland Corp., Motorola, Texas Instru-ments, Lockheed, the National ScienceFoundation, the National Institute ofEducation, and the National Institute ofMental Health. He has a Ph.D. inpsychology from Ohio State University.

Pat Wilson— is manager of theGrangeville smokejumper base, a posi-tion he has held since 1987. He startedhis firefighting career in 1974 on anengine crew with the Idaho Departmentof Lands. He became engine crewforeman in 1976. In 1978 he was anassistant foreman of the now-defunctCoeur d’Alene Hotshots. The nextseason he joined the St. Joe Hotshotsas a lead sawyer. He began smokejump-ing in 1980 in Missoula, transferring toGrangeville in 1981. He became a squadleader in 1983. He served for 2 years onthe forest safety committee and currentlyis a member of the National AerialDelivered Firefighter Study, and a groupthat is rewriting the Smokejumper andParacargo Handbook .

Patrick Withen— a smokejumperbased in McCall, ID, he is assistantprofessor of sociology at CentenaryCollege, Shreveport, LA. His fire experi-ence includes 14 seasons as a smoke-jumper, 1 season on a hotshot crew, and2 seasons on a helitack crew. As a forest

sociologist he spent 1 year conductingbaseline social data collection and socialimpact analysis for landscape analysesand environmental impact statements.He has been a college instructor for 5years. Patrick has a Ph.D. in sociologyand an MBA from Boston College. Healso holds a BS in psychology from theUniversity of Oregon.

Steve Wolf— is a research associateat Klein Associates. He has played a keytechnical role on projects concerned withexpert knowledge and decision support.He was the project leader on a recentlycompleted effort sponsored by the Navyto develop a decision support system forcrew members in a combat informationcenter. He heads a related effort toexamine potential training applications.His current projects include a review ofNational Fire Academy curriculumdesigned to enhance rapid decisionmak-ing on the fireground. He has been amember of a technical team studyinghelicopter pilot safety, allocation methodsused by fire direction officers, and reviewof human-computer interface designs fora surveillance platform developed jointlyby the Army and Air Force. He holds aBS in psychology from Wright StateUniversity, Dayton, OH.

Special thanks to these people for theirassistance during the workshop:

Laurel Chambers, workshop note-taker, Superior Ranger District, LoloNational Forest, Superior, MT

Tim Crawford, Gates of the MountainsMarina, by Helena, MT

Mary Jo Lommen, Mann Gulch EMT,Superior Ranger District, Lolo NationalForest, Superior, MT

Dave Turner, Mann Gulch interpreter,Helena Ranger District, Helena NationalForest, Helena, MT

The following people were invited to theworkshop but were unable to attend:

Paul Broyles, training and safetyspecialist, National Park Service,NIFC, Boise, ID

Jim Cook, superintendent, ArrowheadHotshots and training specialist,National Park Service, NIFC, Boise, ID

Mary Jo Lavin, director, Fire andAviation Management, USDA ForestService, Washington, D.C.

Holly Maloney, squadleader, LoloHotshots, Lolo National Forest,Missoula, MT

Stan Palmer, safety and health groupleader, Bureau of Land Management,NIFC, Boise, ID

Gina Papke, superintendent, Zig ZagHotshots, Mount Hood National Forest,Zig Zag, OR

Bill Russell, acting director, Aviationand Fire Management, Region 3,Albuquerque, NM

Jerry Williams, director, fire operations,Fire and Aviation Management, USDAForest Service, Washington, D.C.

30

Part 4 of 4 Appendix D—Keynote Presentations

Addressing theCommon BehavioralElement in Accidentsand IncidentsCurt C. Braun, Department of Psychology,University of Idaho

Virtually every college student hasfaced the philosophical question, “If atree falls in the woods and no one isthere to hear it, does it make a sound?”The answer of course is no; the fallingtree does not make a sound. Whilemany people struggle with this answer,it is important to remember that theanswer relies, not on the physicsassociated with a falling tree, but ratheron the definition of sound. Sound is asubjective sensation created when theear is stimulated by changes in thesurrounding air pressure. Given thisdefinition, a tree falling in the woodsmakes no sound when an ear is notpresent. A comparable safety questionmight be, “If there is a snag in the woodsand there is no one there, does it posea risk?” Again, the answer would be no.As with the sound example, the answercenters not on the physics of a fallingtree, but rather on the definition of risk,a chance of loss or injury to a human.In the absence of a human, a fallingsnag creates no threat of injury or loss.Although this relationship appearsobvious, it is important to realize thatthere are two components to thisquestion: the snag, and the presence orabsence of the human. Both play a rolein creating a risky situation.

If an individual is injured by a fallingsnag, clearly both had to be present.This situation can easily be representedby the following model:

Environmental Hazard (Snag) +Human = Accident

The role of the snag and the individualin this situation are significantly different.The fact that the snag will eventuallyfall is well known and in contrast to theactions of the human, represents arelative constant. We know that the

snag will eventually fall, but not when.If the environmental hazard remainsessentially constant, only onecomponent is left to vary: the actions ofthe human.

The level of risk created by the snag canbe mitigated or exacerbated by thebehavior of the individual. Injury andloss are more likely when the individualfails to attend to the known risks. Whenthe individual is struck by the fallingsnag, the proximate cause is apparent,inattentiveness. It is not apparent,however, that this was an isolated caseof inattentiveness. This inattentivenessmight represent a general pattern ofbehavior that places the individual atrisk in a variety of situations. Toadequately respond to the accident,consideration must be given to both theproximate cause and the behavioralpattern. Unfortunately, traditional safetyprograms have placed far more empha-sis on the former than on the latter.

Human Behavior andAccidents

Few will argue that most accidents andmishaps are directly related to unsafebehaviors. A review of the national airtraffic control system revealed that 90%of the committed errors could be directlylinked to human inattentiveness, poorjudgment, or poor communications(Danaher, 1980). Mansdorf (1993) listsnine different causes of accidents andattributes all of them to human error inthe form of inadequate training,supervision, and management. Giventhis consensus, the solution is simple;change the behavior where theaccidents occur. Despite the intuitiveappeal of this approach, efforts toincrease safety in this manner often failto produce the anticipated reductions inaccidents. These failures occur becausetraditional safety programs generallyfocus on the unique circumstances andrisks that, like the snag, remain relativelyconstant. Moreover, these programsoften do not consider the broad

spectrum of situations where the samebehavior can also result in an accident.

Krause and Russell (1994) suggest thataccidents result, not from uniquecircumstances or behaviors, but fromthe intentional display of risky behaviorsthat occur with such regularity that theyhave become common practice. Theseauthors contend that an accidentrepresents an unexpected result of anunsafe act that has become part of theworking culture. Despite the best effortsto mandate safety, risky behaviorsincreasingly become acceptablepractice each time they are performedwithout negative consequences. Theprocess is similar to that seen inindividuals who interact with hazardousproducts. Safety researchers havefound an inverse relationship betweensafety behavior and familiarity(Goldhaber & deTurck, 1988). Theprobability that an individual will complywith safety guidelines decreases asfamiliarity with the product increases.

Wildland firefighters are not immune tothis process. In response to the SouthCanyon fire of 1994, Rhoades (1994)writes, “And sometimes, even often, therisks we take in doing our jobs includeviolating the 10 Standard Fire FightingOrders or ignoring the 18 Situationsthat Shout Watch Out .” He furtherwrites, “Nonetheless, very seldom doesour inability to comply with the orderscause us to abandon our tasks...”Rhoades’ statements reflect the fact thatit is possible to violate standard safetypractices without the worry of negativeconsequences. More importantly, how-ever, Rhoades’ comments suggest thatthe violations have occurred with suchgreat regularity that they have becomeaccepted practice in wildfire suppression.

Accident Prevention Froma Behavioral Perspective

An effective prevention program beginsby understanding that accidents oftenreflect the unfortunate outcome of

31

Part 4 of 4hazardous acts that have becomecommon practices and that thesepractices frequently span a multitude ofdifferent job tasks. To be effective, asafety program must: 1) identify the ante-cedent behaviors that result in accidentsand near-miss incidents; 2) determinehow frequently these behaviors occur;3) evaluate training and managementprograms; 4) provide consistent andactive feedback and reinforcement, and5) develop remediation plans.

Identifying Antecedent Behaviors.Traditional accident investigations tendto be very myopic, focusing only on thecircumstances immediately involved inthe accident. The purpose of an investi-gation is to identify the accident’s causewith the aim of creating new procedures,equipment, and standards to eliminateor at least minimize the risk (Mansdorf,1993). This investigative approach,however, must go beyond the traditionalmicroscopic analysis to identify behav-iors that are common in a variety ofaccidents. To facilitate the identificationof these behaviors an investigation teamshould be composed of individuals fromall levels of the work force (Krause &Russell, 1994; Mansdorf, 1993).Moreover, efforts should be taken toreconstruct the accident with the aim ofidentifying the underlying behavioralpatterns that might have precipitated it.Once identified the investigation needsto assess the extent to which thesebehaviors have been present in otherincidents or accidents. Finally, theinvestigation must assess the degree towhich the actions reflect the acceptanceof hazardous and risky behavior ascommon practice.

Assessing the Frequency. To assessthe frequency of unsafe acts, a systemfor reporting accidents, and near-missaccidents must be created. Near-missesplay an important role in assessing thefrequency of risky acts. From thebehavioral perspective, near-missesrepresent accidents without the conse-quences (Krause & Russell, 1994).Moreover, given that unsafe behaviorsinfrequently result in accidents, near-

misses can provide better insight intoemployee safety. Mansdorf (1993)reports that for every serious industrialaccident there are approximately 10minor accidents, 30 property damageaccidents, and 600 near-miss accidents.

The overarching motivation driving areporting system should be the acquisi-tion of reliable and valid data. To facili-tate this process, the reporting systemmust encourage reporting from all levelsof the work force. Moreover, individualsshould be instructed as to their reportingresponsibilities. With regard to thelogistics of the system, every reasonableeffort should be taken to reduce the costof complying with reporting requirements.These efforts might include simplifyingreporting forms, the use on-site or tele-phone based interviewers to whomunsafe acts can be reported, the use ofanonymous data collection systems, thecreation of safety surveys, the use oftrained field observers, or the use ofautomated data collection systems.Such reporting programs might alsoguarantee immunity from disciplinaryactions for individuals who report.

Evaluating Training and Management.There are a variety of questions thatmust be asked when evaluating trainingand management. Are instances of thedesired behavior demonstrated duringtraining? For example, fire sheltertraining has traditionally placed moreemphasis on getting into the shelter thanon other factors such as situationalawareness, site evaluation, groundpreparation, and contingencies all ofwhich are essential to a successfulshelter deployment. Are employeestrained in the selection of the appropriatebehavior? Invariably more than oneoption is available for each situation. Ina situation where a burnover is inevit-able, a firefighter can deploy a fireshelter or attempt to escape. Factorsthat influence this decisionmaking pro-cess must be considered in advance.Training should include techniques andprocedures used to evaluate the variousoptions. Is there a system to continuetraining apart from the classroom? On-

the-job-training (OJT) is a widely usedtechnique but it suffers from manyshortcomings. Trainers are frequentlyunaware of instructional techniques,training occurs only when time is madeavailable, the situation typically dictateswhat skills are learned, and traineesoften take a passive role merely watchingand not demonstrating behavior(Gordon, 1994). Managers and super-visors must assess the extent to whichtraining relies on OJT and developspecific programs to maximize itsusefulness.

After training, are the behaviorspracticed? Just as firefighters exerciseto maintain a level of physical fitness,skills learned in training must bepracticed to ensure competency. In arecent article on decisionmaking in thefire environment Braun and Latapie(1995) noted that training should includethe rehearsal of behaviors that areneeded in stressful conditions. Safetycritical behaviors must be practiced untilthey become automatic. Finally, whatis the perceived priority of safety? Dosupervisors and managers expect safebehaviors? Are firefighters asked towork in high-risk conditions that areoutside of safe parameters? Is there anestablished code of conduct that speci-fies the safe behaviors an individual isexpected to display? Finally, is therean accountability system to which allfirefighters are held? The answers tothese and other questions provide anindication of the priority safety is given.

Feedback and Reinforcement. Theconcepts of training and reinforcementare closely related. At its most basiclevel, training serves to educate an indi-vidual about the various reinforcementcontingencies (Anderson, 1995). Thatis, during training an individual learnsthe actions and behaviors that will bereinforced when training is complete.After training is complete, are the trainedbehaviors expected and reinforced?Moreover, have the trained behaviorsbeen directly or indirectly extinguishedby example or directive? For example,are firefighters more often reinforced for

32

Part 4 of 4taking risks than for demonstratinggood judgment?

While it is important to assess if trainedbehaviors have been reinforced, it isjust as important to determine if unsafebehaviors have been inappropriatelyreinforced by environmental events.Although the ultimate goal of firefightingis fire suppression, a suppressed fire isnot an appropriate reinforcer for firefight-ing behavior. This unsuitability stemsfrom the fact that all fires eventually goout independent of the actions taken byfirefighters. This inevitability makes firesuppression an indiscriminate reinforcer.That is, fire suppression could reinforceboth safe and unsafe behaviors. Somewould agree that factors such asweather often play a larger role in sup-pression than firefighters, but still arguethat firefighters should be reinforced bythe fact that the size of the fire has beenlimited. There might be some truth inthis statement, however, it is notcompletely verifiable because firefightersoften take advantage of areas where thefire would stop on its own (e.g., naturalfuel breaks).

Care should be taken in determiningthe types of reinforcement and feedbackindividuals obtain from the environment.The containment and suppression offires, the saving of structures andresources, and other similar eventsmake poor reinforcers because they areindiscriminate and because they targetthe outcome of behavior and not thebehavior itself. Efforts must be made toreinforce the safe behaviors independentof the outcomes.

Remediation Plans. Shortcomings intraining, supervision, or managementshould not be viewed in isolation but asrepresentative of a company-widepattern of behaviors. Efforts to remedi-ate these shortcomings must endeavorto address both the specific behaviorsand the broader culture. Each planshould identify short-term and long-termobjectives and the criteria against whichthe plan will be evaluated.

Conclusions

Programs aimed at enhancing safety byaddressing the proximate cause of anaccident only consider a small portion ofthe safety picture. Merely addressingthe proximate cause fails to consider thatthe system either directly or indirectlytrains, reinforces, and even expectsemployees to demonstrate hazardousbehavior. An effective safety programmust consider both the proximate causeand the working environment thatpromotes hazardous behavior. Theprogram must identify unsafe behaviorsand assess their prevalence. It mustevaluate training to ensure that individ-uals not only gain the necessary skillsbut are provided with opportunities toexercise and practice those skills. Thesafety program must survey supervisorsand managers to determine if skillslearned in training are actively reinforced,and finally, it must make recommenda-tions that affect behaviors and thesystem that supports them.

References

Anderson, J. R. (1995). Learning andmemory. New York: Wiley & Sons.

Braun, C. C. & Latapie, B. (1995).Human decisionmaking in the fireenvironment. Fire Management Notes,55, 17-20.

Danaher, J. W. (1980). Human error inATC system operations. HumanFactors, 22, 535-545.

Goldhaber, G. M., & deTurck, M. A.(1988). Effects of consumers’ familiaritywith a product on attention to andcompliance with warnings. Journal ofProducts Liability, 11, 29-37.

Gordon, S. E. (1994). Systematictraining program design: Maximizingeffectiveness and minimizing liability.New Jersey: Prentice Hall.

Krause, T. R., & Russell, L. R. (1994).The behavior-based approach toproactive accident investigation.Professional Safety, 39, 22-26.

Mansdorf, S. Z. (1993). Completemanual of industrial safety. New Jersey:Prentice Hall.

Rhoades, Q. (1994, August 26). StormKing: Effective fire fighting calls forbending the rules sometimes. TheMissoulian, pp. A5.

33

Part 4 of 4

*Presented at the U.S. Forest Service Conference, June 12, 1995.

Naturalistic DecisionMaking and WildlandFirefighting*Gary Klein, Ph.D., Klein Associates Inc.582 E. Dayton-Yellow Springs RoadFairborn, OH 45324(513) 873-8166August 8, 1995

The Recognition-Primed DecisionModel describes what people actuallydo when they make difficult decisions.This has many implications for trainingand helping people make decisionsunder stressful situations. It can alsohelp explain the factors behind baddecisions.

The standard method of decision makingis the rational choice model. Under thismodel, the decision maker generates arange of options and a set of criteria forevaluating each option, assigns weightsto the criteria, rates each option, andcalculates which option is best. This isa general, comprehensive, andquantitative model which can beapplied reliably to many situations.Unfortunately, this model is impractical.People making decisions under timepressure, such as fire fighters, don’thave the time or information to generateoptions and the criteria to rate eachoption.

The rational choice model is also toogeneral. It fits each situation vaguely,but no situation exactly. The worst newsis that in studies in which people havebeen asked to follow the rational choicemodel exactly, the decisions they comeup with have been worse than decisionsthey make when they simply use theirown experience base. This model is oflittle value to training because it doesnot apply to most naturalistic settings orto how people actually make decisionswhen faced with complex situationsunder time pressure. Decision aidswhich have been produced to assist withthe application of the rational choicemodel have been largely ineffective.Because of these drawbacks, a field

emerged called Naturalistic DecisionMaking (see Table 1). This fieldemerged because governmentalsponsors such as NASA, FAA, themilitary, and others, realized that theyhad spent a lot of money and builtdecision models that did not work in thefield. They wanted to get away frombuilding analytical models which didn’twork when they were brought intoaction. Naturalistic Decision Makinguses expert decision makers, and triesto find out what they actually go throughin their decision making process.

Instead of restricting decision making tothe “moment of choice,” experts areasked about planning, situationalawareness, and problem solving to findout how these all fit together. This modelis used to understand how people facedecisions in shifting and unclearsituations and under high stakes. Teaminteractions and organizationalconstraints with high stakes are alsoused as factors. For years,researchers had been simply askingcollege sophomores what they would dogiven a set of options, and a clear goal.For Naturalistic Decisionmakingresearch, experts are asked to size upactual situations, using all cues andconstraints to set goals and makedecisions.

The first study I performed to generatemodels and training recommendationsfor decision making under pressure andcertainty was a study for the Army. TheArmy Research Institute wanted somedata on decision making in real, stressfulsituations, and I thought that urbanfirefighters would be a good example ofpeople who had become experts atmaking such decisions. We studiedcommanders who had about 20 yearsof experience, and studied the mostdifficult cases they had. Of the cases westudied, there was an average of fivechanges in the fire and in the way it hadto be handled. About 80 percent of thedecisions were made in less than aminute. As we started the study, wefound that each expert firefighter told usthat they had never made any decisions.They explained that they simply followedprocedures. But as we listened, werealized that in each case, there wasone option which they thought of quickly.They evaluated that one option, and if itseemed viable, they went ahead with it.

We began to wonder how they came upwith that first option and how they wereable to evaluate one option withoutothers for comparison. The strategyused by the firefighters is the basis forthe Recognition-Primed Decision (RPD)Model (see Figure 1). The first level

Positive Features Contrasts

• Studies people with expertise • Studies novices

• Tries to describe • Tries to evaluate

• Takes a broad focus • Takes a narrow focus

• Task context: field settings • Task context: laboratory settingsTime pressure Ample timeShifting conditions Stable conditionsUnclear goals Stated goalsDegraded information Precise informationSubtle cues and patterns Clear inputsTeam interactions Individual tasksOrganizational constraints Individual tasksHigh stakes Low stakes

• Focus on cognitive processes • Focus on analytical strategies

• Relies on Cognitive Task Analysis • Relies on performance measures

Table 1—Features of Naturalistic Decision Making research.

34

Part 4 of 4

consists of a simple match, wheredecision makers experience a situationand match it to a typical situation withwhich they already have experience.Because of this, they know what toexpect. They know what’s going tohappen, they know what the relevantcues are, what the plausible goals are,and a typical action. They are able to doall of this because of their experiencebase. Experience buys them the abilityto size up a situation and know what isgoing on and how to react. That’s whatdecision researchers weren’t learningwhen they studied college sophomoreswho didn’t have an experience base.

An example of the first level of the RPDmodel is a firefighter I interviewed earlyin the process. He explained to me thathe never made decisions. After tryingto press him on the issue, I asked himto describe the last fire he was in. He

told a story of a fairly conventional fire.He described parking the truck, gettingout his hoses, and going into the house.I asked him why he went into the houseinstead of simply working from outside,as I would have been tempted to do.He explained that he obviously had togo in because if he attached it from theoutside, he would just spread it deeperinside the house. He took into accountthe nature of the fire, the distance ofthe house from other buildings, and thestructure of the house. But, even whilehe was attending to these conditions, henever saw himself as making a decision.He never experienced that there wasanother option. He immediately sawwhat needed to be done and did it.

The second level of the model includesdiagnosing the situation. On this level,expectancies are violated. Thefirefighter is trying to build a story to

diagnose the event, and when evidencedoesn’t fit the story, the firefighter hasto come up with a new scenario whichfits the new evidence. There is still nocomparing of options.

On the third level, decision makersevaluate the course of action they havechosen. Originally, we weren’t sure howpeople could evaluate single options ifthey had no other options to compare itto. As we looked through the materialswe were getting, we found that adecision maker would evaluate an optionby playing it out in his/her head. If itworked, they would do it, if it didn’t, theywould modify it, and if modificationsfailed, they would throw it out. In theincidents we studied, commanderssimply generated each option and thenevaluated it for viability. Usually thefirst option an experienced firefightergenerated was a viable option, but they

Figure 1—Recognition-Primed Decision model.

➛➛

➛ ➛

➛

Anomaly

➛

Clarify

➛inference

➛

➛

➛ ➛ ➛no

➛

yes

➛➛

➛

➛

➛

➛

➛

➛

➛ yes

yes,but

Level 1 Level 2 Level 3SIMPLE MATCH DIAGNOSE THE SITUATION EVALUATE COURSE OF ACTION

Experience the Situation Experience the Situation Experience the Situationin a Changing Context in a Changing Context in a Changing Context

DiagnosePerceived as typical [Feature Matching] Is Situation Typical? Perceived as typical[Prototype or Analog] [Story Building] [Prototype or Analog] [Prototype or Analog]

Recognition has four byproducts Recognition has four byproducts Recognition has four byproducts

Expectancies Relevant Cues Expectancies Relevant Cues Expectancies Relevant Cues

Plausible Goals Typical Action Plausible Goals Typical Action Plausible Goals Action 1...n

Evaluate Action (n)[Mental Simulation]

Modify Will it work?

Implement Course of Action Implement Course of Action Implement Course of Action

more data

no

35

Part 4 of 4also understand that they should simplybe satisfying, not optimizing. They willnot necessarily pick the best option.They will pick the first one which ispossible and involves minimal risk.The first viable option is chosen andimproved upon, if necessary. It is notcompared with all other options to seewhich one will be best. As soon as it isdeemed viable, it is chosen and applied.

Naturalistic Decision Making hasimplications for training. Decisiontraining needs to teach people to dealwith ambiguous, confusing situations,with time stress and conflictinginformation. Situation awareness,pattern matching, cue learning, andtypical cases and anomalies can betaught by giving people a biggerexperience base. Training could teachdecision makers how to constructeffective mental models and timehorizons and how to manage underconditions of uncertainty and timepressure.

Methods for providing better traininginclude changes in such things as waysof designing training scenarios. Anotherstrategy is to provide cognitive feedbackwithin After-Action Reviews. This would

do more than point out the mistakeswhich were made in an exercise. Itwould be an attempt to show decisionmakers what went wrong with their sizeup, and why. Another method wouldinclude cognitive modeling and showingexpert/novice contrasts. This would bedone by allowing novice decision makersto watch experts. Novice decisionmakers would also benefit by learningabout common decision failures. On-the-Job Training should be emphasizedrather than simply assuming that oncethe traditional training is finished,decision makers are ready to begin tofunction proficiently. Test andevaluation techniques and trainingdevice specification could also beimproved. All of these might have aneffect on the ability of firefighters todeal with stressful situations.

Why is it that people do make baddecisions? I looked through a databaseof decisions to identify reasons behindbad decisions. We came up with 25decisions which were labeled as poor.Of those, three main reasons for baddecisions emerged. By far, the mostprominent reason was lack ofexperience. A smaller number of poordecisions were due to a lack of timely

information. The third factor was a deminimus explanation. In this situation,the decision maker misinterprets thesituation, all the information is available,but the decision maker finds ways toexplain each clue away, and persists inthe mistaken belief.

The problem of lack of experience hasmany effects (see Figure 2).Inexperienced decision makers lack theunderstanding of situations to be ableto see problems and judge the urgencyof a situation, and properly judge thefeasibility of a course of action. Theseare skills which could be developed toimprove decision making.

The field of Naturalistic Decision Makingresearch is more appropriate thantraditional decisionmaking models forunderstanding how crisis managers,such as firefighters, handle difficultconditions such as time pressure anduncertainty. We have broadened ourfocus from the moment of choice, totake into account situation awareness,planning, and problem solving. By sodoing, we have gained a strongervantage point for understanding errorsand for designing training interventions.

Misrepresent the situation

• De minimus explanations• Alternate SAs are not considered• Difficulty in handling complex:

• Multiple interactive cues• Ill-structured problems

Insensitive to the early signs ofa problem

• Anomalies are not recognized• Urgency is misjudged

Unaware of weaknesses in acourse of action

Figure 2—NDM factors that———— poor decision outcomes.

➛

➛

Lack of experience

• Poor sense of typicality• Poor ability to see patterns

and make discriminations• No basis for expectancies• Missing causal knowledge

➛ ➛

36

Part 4 of 4

Cultural Attitudes andChange in High-Stress,High-Speed TeamsDavid O. Hart, TID, Inc.

What is Decision Making?

As we saw in the other presentations,there are a variety of ways to modeldecision making. The importance hereis that it can be modeled, described, andexamined. By examining decisionmaking as a system, we can learn howattitudes, individual and cultural, affectthe quality of our decisions.

There are as many decision makingdefinitions as there are models. For thisdiscussion we’ll need to have a commonreference to work from when talkingabout decision making. Also, becausewe are talking about organization andteam decision making, we’ll focus thefollowing definitions in that direction. Adefinition of decision making to keep inmind during this discussion is:

The process “of reaching a decisionundertaken by interdependent individualsto achieve a common goal. Whatdistinguishes team decision making isthe existence of more than one informa-tion source and task perspective thatmust be combined to reach a decision.”

Decision Making Factors

Close examination of this definitionreveals many important aspects of thedecision making process in high-stressenvironments. These include, but arenot limited to:

• Ill-structured problems• Uncertain, dynamic environments• Shifting, ill-defined, or competing goals• Action-feedback loops• Time stress• High stakes• Multiple players• Organizational goals and norms

All these factors affect how well thedecision making machine works. If youthink back, you’ve probably encounteredmost (if not all) of these factors duringfire fighting operations.

DM and Attitudes

In this discussion, the factors we’ll beconcerned with are those that relate toand affect cultural attitudes. In general,attitudes that enhance the DM processare seen as positive, and those that actas barriers to effective DM as negative.Many attitudes have both positive andnegative effects. All this may seemintuitively obvious to even the mostcasual observer, but it is important toestablish a common ground before wedelve too deeply into this subject. Inthe spirit of “crawl, walk, run” we’ll needto first understand how attitudes affectthe individual before we can understandthe impacts of cultural attitudes on anorganization.

Attitudes and the Individual

Before we go too much further, we’llneed another definition. This time we’llbe defining attitudes.

The American Heritage Dictionary ofthe English Language defines attitudesas: “a state of mind or a feeling;disposition.” A longer definition is: “Anenduring organizational, motivational,emotional cognitive process with respectto some aspect of the individual’s world.Attitudes and beliefs imbued withemotional and motivational properties.”Another shorter definition, is: “Affect foror against a psychological object.”

They all say the same thing—an attitudeis how you feel about something . Nowthat we know what attitudes are, let’ssee where they come from.

Generally, your experience forms, hasan effect on, or shapes your attitudes.Some attitudes may last only minutes,

others a lifetime. Another way of lookingat it is to say that your attitudes comefrom your values and goals (rememberthose DM factors). So the attitudes youuse as firefighters come from yourtraining and experience as firefighters.

What Do We Do withAttitudes

Attitudes help us make sense out of oursurroundings and allow us to build andmaintain our Situation Awareness (SA).How? By providing each of us a set ofrules and guidelines we use to gatherand process information. Therefore,attitudes aid in our decision making byframing and shaping the information weuse to make our decisions. You couldalmost say that attitudes are imbeddedin every aspect of decision making.Good, bad, or indifferent, attitudesaffect the quality of our decisions.

On a team, the synergy that developscan compensate for attitudinal failuresor barriers in one of its members.Effective teams recognize attitudeproblems and find ways to work aroundthe “attitudinal outages”. A goodexample of this is the issue of womenas crewmembers in combat aircraft.Many male aircrew have a real“attitude” about women in the cockpit.Probable fallout from this barrier isreduced communication, increasedstress, conflict, with a resulting loss ofefficiency and effectiveness. A goodteam will recognize the barrier andreact by:

• Increasing communication to andaround the affected people,

• Closely evaluating the informationsent by the affected parties to weedout any attitude biases,

• Finding ways to reduce stress(knowing military crewmembers,humor would be a likely choice),

• Defusing any conflict before itengages the entire crew.

37

Part 4 of 4We’ve looked at the what, how, andwhy questions regarding attitudes andthe individual, and even looked brieflyat a possible individual attitude outagescenario and the team’s possibleresponse. Now let’s turn our focus toteams.

Attitudes and the Team

Cultural attitudes—what are they, andwhy are they different? As to what theyare, our definition is still valid, but withthis added: the attitude is shared byevery member of the organization.Organizations and teams use attitudesfor the same purpose as individuals, tobuild and maintain their knowledge ofthe environment. The big difference isthat the synergistic effect of teamsmagnifies and multiplies the effect ofattitudes.

The multiplication and magnification cutsboth ways. Positive attitudes provide auniform strength and negative attitudes,uniform weaknesses. An example of apositive effect is providing baselinegoals, values, and priorities (onceagain, remember the DM factors), toestablish a cohesive team more easilyand quickly. Failures are much moreinsidious.

When an attitude fails (e.g., is no longervalid) or is working against a team, itbecomes an attitudinal “blind spot.”Because everyone in the team and/ororganization possesses the attitude, noone can perceive that there is aproblem—there is nothing to compare itagainst. For example, the team has anattitude barrier that inhibitscommunication. By reducing the amountof information flow, and possibly,information quality, there can be asubstantial loss of synergy,cohesiveness, leadership, recognition,awareness, and communication. Allthese elements, working at full capacity,are crucial to effective decision making.

It is important to note that despite theseundesirable results, critiquing andcorrecting the failure is difficult becauseyou can’t “see” the cause.

Where Attitudes Come From

We’ve already determined that anindividual’s attitudes come from his orher values and goals. The same holdstrue for any organization. The culturalattitudes grow out of the organization’svalues and goals. The source for theseattitudes can be either internal orexternal to the organization.

Internal sources are the easiest toidentify. Policy statements, directives,and even official memos are examplesof how organizational goals and valuesmanifest themselves.

Looking to the South Canyon Fire (SCF)incident, the Grand Junction DistrictManagement Team directive that all firesbe “initial attacked and suppressed assoon as possible” is an example of policyworking as a cultural attitude. What yougain from this attitude is a concretedirection for the firefighting teams. Thegoals of their decisions areunambiguous. On the flip-side, thisattitude can become a decisional one-way road. It doesn’t provide a way outof a fire that cannot be suppressed.Also the added emphasis on missionaccomplishment can come into directconflict with existing safety attitudes.

The “can do” attitude identified in theSCF investigation report is common tomany high stress, high speed teams. Ithelps build team cohesion, which isimportant to the team for synthesizinginformation and integrating theindividual perceptions of the situationinto a common perception. But takentoo far, this attitude can have lethalconsequences. By going above andbeyond to complete the job, missionsuccess is prioritized ahead of safety.

We see this in the report where the“can-do” attitude is attributed with thecompromise of the 10 StandardFirefighting Orders (SFOs) and 18Watch Out Situations (WOSs).

When there is a disconnect betweentraining and experience, a barrier toeffective decision making exists. Thisdisconnect causes a gap between theindividual and resulting team perceptionof reality and actual reality. Thisexample is more ambiguous than theprevious two, but when seen in an actualexample, it leaps right out at you. TheSCF report found that “some firefightersfailed to recognize the capability andlimitations of the fire shelters anddeployment sites.” And “somequestioned the value of the fire sheltersunder any conditions and may not havebeen carrying shelters.” It is apparentthat the training received was notsupported or validated by theexperience of the cited firefighters.This kind of gap between perceptionand reality can, and has produced,deadly results.

The final internal example is the attitudeor sense of being part of a larger“family.” This is most often seen as anelitist attitude. In this case we use elitistto mean special, different, or set apart.It is often expressed with the statements“we watch out for our own,” or “we takecare of our own.” This increasedawareness of your teammemberstranslates into an increased safetyawareness. Carried to an extreme, itcan result in a lack of leadership. TheB-52 bomber crash at Fairchild AFB inSpokane was allowed to happenbecause the commanders at the basefailed to ground the pilot for flying theaircraft outside its operational limitsbecause, he was “one of our own,” andfor fear of “ruining his career.”

For external sources of organizationalattitudes, we’ll look at two particular tofirefighting, and one common to theentire federal government.

38

Part 4 of 4Pressure from the public and mediagenerates the attitude that fires with themost public attention should be attackedfirst. Normally, being responsive to theneeds of your customer is seen as apositive goal and attitude. But byallowing people outside theorganization to control priorities, youend up with shifting, ill-defined, orcompeting goals (sound familiar?).

The harsh spotlight of the news mediacan have a similar effect. Anorganization is usually highlightedbecause of some failure or near-failure.The organization usually responds byreacting with abrupt changes in goalsand values, then attitudes, thendecisions. In the case of the SCF, thereaction was increased emphasis onsafety, but unless the spotlight is onsomething that needs to be changed,the resulting changes may not be forthe good of the organization.

The last external example is one thateveryone connected with the federalgovernment, most state governments,and some corporations have felt: “domore with less”. In a perfect world thiswould allow organizations to get themost from their resources.Unfortunately, we don’t live in a perfectworld. In reality, this attitude is a timebomb just waiting to go off.

“Do more with less” pushes people andequipment to perform beyond theircapabilities, usually by sacrificing thenormally accepted margins of safety. Itusually takes a catastrophe many timesworse than the SCF for the federalleadership, from Congress on downthrough each agency involved in theconcerned operation, (in this casewildfire fighting) to realize that you doless with less. Adopting a “do less withless” attitude would mean letting somefires burn themselves out when theydon’t directly threaten the local populace.Unfortunately decisions like theseusually come at an immeasurable cost.

Attitudes, Training, andExperience

Attitudes, training, and experience havea deeply interrelated relationship.Cultural attitudes affect the emphasis oftraining, and experience shapes andmodifies our attitudes. When experienceand training validate each other, thereis usually a positive attitude effect.When they don’t support each other,there’s usually a negative attitude effect.

Start with the training attitude that byemphasizing fire behavior, fuels,weather, and tactics, entrapments will beavoided. Add to that the historically lowfrequency of losses, an experiencebased invulnerability attitude (i.e. “itwon’t happen to me”) can develop. Theoverall experience, expertise, andsuccess of firefighters fosters the attitudethat they can handle any fire (i.e. elitist,can do, or 10:00 fire), which in turn feedsthe training and experience attitude “whyshould we over-learn emergencyprocedures (fire shelter use and bailingout of a situation). From this vantagepoint, it would appear that these attitudesare leading firefighters to lean on luckand circumstance to keep them safe.

The combination of low frequency oflosses (experience), and highlyexperienced teams (experience)conspire to subvert important safetyprocedures and attitudes (training).

Attitude Impacts on SCF

Cultural attitudes played a significantroll at South Canyon. Some of thecultural attitudes that were carried intothe fire were:

• “All fires will be initial attacked andsuppressed as soon as possible.”

• “Highest priority fires are ones thatthreaten life, residences, structures,and utilities.”

• “We can handle the fire.”• “Can do”• “It won’t happen to me.”

This last attitude is a training/experiencetrap stemming from the fire trainingattitude and the fire shelter attitude.

What impact did these attitudes haveon the incident? First, we need torecognize that safety and operationaleffectiveness are opposite sides of thesame coin. The first Standard FireOrder supports this. At South Canyon,the additional emphasis suppressionreceived was both caused by andresulted in the erosion of safety margins.Each time the firefighters “got away with”pushing into their safety margins tosuppress a fire, it reinforced the attitudethat they could do the job with a smallermargin for error. The fact that some ofthe firefighters were uncomfortable withthe situation at South Canyondemonstrates that Grand Junction’ssuppression directive was causing someshifting and competing goals. Thiserosion of the safety attitude coupledwith SA and communication breakdownscritically compromised the team andindividual decision making ability.Among the elements that led to thisbreakdown are physical and mentalfatigue, recognition gaps, weatherinformation not communicated or used,safety concerns not communicated,concerns about who was in charge(leadership) and the numerouscompromises of the SFOs and WOSs.When the blow-up occurred, thesecame together with deadly results. Theattitudes also blocked the last escapepath—dropping tools and packs, buggingout, and using shelters.

After situations like these manyquestions are raised. Some that needto be answered in order to affect anykind of change are:

• Do tactical teams know to increasemeaningful communication during acrisis? Also, do they know how tocommunicate effectively?

• What about pre-planning for crisissituations?

• Do tactical teams get the best informa-tion before and during a fire crisis?

39

Part 4 of 4

Changing Cultural Attitudes

Before we look at examples of how thesechanges are affected, let’s look at whythat change is made.

Why do cultural attitudes change?Because it is recognized that the longterm goals of the organization are notbeing met.

How do you recognize that an attitudeis no longer valid? Since you have no“attitude out” light, you usually know byunwanted results produced by practicingthe behavior associated with the attitude.The feedback from the environment maybe obvious or subtle. Because of theblind spot effect talked about earlier, itis harder for teams to find the offendingattitude than for individuals.Organizations, being larger and morecomplex than their component teams,find it more difficult digging out aninvalid attitude.

Why is it harder for an organization tochange an attitude? You have manymore people needing to change andchange is naturally difficult for people.Because they’re doing something newand different, it take times and effort tomake it stick. Let’s look at a typicalprocess by which organizations canchange attitudes. Then we’ll look tocommercial and military aviation asexamples of organizations that haveundertaken this kind of change.

Preliminary Requirements. Beforethe change process can be started, theorganization, in particular the seniorleadership and managers, needs torecognize that their greatest contributionto this sort of change is providing asupportive environment that will fosterthe growth of the change effort.

Patience, perseverance, andcommitment from the leadership andmanagers is absolutely necessary.Recognizing that this sort of changehappens one person at a time and thatit will be slow and sometimes difficult,they will be supporting the change andtheir own role in the effort.

For the individual, making the changecan be as simple as changing thebehavior associated with the attitude.This can happen very quickly, but maynot have a lasting effect. As soon asthe need for the change has passed,the individual is likely to revert to oldbehavior patterns and start the cycle allover again. Actually changing theattitude is more difficult than changingthe behavior. It takes more time, buthas a more permanent effect. For anorganization, the time and effort isgreatly magnified.

Commitment, or lack thereof, will eithermake or break this type of program.

What needs to be changed? Initiallya survey of the organization should beconducted to determine the attitudes andvalues regarding team effectiveness.Areas that are typically covered in thistype of survey are leadership,communication, recognition andmanagement of stress, needs forachievement, and job satisfaction. Foraccurate data to be gathered the needfor anonymity is essential. In addition across-section of the entire organization,top to bottom, left to right, needs to besampled to prevent inaccurate,misleading, and skewed data. Thisinformation is then used as a benchmarkto measure the change against, and tohelp determine the types of tools tonecessary to make the change.

How does it happen? Using the datafrom the survey, a program of change isdeveloped. Usually this takes the formof training or organizational interventions.The program is usually developed by orin conjunction with professionalsinvolved in this arena. Credibility of thedevelopers, program, and deliverypersonnel is critical to the program’ssuccess. This is the first step in assuringthe buy-in of the front-line teams.

Finally, programs should be designedto fit seamlessly into the culture. Itcan’t be seen as one time fix or justanother training requirement. Tochange the culture, it must be part ofthe culture.

Where does it start? Programs whichwork to improve team attitudes andeffectiveness usually consist of anumber of inter-connected trainingmodules.

Initial “awareness” training is designedto introduce the program and set thestage for the training to follow. It isusually directed at all organizationalmembers who are targeted for change.

A leadership/management “staff” coursefor the senior management is alsoconducted in the initial phases. Theseprograms provide managementpersonnel the essentials to fulfill theirrole in the change process. They needto “walk the talk” if they expect the restof the organization to do the same.

Baseline training is the longest andmost in-depth phase. It provides thebackground, vocabulary, skills, andfeedback the teams need to affect thischange.

Instructors and Evaluators play a specialrole and therefore need special training.This type of training is focused onobserving, instructing, and evaluatingthe new attitude.

Finally, continuation training providesongoing reinforcement of the conceptslearned in the baseline training. For thebest results, it should be practiced in anenvironment as close to actual aspossible.

As with support, training must also runfrom the top down. No one is exemptfrom training, no matter what theirstanding in the organization. Eachphase builds on the previous. Thiscontinuity is necessary so that previoustraining isn’t invalidated by the nextphase. The training that is the mostimportant is usually the most neglected.

Instructor/evaluator and continuationtraining are probably the two most criticalmodules for assuring long-term success.The instructors and evaluators mustembrace the change and its conceptsand procedures, or the training will be

40

Part 4 of 4useless. Lack of buy-in from theinstructors and evaluators can result intraining invalidating training, and evalu-ation invalidating or ignoring training.

Continuation training, on the other hand,keeps the ball rolling. Remember thisis a long term program, not a quick fixBand-Aid. These concepts and skillsneed to be revisited not just annually,but at every training opportunity if it isgoing to be a permanent part of theculture. As with anything new, practice,practice, practice makes perfect. Onefinal, important point regardingcontinuation training—Keep it Fresh!!Nothing will kill a program faster thantired, overused training material. Asnew information becomes available itshould be integrated into the program.

Looking back... we see that this is justa sample of what a program for culturalchange could look like. A real programis much more complex, but then again,real change is much more challenging.

Other organizations have undertaken tochange attitudes within their culture.Most notable is the aviation community.

We’ll look now at commercial and mili-tary aviation to see what brought themthere and what they’ve done and gained.

In the Beginning...

The 1970s saw a number of air carriercrashes. The fact that aircraft crashwasn’t new, but the reasons forcrashing were. More and moreaccidents were being attributed to“human” or “pilot error.” Highlyexperienced, trained, and motivated(sound familiar?) crews were allowingaircraft to crash. Most notable is thePortland DC-8 crash where the crewflew the aircraft out of gas whiletroubleshooting a gear problem on aclear night within sight of theirdestination. Another is the L-1011 thatslowly descended into Floridaeverglades as the crew tried to decidewhat was wrong with a 68¢ lightbulb.The crew was focused on the lightbulband no one was minding the store: why?

The “why?” questions were asked bythe airlines also. Human error was the

answer—but how do you keep it fromhappening? This answer took the form ofCockpit Resource Management (CRM).

A program for change was initiated at anumber of airlines. It probably lookedlike the program we just outlined. Whatthey found was that certain elements inthe human equation needed change.They were, and are, communication,stress management, leadership,decision making, and attitudes.These programs are designed to makethe pilots and flight engineers moreeffective and efficient flight crews.

As the programs became more andmore a part of the airline culture, thebenefits of this type of training wasseen in other areas within thecommunity. They also started seeingsome return on their investment.

A notable (but not isolated) case is theSioux City DC-10 crash. Enroute to theirdestination, the #2 engine, the one in thevertical stabilizer, disintegrated. Piecesof the engine cut through the hydrauliclines for the primary, secondary, andstandby systems. Without hydraulic

41

Part 4 of 4power, the pilots were unable to controlany of the flight control surfaces. By allrights, the aircraft should have crashed,killing everyone aboard. That’s whatthe engineers at the airline and aircraftmanufacturer said. But Capt. Al Haynesattributes his and the passengers’survival directly to CRM. The open,continuous communication, creativesynergy, and their recognizing and usingall available resources are principles atthe heart of CRM, and were the onesused successfully by the flight crew.

Increased focus on and awareness ofeffective and efficient flight operationshelped to broaden the scope of theprogram. The first to be brought intothe fold were the cabin crew, hence thename change to Crew ResourceManagement. Then it spread to themaintenance organizations.

In the early to mid-1980’s militaryaviation became aware of the benefitsof CRM. The USAF Military AirliftCommand (MAC) was the first to comeon board. Their operations were theclosest to the airlines, so it was natural

for them to see the benefits first. MAC“spun” the airline programs to better fittheir environment. The military wasinterested in the effectiveness andefficiency issues, but were moreinterested in CRM’s major by-product:SAFETY. In an environment whereyour enemy is actively trying to reduceyou to an aluminum rain shower, aprogram that keeps you from doingyour enemies’ job is always attractive!

Today, CRM is an inseparable part ofthe airline culture. Human factorsrelated accident rates are down,incidents are down, safety is up and sois efficiency. The program is working.

As for the military, the change is stilltaking root. Military CRM hasn’t reachedthe stage the airlines have, but then aswe have said, these things take time. Ithas also moved out of the aircraft arena.Other military units are seeing thebenefits of CRM. Maintenance, testengineers and pilots, and special forcesunits are just a few that have embracedthe concepts of CRM.

Last Words

Changing a cultural attitude can be adaunting process. But in thisenvironment, as in some of the otherswe’ve talked about, ignoring an attitudethat is in conflict with the organization’sgoals and values is not just inconvenient,it’s downright lethal.

By believing that what you’re doing isimportant, you will be able to make thechanges in your culture. These changeswill have far reaching benefits for theindividual and the organization in safety,decision making, and operationaleffectiveness.

—TIG, Inc. is a consulting company in Aurora,Colorado. We specialize in the delivery anddevelopment of Crew Resource Management(CRM) and human factors training. TIG, Inc. iscurrently providing services to the Army GuardSpecial Forces and The USAF Reserves flyingand maintenance organizations.

42

Part 4 of 4

1Paper presented at Decision Workshop on Improving Wildland Firefighter Performance Under Stressful, Risky Conditions: Toward Better Decisions on the Fireline and More ResilientOrganizations, Missoula, Montana, June 12-16, 1995. I acknowledge with appreciation the generous assistance of Ted Putnam in the development of these arguments, as well as the inputsfrom other participants in the decision workshop. I also thank David Thomas and Paul Gleason for their continuing encouragement. Please direct any correspondence concerning thismanuscript to Dr. Karl E. Weick, School of Business Administration, University of Michigan, Ann Arbor, MI 48109-1234. Telephone: (313) 763-1339.

not, the departing Incident Commandersays in his statement, “I knew (ia.) thatMackey would look (sic) at fire from theair before they jumped and that hewould make a decision on what to dowith it after we left. I did not feel thatsmokejumpers needed additionalguidance” (Report, 1994, p. A 5-9).Mackey got off to a bad start, and thequality of the briefings didn’t improvemuch from then on. For example, thePrineville Hot Shots were not told howGambel Oak burns when it is dry, norwere they told that in previous days,fires had made spectacular runsthrough this material in Colorado.

Why so much casualness? Onepossibility is that everyone seriouslyunderestimated how much continuingeffort and shared information it takes tobuild coordination and hold it together,especially during transitions from aninitial attack to an extended initial attack,from one level of complexity to anotherlevel, and from one organization toanother. The investigation team, on p. 6,states the following: “as is typical inextended attack situations, firefightinggroups arrived on the fire at intervalsfrom dispersed locations and blendedinto the existing organization.” The keyword there is “blended.” Blendingsounds like something that occursautomatically not something that peoplework at. Many would say it’s especiallyhard to blend into an “existingorganization” if that organization itself isinvisible, as was the case for somepeople at South Canyon. Some peopletrying to blend did not know who theIncident Commander was, or whichradio traffic had the force of authority,or what the suppression strategy wassince it seemed counter-intuitive.

The questions that need to be pursuedare, why does briefing continue to betreated casually and what does betterbriefing sound like? Back in 1949,during the investigation of Mann Gulch,Henry Thol’s father understood the

South Canyon Revisited:Lessons from HighReliabilityOrganizations 1

Karl E. Weick, University of Michigan

In this paper I want to explore the ideathat organizing to prevent wildland firedisasters such as the South Canyon Fireon July 6, 1994 in which 14 people losttheir lives, is an ongoing struggle foralertness. My intention is to look moreclosely at that struggle. I want to do 4things. First, I want to discuss 4 piecesof my earlier analysis of the Mann Gulchfire that seem relevant to South Canyon.In particular, I want to discuss briefings,leadership, tools, and wisdom.

Second, I want to discuss organizationalissues at South Canyon that are lessvisible in Mann Gulch. These includediscrepancies, levels of experience, thewill to communicate, and Watch Outsinvolving management. Third, I want totouch on solutions. And I want toconclude by discussing some questionsabout South Canyon that continue tohaunt me.

Similarities Between MannGulch and South Canyon

Briefings . The struggle for alertness atMann Gulch was undermined by manyof the same things that undermined it atSouth Canyon, one of which is briefings.Briefings are an attempt to give peoplein a crew a common framework inadvance including assumptions aboutwhat they may face, how it will develop,and how the crew will function andupdate its understanding of what isgoing on.

At Mann Gulch, the crew of 14essentially proceeded without much ofa briefing. They basically knew only

that they were jumping on a fire thatwould likely be out by 10:00 the nextmorning. After landing, all some ofthem knew was that Dodge hadscouted the fire on the South slope withHarrison, had used the phrase “deathtrap” to describe what he found, andhad ordered the second-in-commandWilliam Hellman to march the crew downthe North slope toward the MissouriRiver. Dodge didn’t say whether thistactic was to escape the death trap orto position the crew to fight the fire, orsimply to get closer to the river. Whenthe fire spotted to the North side of thegulch, Dodge turned the crew aroundand angled them up toward the ridge,and soon ordered them to drop theirtools, and then to enter an escape fire,all without verbalizing his reasons(Dodge, 1949, p. 121). Since the crewdid not know each other well, sinceDodge knew only 3 of them, sinceseveral were on their first jump, andsince Dodge himself was rusty onleading a crew (Maclean, 1992, p. 41), itwas imperative to build some commonunderstanding and common action intothis assortment of strangers. Thatdidn’t happen.

But neither did it happen 45 years laterat South Canyon. The South Canyonaccident investigation team allocatedalmost a full page (Report of the SouthCanyon Fire Investigation Team, 1994,p. 26: hereafter referred to simply asReport, 1994) of their report to “Safetybriefings” as a “significant contribution”to the 14 deaths. The hand-off of thefire the evening of July 5 from the BLMcrew to the smokejumpers and Jumper-in-charge Mackey is a good example ofhow not to brief people. The hand-off isby radio rather than face to face, is madeafter the BLM crew who know theterrain and foliage has left the scene,and the jumpers inherit a handline whichis partially constructed but already lostby the time they collect their gear andare ready to extend it. Without checkingwhether the assumption is correct or

43

Part 4 of 4trained to wear steel-toed shoes at alltimes and often refuse to take them offwhen they are ordered to abandon asinking ship. Fighter pilots report beingreluctant to eject from the “warm womb”and “cocoon” of oxygen in a cockpit thatis out of control into a far more harshenvironment. It is just as hard to dropshoes or an aircraft as it is to drop apulaski and a pack.

At Mann Gulch, Dodge told his crew to“drop all heavy tools” 200 yards afterthey turned upslope. According toSallee (1949, pp. 75-76) and Rumsey(1949, p. 103) people either threw awayeverything or nothing. Dodge in histestimony said he “didn’t know untillater that they had discarded shovelsand pulaskis” (1949, p. 118). Salleereported that with the fire racing at them,smokechaser Harrison was sittingresting “and he still had his pack on”(Sallee, 1949, p. 88).

This same pattern was repeated atSouth Canyon. Some of the smoke-jumpers who deployed their sheltersabove the lunch spot, did drop theirtools. But in doing so, they were struckby the enormous symbolic significanceof what they were doing. One observedthat putting down a saw was likerunning up a white flag (Rhoadesstatement); another (Petrilli), that the“Pucker factor” went up a notch (Report,1994, p. A5-69).

What about those who didn’t drop theirtools? If dropping your tools signifiesyou’re in deep trouble, keeping themmay help you feel you’re safe. To holdonto your tools is to stay in control, toremain a firefighter rather than a victim,to appear calm. I’m still in it. This is notjust an issue of symbolism since toolsare needed to scrape an area clearbefore deploying a fire shelter. But thereluctance to drop tools may come fromother sources such as economics,habits, avoidance of failure, predictionsof fire behavior, and social dynamics.Equipment is expensive and jumpers, atleast, are told repeatedly and early intheir training to carry out everything that

essentials of a briefing even if much ofhis emotional testimony (“I owe this tomy boy”, p. 201) was tough to follow.“Usually the foreman he always lookedout for all, to take care of anything thathappened. We always looked out forthat before he put the men on the fireline. He had something to fall back on. . . let’s go in there boys, the wind isn’tblowing now. We’ll go in there. Butwatch out, the wind can change anymoment” (Thol, 1949, p. 200). Morerecently, researchers have studiedeffective cockpit crews in aircraft andhave found that better briefing leads tobetter performance. This is relevantbecause in cockpits, as well as on firelines, people often work with strangers.In particular, effective leaders establishand reaffirm norms of conduct forbehavior in the group, and insist thatpeople keep each other informed onwhat they were doing and the reasonsfor their actions and the situational modelthat gave rise to those reasons andactions. Almost no one at Mann Gulchor South Canyon heard someone say,

1) Here’s what I think we face:2) Here’s what I think we should do;3) Here’s why;4) Here’s what we should keep our

eye on;5) Now, talk to me.

Leadership. But Mann Gulch andSouth Canyon are similar not only intheir casual briefings. There wasuncertainty about leadership in bothcases. At Mann Gulch, leadershipmoved uneasily among Navon,Hellman & Dodge. At South Canyon, itmoved uneasily among Blanco,Mackey, Longanecker, Shephard,among others. At Mann Gulch, as atSouth Canyon, crew members were notclosely acquainted with their foremendue to continual rotation of peopleamong crews and assignments. (Fite,1949, p. 28). Dodge knew only 3people in his crew, Hellman, McVey,and Thol (Dodge, 1949, p. 125).Hellman, who was better acquaintedwith the men (Dodge, 1949, p. 125)was near the front of the line as they

raced uphill (Sallee, 1949, p. 76) andreportedly said “to hell with that, I’mgetting out of here,” when Dodge orderedpeople to jump into his escape fire.

At Mann Gulch people were tornbetween 2 conflicting influences. But,the same thing happened at SouthCanyon. Haugh and Erickson bothyelled at the retreating Hotshots to droptheir tools (Report, 1994, p. 16) and runfor the ridge while Thrash, who was atthe head of the line of jumpers andhotshots stopped and began to deployhis fire shelter as did smokejumperRoth. Hipke and Blecha said inessence, to hell with that, I’m gettingout of here and continued to run.

This similarity may be merely a coinci-dence. It may be more significant. Itseems worth exploring, however,because it adds uncertainty to a situationthat already has lots of puzzles. Uncer-tainty about leaders puts increaseddemands on crews, dispatchers, andpilots at a time when they are close tooverload. Uncertainty pulls groups apartwhich, makes them more susceptible topanic (Weick, 1993, pp. 637-638). Anduncertainty in the face of unclear leader-ship often cuts off the flow of informationbecause people don’t know who to sendit to and responsibility keeps shifting atwill. As we will see later, uncertaintiesabout leadership were not confined toSouth Canyon. They extended upthrough the organization and this setsthe tone for actions reflected throughoutthe organization.

Tools. A small, but powerful similaritybetween Mann Gulch and SouthCanyon is that, in both cases, whenpeople were fleeing the blowup andwere told to drop their tools so theycould move faster, some resisted.Several calculations suggest that thisresistance may have cost them theirlives (Report, 1994, p. A3-5). Theywould have been able to move 15-20%faster (Putnam, 1994) without theirpacks and tools. Firefighters are not theonly people who are reluctant to droptheir tools. Naval seamen on ships are

44

Part 4 of 4know is only part of what could beknown, and therefore, you need to stayalert. You need to avoid excessconfidence that you know everything andexcess caution that you know nothing, ifyou want to stay flexible.

Wise organizations know what they don’tknow. They know two things: first, theyknow that they have not experienced allpossible failure modes and second, theyknow that their technology is still capableof generating surprises (Schulman,1993). Thus, when they act on the basisof their past experience, wise organiza-tions act as if that experience is bothcredible and limited. They simultane-ously believe and doubt they know whatis up. Consider the case of a near missor a close call. The fascinating thingabout a near miss is that, “Every time apilot avoids a collision, the eventprovides evidence both for the threatand for its irrelevance. It is not clearwhether the learning should emphasizehow close the organization came todisaster, thus the reality of danger in theguise of safety, or the fact that disasterwas avoided, thus the reality of safety inthe guise of danger” (March, Sproull, andTamuz, 1991, p. 10). If the moment isinterpreted as safety in the guise ofdanger, then learning should bediminished because “more thoroughinvestigations, more accurate reporting,deeper imagination, and greater sharingof information” are all discouraged(Sagan, 1993, p. 247). The attitude ofwisdom sees a near miss as evidencethat the system is both safe and vulner-able, that people must remain alert, andthat a safe environment is not measuredby an absence of accidents (that out-come is largely dependent on luck), butis the result of active identification ofhazards and their elimination (Allinson,1993, p. 186).

At Mann Gulch, people believed theywere fighting a fire that would be out by10:00 the next morning and failed toraise questions about whether thisexpectation remained accurate. AtSouth Canyon people believed theycould “hook” the fire before the winds

is dropped to them. Habits built upduring training are much more likely toinvolve moving with tools in hand,rather than moving and discarding tools.People have no idea what it feels like torun and discard tools or even how to doit. Rhoades in his statement mentionsthat as he was running to escape theSouth Canyon fire he kept looking for aplace to put the saw down so it wouldn’tget burned, a search which undoubtedlyslowed his progress. In his words, “atsome point, about 300 yds. up thehill....I then realized I still had my sawover my shoulder! I irrationally startedlooking for a place to put it down whereit wouldn’t get burned. I found a place Iit (sic) didn’t, though the others’ sawsdid. I remember thinking I can’t believeI’m putting down my saw.” These wordshave even more impact when it isrecalled that, among the fatalities,firefighter #10 (Putnam, 1994) wasfound with a saw handle still in his hand.To discard one’s tools may signify morethan giving up control, it may also be anadmission of failure which, in a “can do”culture, is a devastating thing to admit.

There is a further complication with theseemingly simple act of dropping one’stools. If people drop their tools, theystill face a tough choice, namely, do Inow run faster or do I stop and deploymy shelter? It is tough to do bothalthough some people at South Canyontried. Running faster and stopping todeploy are incompatible and uncertaintyabout which one to do may compelpeople simply to keep doing more ofwhat they are already doing, namely,running with tools. To keep running isto postpone having to make a tougherchoice, especially if the person feelsboth exhausted and uncertain how safethe shelter really is. People may alsohold onto tools because their predictionsof fire behavior suggest that the firewon’t reach them. This is a clearpossibility at South Canyon. As the firemoved toward the hotshots and jumpersmoving North along the fireline, itrepeatedly was channeled toward theridgeline along draws that ran at rightangles to their movement. This fire

behavior could have created theimpression that the crew was at the flankrather than the head of the fire whichmeant there was no need to drop tools.

Finally, people may hold onto their toolsas a simple result of social dynamicswhen people are lined up. If the firstperson in a line of people moving up anescape route keeps his or her tools, thenthe second person in line who sees thismay conclude that the first person is notscared. Having concluded that there isno cause for worry or that I’m not goingto be the only one who goes backwithout tools, the second person alsoretains his or her tools and is observedto do so by the third person in line whosimilarly infers less danger than mayexist. Each person individually may befearful, but mistakenly concludes thateveryone else is calm. Thus, thesituation appears to be safe except thatno one actually believes that it is. Theactions of the last person in line, theone whose back feels most intenselythe heat of the blowup are observed byno one, which means it is tough toconvey the gravity of the situation backup to the front of the line.

What hasn’t changed in 45 years is thepower of symbols. Packs and saws maybe heavy and slow one’s pace. But thatmay be one of their less importantqualities. More significant may be theirability to reduce one’s sense of danger.If throwing tools is a sign of surrender,keeping them may be a sign of astandoff or victory. It may be importantfor trainers to emphasize, “Look people,you’re going to want to hang onto thisstuff. Don’t! It could cost you your life.

Wisdom. The fourth aspect of myMann Gulch analysis that fits SouthCanyon centers on the idea of wisdom.To understand why the idea of wisdomfits here, you need to understand firstthat wisdom is a mixture of knowledgeand ignorance. When one of themgrows, so does the other. To knowsomething better is also to discover thatnew questions about it are raised.Wisdom is an attitude that what you

45

Part 4 of 4troublesome discrepancy. Buildingdirect line downhill is dangerous.Longanecker said, “going downhill directis a bad deal” (Report, 1994, A5-52).Archuleta asks, why are we punching inline? Erickson asks, “Where are thesafe areas?” and hears the answer,“there really aren’t any.” Rhoades,Doehring, and Shelton overhear thisconversation. But the decision is madeto build a direct line anyway, whichleaves everyone tense. They believethat the action is dangerous, yet theyare doing it. What makes this reallytroubling is that the decision is a public,irrevocable, choice. There is goodresearch evidence (e.g., O’Reilly andCaldwell, 1981; Salancik, 1977) thatwhen people make choices of this kind,they are more likely to change theirbeliefs so that they become consistentwith the action they are now committedto. In this case people should begin tobelieve that building direct line downhillis safe after all in order to justify whatthey are actually doing.

And that’s what seemed to happen.Listen to how Quentin Rhoades in hisown words, handled things: “I resolvednot to go down that hill digging line . . .Smokejumpers arrived and starteddigging line. I remember thinking that Imust have missed something. I hadn’tbeen on a fire since August 18, 1992and I felt a little green.” Rhoadesconvinces himself that the main reasonthe situation seems dangerous is thatit’s his fault, he’s rusty, he’s missedsomething, which means the situation isnot as dangerous as it looks. Otherpeople resolve the discrepancy in otherways. They convince themselves thatthe leaders know what they’re doing,that it won’t take long to cut the line, thatthe predicted weather front won’t bethat strong, that they can “hook the firebefore the front passed” (Report, 1994,p. A5-53), that the crews are really ontop of this job, and that more resourcesare coming (Report, 1994, p. A5-47).There is a grain of truth in all of thoseexplanations. But people also have astake in needing them to be true, sincethey reduce the tension associated withdoing something they believe to be

would build and they presumed thatlookouts and a commander had the bigpicture even though the firefighters hadseen no evidence of this.

The attitude of wisdom is one way toremain alert, because it leads people toremain open to what is happening and torely cautiously on their past experience.I’ve always been struck by evidencesuggesting that there are certain periodsduring a person’s career, when they aremost in danger of getting injured orkilled. Police, for example, are in mostdanger of being shot during their 5thyear on the force. Firefighters are inmost danger of fireline accidents eitherin their first 2 years or after 10-15 yearsof experience (Pyne, 1984, p. 391).Young firefighters are vulnerablebecause of their inability to recognizehazardous situations. The moreexperienced firefighters are vulnerablebecause they presume they’ve seen itall, they have less openness to newdata, thus the validity of their modelsdecreases. The unexpected gets them.

Crews and commanders need to keeplearning and updating their models.This won’t happen if they presume thatnothing about fires can surprise them, ifnear misses are treated as testimonialsto safe practices, and if they are certainthat they’ve experienced all possibleways in which a system can fail. Theseattitudes won’t change if they reflectsimilar attitudes in top management.You may recall that Maclean felt “theForest Service wanted to downplay theexplosive nature of the Mann Gulch fireto protect itself against public chargesthat its ignorance of fire behavior wasresponsible for the tragedy” (Maclean,1992, p. 125). The key word there is“ignorance.” The service doesn’t wantto appear ignorant. Nor do it’s crews.The price of creating this impressionmay be a loss in vigilance, learning,and wisdom.

It is tempting in a world of boldness andaggressive attacks, to conclude thatthere is no place for doubt. But asThoele (1994) has suggested the bestfirefighters do not confuse risk with

recklessness, and they are able “to say‘no’ without sustaining dents in theirmachismo” (p. 28). That’s what wisdomis about, and why it’s worth striving for.

Differences Between MannGulch and South Canyon

Discrepancies Between Beliefs andActions. Having suggested at least 4ways in which dynamics of organizingin South Canyon replay themes thatunfolded earlier in Mann Gulch, I nowwant to explore some additional issuesthat were less visible in Mann Gulch butthat stand out in South Canyon.

The first of these is the unusually largenumber of inconsistencies betweenbeliefs and actions at South Canyon.I want to dwell on these because theysuggest one reason why peoplepersisted so long doing things thatviolated fire orders and watch outs.

A recurring belief among people fightingwildland fires is that some of the firesthey fight are on worthless land. Thiswas a prominent issue at Mann Gulch.As Earl Cooley (1984) put it, “One of themain questions was why we risked livesand spent many thousands of dollars tosave scrubby timber and cheatgrass”(p. 91). A basic discrepancy thatfirefighters and overhead face over andover is between their belief that the landis worthless and the reality that they arerisking their lives to defend it. The actionof defending is inconsistent with thebelief that the area is worthless.Contradictions such as this causetension and continue to do so until theperson either changes the belief—theland is more valuable than it looks—orchanges the action, and uses low prioritysuppression tactics. Either changereduces the inconsistency.

Let’s extend this scenario to SouthCanyon and a key decision, the decisionmade at 9:30 the morning of July 6 tocut a direct fireline, downhill (Report,1994, p. A4-6). What is noteworthyabout this decision is that it involves a

46

Part 4 of 4somewhat chilling connotation to thepersonnel category, “Overhead.”Experience is unevenly distributedacross the several activities at SouthCanyon and does not always line upwith authority. There are no clearmechanisms to mobilize and focus andimplement the experience that isscattered around. And finally, everyoneis accessing their experience underincreasing amounts of stress, whichmeans they are likely to fall back onthose habits and understandings theyhave overlearned (Weick, 1990, pp.576-577). Unfortunately, these may bethe very habits and understandings thatare least relevant to the uniqueconditions in South Canyon.

There are at least three reasons weneed to tackle the issue of experienceand how it is mobilized. First, animportant finding from studies of highreliability organizations is that they havemultiple structures. Aircraft carriers, forexample, have a bureaucratichierarchical structure for normalfunctioning during slack times, adifferent structure built around expertisefor “high tempo” periods of extendedflight operations, and a third structureexplicitly designed for emergencies.High tempo structures are especiallyrelevant for wildland firefighting whererank in the formal hierarchy does notalways coincide with technical expertise.LaPorte and Consolini (1991) describea high tempo structure on carriers thisway: “Contingencies may arise thatthreaten potential failures and increasethe risk of harm and loss of operationalcapacity. In the face of such surprises ,there is a need for rapid adjustmentthat can only rarely be directed fromhierarchical levels that are removedfrom the arena of operational problems.As would be expected, superiors havedifficulty in comprehending enoughabout the technical or operational situa-tion to intervene in a timely, confidentway. In such times, organizationalnorms dictate noninterference withoperators, who are expected to useconsiderable discretion.

dangerous. The trouble is, they nowhave a vested interest in not seeingwarning signals. If they do notice thesesignals, then their whole sense of whatis happening collapses. Listen again towhat Rhoades says: “My ditty bagcontained a copy of standard fire ordersand watch situations. I consideredlooking at it, but didn’t. I knew we wereviolating too many to contemplate.”

When people take public, irrevocableactions for which they feel responsibility,their mind set is to justify those actionsand to assemble evidence that showsthe action makes sense (Ross & Staw,1986). They are not indifferent towardevidence that raises doubt about theaction. Instead they avoid, discredit,ignore, or minimize this contraryevidence and keep looking for positivereasons that justify continuing the action.People who justify their actions persist,or in the words of the investigating team,“strategy and tactics were not adjustedto compensate for observed andpotential extreme fire behavior” (p. 35).

I have dwelt on this one decision atSouth Canyon to show how peoplejustify their actions and in doing so,become more committed to continuingthose actions. There are several otherdiscrepancies that could be analyzedthe same way, such as the belief thatthis was a low priority fire yet Type 1crews were put on it; the policy that twoor three trees burning is a standardsmokejumper dispatch (French), yetjumpers were not dispatchedimmediately; the belief that this is apotentially serious fire, yet a crew walksoff it the night of the 5th; the belief thatretardant works only at certain stagesof a fire, yet requests for it at that stageare refused; aerial reconnaissance thatspots fingers of fire in west drainage onJuly 6, yet these are not drawn on themap (Report, 1994, pp. 26, A5-70). Mypoint is not simply that there werediscrepancies at South Canyon. Life isfull of discrepancies and people manageto deal with them by sizing up pro andcon evidence. My point is that, keydiscrepancies at South Canyon seemedto occur in a context where people got

locked into public irrevocable, volitionalactions, and had to justify those actions.These justifications made them morecommitted to those actions, which ledthem to persist longer in executingthose actions despite growing dangers.Notice that the people who would bespared from this process of escalationwould be those who were forced to cutline (there is low choice), people whosaw escape routes, (the action isrevocable) and people who did notexpress their views in public (thedecision is not linked to them asindividuals).

Levels of Experience. Earlier Imentioned that experience has both anupside and a downside. The upside isthat it gives you more patterns that canbe retrieved and matched with currentpuzzles to make sense of them. Thedownside is that more experience cansometimes lead to less openness tonovel inputs and less updating of themodels one uses. Failures to reviseoften produce ugly surprises.

I want to dig deeper into the issue ofexperience levels at South Canyon,partly because the accident investiga-tion team seemed reluctant to do so. Isay this because if you look at the FireEntrapment Investigation and ReviewGuidelines (Report, 1994, pp. A12-3 toA12-11) which they followed religiouslyin structuring their report, the onlycategory out of the 28 that they omittedwas category 23, “V. Involved personnelprofiles - Experience levels” (Report,1994, p. A12-7). This omission may bedue to the fact that, on paper everyoneis qualified. But just because they’requalified on paper, doesn’t mean thattheir experience is deployed well in thisincident or sufficient to handle itschanging character or easily adapted toit. Issues of experience levels at SouthCanyon are complicated, difficult tountangle, and touchy when untangled.But that’s no reason to avoid them.

The overall level of relevant experiencefor leadership appears to be low.Several people appear to be in over theirheads, which gives a whole new and

47

Part 4 of 4He starts with a sloppy hand-off theevening of July 5 and an unfinishedproject which he is unable to continue.He’s dropped on unfamiliar terrain, attwilight, with rolling debris and steepslopes. The crew is unable to get muchsleep. The resources (two Type 1 crews)that Mackey requests the night of the 5tharrive in small numbers at unpredictableintervals the next day (8 jumpers at10:00 a.m., 10 hotshots at 12:30 p.m.,another 10 hotshots at 3:00 p.m.) andMackey is not even sure they’ll come atall since he’s been told his fire is lowpriority. When there is disagreementabout building line direct and downhill,the incident commander does notresolve it and the hotshot superintendentdoes not seem to question the strategywhen he arrives around noon (Report,1994 pp. A4-6, A4-7).

At some level Mackey knows the downhillstrategy is risky because, in response toa flare-up at 10:35 AM, he begins to pullthe crew out (Report, 1994 p. A5-70)only to have that decision questioned byLonganecker who suggests doing bucketdrops. The drops are made and thecrew resumes cutting line. Not longafter this Rhoades observed that “Donlooked terrible.” Still later, when the sawRhoades is using breaks down, Mackeyoffered to sharpen it and help him cutline. This looks like a clear instance ofa person falling back on overlearnedbehavior when that person is underpressure. Mackey discards the lessfamiliar activity of keeping your head upand supervising for the more familiaractivity of keeping your head down andcutting line.

I mention this example to make the pointthat when demands exceed capabilities,which is the basic condition under whichpeople experience stress (McGrath,1976), this is seldom simply the fault ofan individual. The buck doesn’t stop withthat person. Instead, the buck stopseverywhere (Allinson, 1993). The peoplearound Mackey made his assignmentharder and reduced his capabilities tohandle it. The resulting pressure madeit harder for Mackey to gain access tothe experience he already had, which

Authority patterns shift to a basis offunctional skill. Collegial authority (anddecision) patterns overlay bureaucraticones as the tempo of operationsincreases. Formal rank and statusdeclines as a reason for obedience.Hierarchical rank defers to the technicalexpertise often held by those of lowerformal rank. Chiefs (senior noncom-missioned officers) advise commanders,gently direct lieutenants, and cowensigns. Criticality, hazards, and sophis-tication of operations prompt a kind offunctional discipline, a professionaliza-tion of the work teams. Feedback and(sometimes conflictual) negotiationsincrease in importance; feedback about“how goes it” is sought and valued”(p.32).

People in South Canyon did not seem tohave the capability to form a high tempostructure where influence flowed fromexpertise and experience, rather thanfrom the formal chain of command. Inpart, the problem was that it was neverclear where the relevant expertise waslocated so that the structure could formaround it. Furthermore, there was noclear chain of command that could deferto more experienced people nor wasthere a clearly understood set of signalsby which such a shift in structure couldbe conveyed immediately andunequivocally to everyone.

A second reason the issue of experienceis important is because it has thepotential to create a smarter system thatsenses more. A key idea in systemdesign is the notion of requisite variety:it takes a complex system to compre-hend a complex environment (Miller,1993). Analyses of South Canyon thatare consistent with this principle havealready begun to appear. For example,Topic 3.5 in the IMRT review states thatmanagers should “match qualified inci-dent commanders with the complexityof incidents” (Wildfire, Vol. 3, No. 4, Dec.1994, p. 46). That’s requisite variety.Inadequate requisite variety occurs whena less complex incident commander, ora less complex jumper crew, or a lesscomplex dispatcher, cannot adequatelycomprehend a more complex event.

Requisite variety that is more adequatecan be illustrated by a crew of smoke-jumpers who have had prior experienceas hotshots. Such a crew has thecapability to function either in a moreindependent jumper mode or a moredisciplined hotshot mode, which givesthem a larger variety of ways to copewith a larger variety of fire behaviors.

The notion of requisite variety alsoalerts us to a hidden danger insuccessful firefighting. There is growingevidence that success leads to systemsimplification (Miller, 1993), which meanssuccessful systems steadily becomeless sensitive to complex changesaround them. This insensitivityculminates in a sudden string of failuresand the horrifying realization that onehas become obsolete and faces a nasty,prolonged period of playing catch-up.

Again, the lesson from high reliabilityorganizations such as the DiabloCanyon nuclear power plant is the needto cultivate diverse experiences, variety,multiple points of view, and conceptualslack (Schulman, 1993) so that peoplehave a better sense of the complexitythey face. And, there also need to bewell-learned, trusted, procedures tohandle the inevitable conflicts that arisewhen people make different interpreta-tions, such as when a Fire ManagementOfficer and a Hotshot superintendentdiffer on how the fire should be fought.

The third and final nuance of experiencethat I want to raise is the question ofwhat happens when you are at the limitsof your experience where demandsexceed capabilities? And what can bedone about it?

For the sake of illustration, let’s look atjumper Mackey who was jumper-in-charge at South Canyon and who hadjust recently been given a permanentappointment. What’s interesting andtroubling about Mackey’s position is thatthe system makes it hard for him to do agood job on this fire. If we put ourselvesin Mackey’s shoes we discover that heis in a bad spot almost from the start.

48

Part 4 of 4work at a level of stress which is alreadyquite high. It doesn’t take muchadditional stress before the quality oftheir judgment and thinking may beginto suffer. Radio discipline is practicallynon-existent (Report, 1994, p. A5-37).Dispatch keeps reminding people thatSouth Canyon is a low priority fire, thatthere is nothing out of the ordinary(LaDou statement), and they keepsaying “Roger” to all requests forresources without any feedback as tohow and when the request will behandled, if at all. Requests for retardantare denied, weather briefings areunevenly distributed, and no one takesresponsibility for better distribution. TheIC is invisible (Report, 1994, p. A5-67)and there is no guidance for helicopteruse which means that people competecontinuously (Byers statement) for itsservices (Report, 1994, p. A5-22).These poorly integrated managerialdecisions are spread over the periodfrom July 2nd through the 6th and mayreflect even earlier decisions aboutsafety and how people are to be treated.

The questionable decisions continueafter the blowup, suggesting that theincident within the incident ismishandled. Aircraft are kept circlingabove the blowup for 45 minutes in 50knot winds (Ferneau statement). Thegovernor is allowed to tie up a key phoneconnection for 15 minutes which delaysrescue efforts for the people deployedin shelters. Helicopter pilot Good, whoseems to have a stunning amount ofendurance and resilience, is still beingordered around at 9:00 at night, this timeto fly body bags in. He speaks for alarger group when he refuses, saying,“I’ve had enough” (Report, 1994, p. A5-50).

These are all symptoms of problems farremoved from the crew boss on theground, and our job is to diagnosesymptoms of what. Many would saythese are symptoms of problems incommunication. This is what theHotshots said: “The crew wants to knowwhere the communications broke downwith the red flag warning” (Report, 1994,p. A5-81). The answer to their question

increased pressure when his decisionswere questioned, which gave him evenless access to his experience until hewas caught in a vicious circle where hedid what he had always done on fires,namely cut line rather than supervise.The Hotshots had no idea somethinglike this might be developing, and whenthey saw Mackey, he seemed to bemoving around and checking, which iswhat overhead is supposed to do.

The system let Mackey down. It did littleto remove or redistribute pressures, itdid little to simplify his assignment, andit did little to monitor the fact that he andothers had less and less energy to copewith growing complexities. The crewwas losing variety and alertness, and noone spotted this or slowed the loss, oraltered the work so that whateveralertness remained was sufficient.

Communication. In the precedingdiscussion of levels of experience, Isteadily enlarged the size of the relevantorganizational unit from jumpers andhotshots and South Canyon overhead,to the system in general includingdispatcher, interagency coordinators,and top management. I did so in thehope that we would not fall into the trapof glibly saying that South Canyon isanother instance of operator error, butwould instead incorporate a larger,earlier, higher set of design decisionsas significant contributors to theincident.

You may recall that the teaminvestigating South Canyon felt that“Management support and dispatchcoordination” were not “significantcontributors” to the disaster, but merely“influenced” it (Report, 1994, p. 33). Imention this partly because noteveryone agrees with this assessment(e.g., OHSA, IMRT), and partly becausethis is the same kind of questionableassessment that was made by the teaminvestigating the Challenger disaster.In the Challenger report, the MainCause of the disaster was listed as“failure in the joint between the 2 lowersegments of the right solid rocket motor”(Allinson, 1993, p. 111) and the

Contributing Cause was listed as flawsin the decision making process. Theimplication of such an analysis is thatpeople should devote the brunt of theirenergy to correcting the main cause.

Allinson (1993, p. 111) among othershas argued that the Challengerinvestigating team had their prioritiesreversed. The failure was set in motionby actions and choices that said it wassafe to launch and by the decision tolaunch itself. The defect in the O-ringcan’t harm anyone as long as thatdefect stays on the ground. The factthat a defective design even “existed atall was the result of previous decisionsto select this design. That it was allowedto continue to exist was the result ofprevious decisions not to alter it, despiterepeated warnings. That it was allowedto be in use in unsuitable weatherconditions was also the result ofdecisions made to allow it to operatedespite the danger that the weatherconditions represented. [Allinsonconcludes by saying] It seems moreappropriate, then to describe thetechnical defect of the Challenger withthe term “proximate cause” andmanagement’s decision to launch theChallenger without an adequate regardfor safety, the ‘primary cause’ ” (p. 113).

Since the South Canyon report focuseson the crews cutting line, it is difficult tospot earlier administrative decisions thatare potentially significant. But there iscertainly no shortage of possibilities.Crews at South Canyon are told to beaggressive but are given little supportto do so and later are faulted for beingtoo aggressive. Prineville Hotshots arerequested and then treated poorly whenthey arrive at the Glenwood Springsoffice at 8:00 a.m. on the 6th (Report,1994, p. A5-80) where they are forcedto look around to find tools and then goto the 7-11 to get food. Theirunderstandable agitation at beinghandled this way probably does notdisappear the moment they get toSouth Canyon. Instead, much like themarried pilot who takes command of anairplane shortly after an intensedomestic quarrel, the crew starts their

49

Part 4 of 4not desired, or would not be passed on.They may not have asked becausethey thought they had all the answersor wouldn’t get them anyway. And theymay not have passed on informationbecause they assumed it would notreceive a hearing. If any of thesepossibilities are true, and if people alsobelieve that no news is good news,then wildland firefighting is a thousandadministrative accidents waiting tohappen and is even more dangerousthan people realize. Fire is not theproblem. The problems are alertness,trust, trustworthiness, respect, candor,and “the will to communicate” (Allinson,1993, p. 41), a list that fits Mann Gulchas much as it fits South Canyon. Thedifference is that in South Canyon, thelist applies to a more dispersed set ofpeople with a more diverse set ofinterdependent tasks.

Safety attitudes are inherent in goodmanagement practice rather thansomething that are tacked on. Free flowof information is good managementpractice, gets things done, and saveslives. If people fail to pass alonginformation, fail to listen attentively, andfail to elicit information actively, that isbad management and unsafemanagement. I suspect Stephen Pyne(1984, p. 394) has it about right whenhe said that “All too often ‘safety’ is acosmetic, a mandated and barelytolerated veneer of declarations,memorandums, task force reports,safety officers, and exhortations thathas little relevance to the conduct ofpractical affairs. Something is taughtas a ‘safe’ procedure rather than theonly procedure. Safety is somethingadded to a program, not somethingintegral to it . . . Most safety programsfail at the bottom because they are nottruly practiced at the top.”

Watch Outs for Administrators. In thecontext of a closer look at administrators,it makes sense to look at the 10 fireorders and 18 Watch Outs that arepotential guidelines for firefighters,guidelines that remained on a cardinside Rhoades’ ditty bag, untouchedand unread. I want to make two points

about a breakdown is that the commun-ications broke down everywhere, whichis an inevitable diagnosis when youargue that the buck stops everywhere.

Here’s what good communication lookslike. The example comes fromWinston Churchill. When he discoveredto his horror that Singapore wasvulnerable to a Japanese land invasionduring WWII, Churchill said, “I ought tohave known. My advisers ought to haveknown and I ought to have been told andI ought to have asked” (Allinson, 1993,p. 11). Notice how much complexityChurchill has described. There is no onecause for this disaster. Churchill couldhave known. Others should have known.Those who should have known, shouldhave informed Churchill without hisasking. If others did not know, theyshould have found out and informedChurchill on their own without waiting forhim to ask. If they didn’t know, Churchill,by inquiring of them, might have proddedthem to find out. If they had known butfailed to speak up, Churchill, by inquiring,may have been given the necessaryinformation. Any of these eventualitiesmight have changed the course ofevents (adapted from Allinson 1993,pp. 11-12).

It is everyone’s responsibility tochallenge and to respond to thechallenges in a trustworthy manner, andto listen carefully and respectfully to theresponse. When people fail to engagein respectful interactions (Weick, 1993,pp. 642-644), things can get dangerous.Let me suggest why that happens.

One possibility in wildland firefighting isthat a norm has developed which saysessentially, no news is good news.Partly because people on crews areindependent, adventuresome, take-charge people; partly because radiotraffic is so hard to control; partlybecause there are no detailed andsystematic communication protocols fordispatchers and crew leaders toexchange information about changes infire status; and partly because peoplepresume the basic task itself isstraightforward, a failure to report is

treated as a positive message thatthings are OK. Notice, that if things arenot OK and people are preoccupied andunable to send a message, this too willresult in a failure to report.

Thus, no reporting can mean eitherthings are OK or things are not OK. TheZebrugge Ferry disaster on March 6,1987 involved this very misunder-standing. The person responsible forclosing the bow doors of the ferry didnot report any deficiency to the Captain,not because there was none, butbecause he had fallen asleep beforeclosing the doors. The Captain steamedinto the channel unaware that the doorswere open and water was flowing intothe vessel. Five minutes after leavingthe coast of Zebrugge, the ferry Heroldof Free Enterprise capsized, sank, and193 lives were lost. The buck stopseverywhere on this incident. Virtuallythe same scenario happened 5 yearsearlier on October 29, 1983 aboard theferry Pride , but was caught before theship capsized. At that time, the Masterurgently communicated withmanagement requesting that there besome indication on the bridge whetherthe watertight doors were closed or not(Allinson, 1993, p. 203). Managementdid not listen. Their responses to thisrequest are preserved in the accidentinvestigation and included remarkssuch as, “Nice, but don’t we alreadypay someone?”; “Assume the guy whoshuts the doors tells the bridge if thereis a problem”; “My goodness.” Peopleat the top didn’t feel it was part of theirjob to inquire, or to listen attentively, orto pass along information. So there isno reason for the Masters’ of the vesselsto act differently if this is the preferredcommunication style at the Peninsulaand Oriental Steam NavigationCompany (Allinson, 1993, p. 195).

People associated with South Canyondidn’t know a lot of things they shouldhave known. This raises at least 3questions: why weren’t they told, whydidn’t they ask, and why didn’t they tellwhat they knew? They may not havebeen told because others thought theinformation would have no effect, was

50

Part 4 of 4Safe areas for administrators arecreated by such things as clear normsabout the relationship between failureand learning, secret ballots, anonymousreporting of near misses, access tobrainstorming where evaluation of ideasis intentionally suspended, theequivalent of a penalty box wherepeople who commit glaring errors areput for a finite period of time after whichthey rejoin the action, and availability of3rd parties to mediate conflicts that aredifficult to resolve. It is the veryavailability of these safe areas thatallows administrators to act in a candidmanner that can then be mirrored onthe fireline.

If a firecrew sees that management isviolating its own version of LCES, theyshould be just as wary and alert as ifthey saw themselves violating LCES atthe fire itself. The dangers, in eithercase, are real, immediate, and serious.

Moving Toward Solutions

My analysis so far has been largelyspeculative and has consisted ofextrapolations from what is knownabout high reliability organizations toseemingly analogous circumstances inSouth Canyon. Given the tentativequality of this diagnosis, it is prematureto talk about remedies. Nevertheless,remedies have already been implied inwhat I’ve said and I want to illustratebriefly some directions in which thoseimplications point.

1. If leadership is an issue, then itseems important to look more closely atthe possible pathways by which onecan become a smokejumper foreman,whether the route is through expertisewith parachutes, or with leadership, orwith fires. Depending on which route isfavored, people in the field could havevery different habits they fall back onwhen put under pressure.

about these two lists. First, I thinkfirefighters should begin to compile a listof Watch Outs for administrators. In thesame way that the current 18 WatchOuts alert crews to increased hazardsat the site of the fire itself, administratorWatch Outs would alert crews toconditions back at headquarters thatare just as hazardous as the fire itself.Recall that Longanecker (Report, 1994,p. A5-54) proposed just such a watchout in his statement after South Canyon:Watch out “when you don’t receive theresources that you need or you aredebating with the dispatcher about theresources you need.” A handful of otherWatch Outs might include, Watch out,

1. When the governor is in town (Report,1994, p. AF-64);

2. When interagency ties are strained(Report, 1994, p. A5-63);

3. When dispatchers keep track of thingsin their head rather than on paper;

4. When the norms for radio disciplineare loose (Report, 1994, p. A5-37);

5. When people are reluctant to ask forhelp;

6 When administrators are getting on-the-job training;

7. When administrators say “keep itsimple;”

8. When the overhead is tough to find(Caballero statement); and

9. When you don’t know which office toreport to, you think about it, andhaving thought about it you then goto the wrong one (Taft statement).

The second point I want to make is thata good place to start in developing a listof administrative Watch Outs is withexisting efforts to boil the ten fire ordersdown to the acronym LCES (Gleason,1991). If lookouts, communication,escape routes, and safe areas, are goodenough for firefighters, they are goodenough for administrators. Theprinciples are essentially the same ineither case. For example, theadministrative counterpart of lookouts isa person with the big picture. In nuclearpower plant control rooms, there is aperson called the shift foreman (Weick,1987, p. 116) whose sole responsibility

is to maintain the big picture. The mosteffective aircraft cockpit crews are thosein which, during an emergency, theaircraft is flown by the first officer (co-pilot) not the captain and the captainplans how to deal with the emergencyand tracks progress.

Although, I have already discussedcommunication , a good way toillustrate it is by a surprising finding instudies of captains who lead the bestaircraft crews. Investigators found thatthese captains readily acknowledge thattheir decision making ability is not asgood in times of emergency as it is atother times (Helmreich, Foushee,Benson, & Russini, 1986). Captainswho are the worst leaders, say thattheir decision making ability is just asgood in time of emergency as it is atother times. Poor leaders don’t listenbecause they don’t think they need to.Good leaders don’t fall into that trap.Recall an earlier point I made that apotential trap when people gainexperience is that they lose opennessto new information. Here we see clearevidence that good pilots—and byextension, good leaders in general—don’t let that happen.

Escape routes for administratorsconsist of things like options, revocableactions, pulling the plug, seeing thetemptation to escalate a commitment tosalvage a losing cause and thenavoiding it. The scary thing aboutadministrative escape routes, is thatsometimes they are used to denyindividual responsibility and to pass thebuck. That’s the mind set that we wantto undercut with a culture where thebuck stops everywhere. Managersresponsible for treating people withrespect need to have the welfare ofthose people in mind and not just theirown reputation, when they vow never toget into anything without having a wayout for everyone. Safe flight operationson aircraft carriers are made possiblebecause that’s precisely what managersbelieve and put into practice (Weick &Roberts, 1993).

51

Part 4 of 42. If people feel there are too many rulesbinding on firefighters (10 fire orders +18 Watch Out situations + 4 LCES + 4common denominators + 3 sources ofjudgment error [ignorance, casualness,distraction] + 9 guidelines for indirect/downhill line construction = 48, inGleason, 1994, pp. 24-25), and if fire-fighters say they need to violate ordersto keep fires from growing (Rhoades,1994, p. 22), then clearly some prioritysetting is in order. It is here where Ithink it makes sense to talk about simul-taneous centralization and decentraliza-tion. What you want to do is centralizeeveryone in terms of 3 or 4 key valueswhich are treated as non-discretionaryand imperative (LCES?), anddecentralize the others issues so thatthey serve as guidelines and a platformfor improvisation to meet unanticipatedlocal conditions. I have no idea what thefinal partitioning of Gleason’s 48 guideswould look like. I do know that discus-sions to hammer out such a partitioningwould strengthen the will to communicate.

3. I would pay close attention to whatpeople overlearn during their training,since this is what they are most likely todo when put under pressure. Forexample, the 23 people (Report, 1994,p.14), who fled from the ridgeline didnot take the shorter, safer, more directroute used by Haugh, Hipke, andErickson, but instead ran out the sameway they had hiked in, which exposedthem to more danger for a longer period.If firefighters haven’t practiced and over-learned shelter deployment, or droppingtheir tools, or using a checklist, orwatching out for the safety of a buddy,or running from fire as fast as possible(Maclean, 1992, p. 272) over and over,then it’s a safe bet they won’t do thosethings either when they are underintense pressure.

4. I think Dave Thomas (1994, pp. 45-48) is right in his insistence that firestories and case studies are a crucialmeans to extend people’s repertoire ofexperience, even if that experience issecond-hand. There certainly areenough “old fire dogs” around to make

it possible for live cases to be made aregular part of training. Our research onsocialization of newcomers on aircraftcarriers suggests that old hands whotell war stories are an invaluable sourceof training. Remember, we’re talkingabout organizations in which it is hard tolearn by trial and error. The next errormay be the last trial. If trial and errorlearning is limited, then case studiesbecome very important.

5. I think there is a key training lessonin the recent experience with airlinetraining in cockpit crew management.This training didn’t have much effect orcredibility until the people being trainedwere put in flight simulators where theysolved in-flight problems and werevideo-taped doing so (Helmreich andFoushee, 1993, p. 28). Pilots sawthemselves actually committing theerrors that up to then, had only beendescribed in dry classroom lectures.And what may have been most crucialin this Line Oriented Flight Training isthat each videotape was erasedimmediately after the performance hadbeen critiqued. Videotapes of crewinteraction during fires, of dispatchersallocating scarce resources, or ofadministrators briefing local propertyowners, all could prove to be a valuablewindow on just how well the struggle foralertness is being waged.

I know these are all small solutions topotentially big problems. But they are astart, they can be done in parallel, theycan be done simultaneously in differentplaces, and they may stimulate a betterset of starting points.

Lingering Questions

Even though I have some hunchesabout what might have been going onin South Canyon, there are somequestions that continue to baffle me.For example, how is it possible that somany fire orders and Watch Outs werebeing violated (20/28 were violatedaccording to the South Canyon

investigation team, p. 3), enoughviolations that Rhoades was scared tocount them, yet Ryerson is quoted in theWall Street Journal (8-22-94) as saying“it happened fast enough that none of usknew we were in danger . . . It happenedin a matter of seconds” (Page A1,column 1) and Blanco called dispatchshortly before the blowup “and toldthem that things looked good” (Report,1994, p. A5-11)? I realize that Ryersonprobably means the blowup itselfhappened fast, yet conditions had beensteadily worsening and the blowup wasnot the first moment people senseddanger. People either weren’t keepingscore of the number of violations, ordidn’t want to know the score, orbecause they arrived at different timeswith different information had a differentsense of the number of violations.

A different set of questions concernsthe role that groups play. Why didn’tthe Prineville Hotshots speed up, lookback, drop their tools? Perhaps theydidn’t think they were in great danger.The fire could have burned straightuphill toward the lunchspot. But whatwe may also be seeing here is the flipside of what I think happened at MannGulch. At Mann Gulch the groupdisintegrated, which led to a loss ofmeaning and then to somethingapproximating panic. At South Canyonthe group remained together (Report,1994, p. A4-10) and things stayedmeaningful, but people held onto thewrong meaning. Imagine what a typicalhotshot might be thinking. Ericksonand Haugh are strangers and jumpersto boot; they are saying “run,” but thishas been a sloppy operation from thestart. Furthermore, we didn’t hearanything about a weather front nor didwe hear the argument about cuttingdirect line downhill, so presumably we’resafe and they’re probably exaggerating.

It may be that group ties were too tightamong the hotshots, the level of concernwas too low, and the meaning persisted,like it did at Mann Gulch, that this is justone more 10:00 fire.

52

Part 4 of 4function of deflecting blame from admini-strators onto crews, and are only inci-dentally relevant to safe practice. Withthis many guidelines in place, it’s fairlyeasy after the fact for administrators tospot at least one violation that occurredand to spotlight it as THE cause of theaccident.

My point here is not to be cynical.Instead, I want to raise the possibilitythat the system may know less aboutfirefighting than it thinks it does. Themultiple guidelines give the impressionthat much is known, but the guidelinesmay be redundant, they may say thesame thing in several different ways.The result may be that when peopletake these guidelines seriously theyreduce their ability to sense subtlevariations in fire behavior and thereforeundertake more dangerous actions.The guidelines may shield management,but they also may create blindspots forfirefighters. I think that possibility needsto be explored carefully.

If it turns out that the 48 guidelines sayjust a handful of different things andanticipate a relatively limited set ofvariations in fire behavior, then effortsshould be made to develop a morecomprehensive, more varied set ofguides. If it turns out that all 48 aredifferent, varied, and necessary, then itwould seem important either to prioritizethem as mentioned earlier, or divide upresponsibility for them among the crew.If there are 48 guidelines and 10 crew,then each crew member would beassigned 5 guidelines to monitor,champion, and communicate.

Conclusion

Something that both Mann Gulch andSouth Canyon share in common is aseries of events in which something verysmall escalated into something mon-strous. A good example of two eventsthat can be caught in an escalating spiralthat starts small and ends monstrous arethe events of “fear” and “understanding.”As fear increases, understanding

Perhaps there is such a thing as a groupbeing too disciplined and too cohesive.High cohesion wards off panic, but italso encourages groupthink and wardsoff more disturbing and more variedmeanings of what may be happening.Variety may have been crucial tosurviving this incident. The 12 peopleclimbing up the fireline toward the ridgeall did the same thing and perished.The other 37 people on the mountaindid different things, most of whichworked. Three ran to the top of thefireline; 8 ran above the lunch spot anddeployed shelters; 1 stayed at thelunchspot; 23 headed for Helitack 2 butthen stopped and went down variousportions of the east drainage; and all ofthese people lived. Two people tried tomake it to Helitack 2, but failed. To putit in the most extreme form, thehotshots didn’t panic and that may havebeen their problem. If they had comecloser to doing so they might havelived. I know how bizarre that sounds.But it’s important to realize that we aredealing with strong, competing, humantendencies toward independence andconformity. That lies behind respectfulinteraction. People need sufficient socialsupport to stay calm and sufficientindependence to be innovative. Peoplewho fight wildland fires aren’t freed fromthis dilemma simply because they arebold. As long as crews and danger anddifferent experiences mix together, wecan expect puzzling outcomes.

Notice that we can take a totally differentapproach in analyzing the Hotshots’behavior. Earlier, I argued that becausethey were poorly treated in GlenwoodSprings, they may have been undersome stress when they got to SouthCanyon. If, in addition, they had doubtsabout the safety of what they weredoing, then the level of stress might havebeen quite high when they were orderedto retreat to the ridge. If, during hotshottraining, people overlearn paramilitarydiscipline, regimentation, and obedience,then we would expect this pattern ofdiscipline to be especially visible underhigh stress. The general idea is thatwhen stress increases, people fall backon overlearned habits. Thus, the brisk,

well-spaced, steady march up thefireline toward the ridge with tools inhand, may represent the behavior of agroup under enormous pressure ratherthan that of a group that is relativelycalm and thinks this is just another fire,albeit one that has been has beenmanaged a bit more poorly than usual.

A further puzzle at South Canyonconcerns the possibility that this fire fellin a kind of “no man’s land” at a crucialperiod. Jumpers who dropped on thefire the night of August 5th found a firethat seemed larger than an initial attackfire for which they are experts. Whenthe shots began trickling in aroundnoon on August 6th, they found a firethat seemed smaller than fires forwhich they are experts. The result is afuzzy situation where the fire is too bigfor some, too small for others, and tooforeign to the experience of the peoplein charge. The problem may not bethat a transition was mishandled andresulted in fatalities. Rather, theproblem at South Canyon may havebeen that the complexity of the fire felloutside the scope of everyone whotried to control it. If that’s plausible thenit suggests the need for rethinking theadequacy of existing fire categories andtheir matchup with training andexpertise. Problems may occur notonly when fires move from onecategory to another, but also when theydefy categorization in the first place.

As a final lingering question, I wonder if48 guidelines might be too fewguidelines to be of much help tofirefighters? There seems to be lots ofoverlap and similarity among theguidelines, so much so in fact that if westudy them closely, we might discoverthat they have too little variety to matchthe large amount of variety in wildlandfires. If that were possible, then itwould explain why firefighters feel theyhave to violate orders. They do so toregain the variety of attack they feel isnecessary to combat the variety in thefire they face. The possibility that 48guidelines actually reduce requisitevariety is also consistent with the ideathat these guidelines may serve the

53

Part 4 of 4decreases, which causes fear toincrease even more, which leads toeven less understanding, and thisescalation increases until somethingexplodes. That could be what happensas people discuss how to prevent moreSouth Canyons. But if the discussionleads to more understanding, then wecreate a world where more fear leads tomore discussion which leads to moreunderstanding which leads to less fear.My remarks should be understood asan invitation to discussions that improveour understanding and lessen our fears.

References

Allinson, R. E. (1993). Globaldisasters. NY: Prentice-Hall.

Cooley, E. (1984). Trimotor and trail.Missoula, Mont.: Missoula Press.

Dodge, W. (1949). Testimony. MannGulch Transcript (pp. 117-125).Service Washington, DC: US ForestService.

Fite, F. (1949). Testimony. MannGulch Transcript (pp. 26-32).Washington, DC: US ForestService.

Gleason, P. (1991). LCES-A key tosafety in the wildland fireenvironment. Fire ManagementNotes, 52 (4), 9.

Gleason, P. (1994). Unprepared for theworst case scenario. Wildfire, 3 (3),23-26.

Helmreich, R. L., & Foushee, H.C.(1993). Why crew resourcemanagement? Empirical andtheoretical bases of human factorstraining in aviation. InE. Weiner, B. G. Kanki, & R.L.Helmreich (Eds.), Cockpit resourcemanagement. (pp. 3-45). SanDiego: Academic Press.

Helmreich, R. L., Foushee, H. C.,Benson, R., & Russini, W. (1986).Cockpit management attitudes:Exploring the attitude-performancelinkage. Aviation, Space, andEnvironmental Medicine, 57, 1198-1200.

Maclean, N. (1992). Young men andfire. Chicago: University of ChicagoPress.

March, J. G., Sproull, L. S., & Tomuz,M. (1991). Learning from samplesof one or fewer. OrganizationScience, 2, 1-13.

McGrath, J. E. (1976). Stress andbehavior in organizations. In M. D.Dunnette (Ed.), Handbook inindustrial and organizationalpsychology (pp. 1351-1395).Chicago: Rand-McNally.

Miller, D. (1993). The architecture ofsimplicity. Academy ofManagement Review, 18, 116-138.

O’Reilly, C. A., & Caldwell, D. F.(1981). The commitment and jobtenure of new employees: Someevidence of postdecisionaljustification. Administrative ScienceQuarterly, 26, 597-616.

Putnam, T. (1994). Analysis of escapeefforts and personal protectiveequipment on the South Canyonfire. USDA Forest Service,Missoula Technology andDevelopment Center.

Pyne, S. J. (1984). Introduction towildland fire. NY: Wiley. Report ofthe South Canyon Fire AccidentInvestigation Team. August 17,1994.

Rhoades, Q. (1994). Effective firefighting calls for bending the rulessometimes. Wildfire, 3 (3), 22.

Ross, J., & Staw, B. M. (1986). Expo86: An escalation prototype.Administrative Science Quarterly,31:274-297.

Rumsey, W. (1949). Testimony. MannGulch Transcript (pp. 97-109).

Sagan, S. D. (1993). The limits ofsafety. Princeton University Press.

Salancik, G. R. (1977). Commitmentand the control of organizationalbehavior and belief. In B. M. Stawand G. R. Salancik (Eds.), Newdirections in organization behavior.(pp 1-54). Chicago: St Clair.

Sallee, R. (1949). Testimony. MannGulch Transcript (pp. 69-89).Washington, D.C.: U.S. ForestService.

Schulman, P. R. (1993). Thenegotiated order of organizationalreliability. Administration andSociety, 25, 353-372.

Thoele, M. (1994). Firefightersemphasize safety, but. Wildfire, 3(3), 27-29.

Thol, H. J. (1949). Testimony. MannGulch Transcript (pp. 183-202).Washington, D.C.: U.S. ForestService.

Thomas, D. (1994). A case for firebehavior case studies. Wildfire, 3(3), 45-47.

Weick, K. E. (1987). Organizationalculture as a source of highreliability. California ManagementReview, 29 (2), 112-127.

Weick, K. E. (1990). The vulnerablesystem: An analysis of the Tenerifeair disaster. Journal ofManagement, 16, 571-593.

Weick, K. E. (1993). The collapse ofsensemaking in organizations:TheMann Gulch disaster.Administrative Science Quarterly,38, 628-652.

Weick, K. E., & Roberts, K. (1993).Collective mind in organizations:Heedful interrelating on flight decks.Administrative Science Quarterly,38, 357-381.

54

Part 4 of 4 Appendix E—Related Reports

The Collapse ofDecisionmaking andOrganizational Structureon Storm King MountainTed Putnam, Ph.D., Protective Clothing andEquipment Specialist, Missoula Technologyand Development Center, May 1996

Stress, fear, and panic predictably leadto the collapse of clear thinking andorganizational structure. While thesepsychological and social processes havebeen well studied by the military and theaircraft industry (Cockpit ResourceManagement) (Weick 1990 and Wiener,Kanki, and Helmrich 1993), the wildlandfire community has not supported similarresearch for the fireline. The fatal wild-land fire entrapments of recent memoryhave a tragic common denominator—human error. The lesson is clear: study-ing the human side of fatal wildland fireaccidents is overdue.

Historically, wildland fire fatality investi-gations focus on external factors like firebehavior, fuels, weather, and equipment.Human and organizational failures areseldom discussed. When individualfirefighters and support personnel aresingled out, it’s often to fix blame in thesame way we blame fire behavior orfuels. This is wrong-headed and danger-ous, because it ignores what I think is anunderlying cause of firefighter deaths—the difficulty individuals have to consist-ently make good decisions under stress.

There’s no question individuals must beheld accountable for their performance.But the fire community must begindetermining at psychological and sociallevels why failures occur. The goalshould not be to fix blame. Rather, itshould be to give people a better under-standing of how stress, fear, and paniccombine to erode rational thinking andhow to counter this process. Over theyears, we’ve made substantial progressin modeling and understanding theexternal factors in wildland fire suppres-sion, and too little in improving thinking,leadership, and crew interactions.

Decisionmaking—A Telling Model

Human thinking and decisionmakinghave been studied and modeled. Thedecision process is essentially additive:A+B+C. For example, a decision to buildfireline may be characterized by fire-fighters (FFa, FFb, FFc, FFd) basingtheir choice on these factors:

FB—fire behaviorW—weatherFL—fuelsE—equipmentP—personnel, experience, skillS—safetyM—expectations of management

Numerous studies show no matter howmany factors are important, the humanmind normally can handle only aboutseven factors (e.g., seven-digit telephonenumbers). People differ both as to howmany factors they use and the valueplaced on these factors. In this modeling,the first factor is the one each firefighterpays the most attention to with the otherfactors added in decreasing level ofimportance.

So the decisionmaking processingleading to fireline building could bemodeled:

FFa = M+W+FB+S+P+E+FLFFb = S+P+M+FBFFc = FB+P+EFFd = P+E+S+FB+W

Although their decisions were the same,they arrived at them through quitedifferent factor evaluations.

However, in situations that create stress,fear, and panic, minds regress towardsimpler, more habitual thinking. Thisregression could be modeled:

FFa = M+W (Get the work done, weatherpermitting)

FFb = S (Safety first)FFc = FB (Fire behavior most important)FFd = P+E (People and equipment

dominant)

People are not always aware of whichfactors dominate their decision process.Although we say “safety first,” this doesnot mean it’s necessarily first in actualdecisions. Also, people are seldomaware of the few factors they actually areprocessing, so they tend to be overcon-fident in their decisionmaking ability.Although people are unable to use all theavailable information for decisionmaking,especially when under stress, computershave no such limitation. Computersprocess information interactively, AxBxC,and can use most of the available infor-mation for better decisions. People arevery good at determining the state ofeach factor, the inputs, but not so goodat integrating all the factors to make adecision. While computers are of help toincident management teams, normallythey aren’t available for extended initialattack.

So when fireline conditions are routine,most people would reach similar deci-sions because they are more aware andtake more information into account.When fireline conditions worsen,decisions are more at the mercy of theone or two factors individuals are stillprocessing and their level of experience.In the example above, under stressfulconditions even though each firefighter’smain factors differ, if they readilycommunicate as a crew, most of thefactors are still present. Althoughindividual decisions are additive, wheregood communications exist, the groupdecision can approach the better inter-active process.

Studies also show that our linear thinkingtends to underestimate hazards, particu-larly if the hazard is increasing at alogarithmic or exponential rate as canhappen on the fireline. An examplewould be estimating rates of fire spread.A computer would give the betterdecision in a heartbeat. People wouldtend to underestimate the rate of spreadand have difficulty deciding on anappropriate course of action. And so itis important to understand the limits ofhow we process information and thecommon types of errors that can occur.

55

Part 4 of 4

Leadership and GroupBehavior

Stress, fear, and panic take their toll atall levels of the wildland firefightingorganization. Under stress, leadershipbecomes more dogmatic and self-centered. It regresses toward morehabituated behavior. Groups tend tofragment under stress into smaller unitsor to stick together and follow their leaderwithout joining the decisionmaking pro-cess. Either way, most of the informationavailable for the best decisions is notutilized.

An extensive 12-year study of ForestService field crews conducted by sociol-ogist Jon Driessen (1990) showed thereis an inverse correlation between crewcohesion and accident rates. The studyalso identified factors fostering cohesion.Driessen found it takes about 6 weeksfor good crew cohesion to take effect.So firefighting crews are predisposedtoward accidents until they becomecohesive units. Unfortunately, this typeof information is not normally consideredeven when sending crews to riskier fires.

An excellent case study of leadershipunder stress on a smaller scale is Dr.Karl E. Weick’s The Collapse of Sense-making in Organizations: The MannGulch Disaster (Weick 1993). Althoughthe leadership and organizational struc-ture discussed are based on NormanMaclean’s Young Men and Fire , Weick’sanalysis is thought-provoking. It is alsohaunting because the South CanyonFire Investigation report shows thehuman and organizational failures onStorm King Mountain are similar to thosehe hypothesizes happened at MannGulch 45 years earlier.

Risk-Taking inWildland Firefighting

First, wildland fires cannot be foughtwithout risk. Making decisions while atrisk assumes firefighters can evaluatethe likelihoods of various states of nature.On larger fires, with structured incidentmanagement teams (IMT), specialists,and portable weather stations, etc., thelikelihoods are more objective and out-comes are better predicted. An excellentstudy of leadership under stress on alarger (IMT) scale is Taynor, Klein, andThordsen’s 1987 article, DistributedDecisionmaking in Wildland Firefight-ing . They describe the IMT as a veryrobust organization due to lengthyexperience levels, the commonexperience of working together,excellent communication structure, andwell-defined, well-practiced roles. Incontrast, on smaller fires, the likelihoodsare more subjective, based on skill andexperience rather than instruments.When small fires grow larger and morecomplex, such subjective estimatesbecome less accurate, and decision-making regresses to a reliance on fewerand fewer factors. The result is a failureto keep up with rapidly changing condi-tions, and people on the fireline are putat greater risk.

Second, risk-taking is subject toperceived and actual rewards andpunishments. When we attach a stigmato deploying a fire shelter, we bias fire-fighters into taking more risks to escape.If there’s a stigma associated withdropping packs and tools, firefighters willcarry everything while trying to outrun afire. If a stigma is attached to abandoninga fire or the fireline, firefighters will takemore risks to control a fire. The variouspayoffs associated with risk-taking arenot necessarily those managers claimare operating. We need professionalsspecializing in the study of decisionmak-ing under stress to interview managersand firefighters, so we can begin tobetter understand actual risk-taking onthe fireline.

Collapse ofDecisionmaking onStorm King Mountain

On the South Canyon Fire the firstdecision failures occurred at the BLM(Bureau of Land Management) districtlevel. Although the fire started July 2 ina fire exclusion zone, resources did notreach the fire until July 5. It was theworst fire season in years and localresources were stressed. Holding costsdown and making do with localresources dominated decisionmaking.From our earlier analysis, we can predicta tendency to fall back on habituatedtactics, such as letting the fire go until alocal crew is available. Although manycrews were available nationally, thedistrict did not request help until July 5.The longer initial attack was delayed, thegreater the risk the firefighters faced.

An incident commander (IC) from thelocal BLM district arrived on the fire themorning of July 5. But because ofmechanical problems with their chainsaws, the IC and crew left the fire thatevening as a load of smokejumpers weredropped onto a nearby ridge. The firstperson out the door of the jumper aircraftbecame the jumper-in-charge (JIC). Viaradio the IC turned the fire over to theJIC. This situation raises two immediateleadership questions: Why did the ICleave the fire? Was first experiencedperson out the door the best way tochoose the JIC?

The jumpers fought the fire most of thenight as it continued to grow in size. Inresponse, the JIC ordered two moreType I crews. The IC returned with hiscrew the morning of July 6. By 10:30a.m., a second load of jumpers arrived,and the JIC of that plane load becamethe line scout (LS). The IC and his crewstayed on top of the ridge buildingfireline, while the jumpers beganconstructing fireline downhill on the westflank. At 12:30 p.m., 10 members of thePrineville Hotshots (PHS), including

56

Part 4 of 4their superintendent, arrived at the fire.The IC, JIC, and PHS superintendentagreed to send 9 PHS down to helpbuild fireline on the west flank. At 3:00p.m., the remaining 10 PHS arrived atthe fire and stayed on top of the ridgewith their superintendent to help the ICand his local crew.

So the organization structure before theblowup was:

Location Local NationalResources Resources

Ridgetop 9 BLM District 11 PHS2 USFS District2 Helitack

West flank None 9 PHS8 Missoula SJ4 McCall SJ2 North Cascades SJ1 West Yellowstone SJ1 Grangeville SJ

All the ingredients were in place for acatastrophe. Three local crews (BLM,USFS, Helitack), the Prineville crew splitinto two groups, and jumpers from fivedifferent bases led by two somewhatrandomly selected JIC’s were throwntogether and asked to perform as a teamunder increasingly unstable conditions.Neither leadership roles nor a cohesiveorganizational structure stabilized beforethe blowup.

On the west flank, a group of ninesmokejumpers split off to construct fire-line to the southwest, forming a thirdgroup. These three groups began tofocus on their own immediate problemsand communications among themcontinued to decline. As the wind pickedup after 3:00 p.m., so did fire activity andfirefighter stress levels. And, predictably,decisionmaking and organization col-lapsed inward, with fatal consequences.

From the South Canyon Fire Investi-gation report and witness testimony, wecan find signs of collapse similar to thoseWeick identified in his analysis of MannGulch, including:

• Leadership questioned and challenged(for incident commander, jumper-in-charge, and line scout).

• Decisions questioned.

• Most experienced people not consultedand locked out of decision process.

• Poor communication concerningdeteriorating conditions—especiallyamong groups.

• Continued fragmentation into smallergroups.

• Decreased talking within groups.

• Failure to integrate vital, availableinformation when changes occurred.

• Failure to act on the weight of theevidence.

• Underestimating the current andpotential fire behavior.

Once the blowup occurred, in the ensu-ing stress, fear, and panic, people’sactions followed classic lines ofregressing to more habituated patternsof behavior:

• On the ridgetop all but two people ranout the east drainage, a potential deathtrap. This was not a matter of thoughtas much as regression—going backthe way you had come in.

• The two helitack refused to go into theeast drainage and ran back along theridge they had been dropped off on,possibly looking for a copter pickup site.

• The west flank SJ and PHS went backup the fireline they had been digging.

• Virtually all the escaping firefighterscarried their tools and packs eventhough it cost many of them their lives(Putnam, 1994).

• Even when the firefighters were yelledat to drop their tools and equipment,they did not. This deeply ingrainedresponse pattern resulted in fatalities.

• Even though their lives were at stake,very few firefighters made any attemptto use their fire shelters, resulting in a

higher number of fatalities (Putnam1994).

• Although firefighters knew what fireshelters were and how to open them,they clearly did not know how to usethem effectively or where they wouldwork best.

Training to MakeDecisions Under Stress

Courses such as Cockpit ResourceManagement train crews to counteractthe natural tendencies for behavioralregression. Countermeasures mentionedby Weick and others include:

• Nonstop communication, both verbaland nonverbal is crucial, especiallywhen people first come together.

• Survival goals (threat recognition,escape, shelter use) must be over-learned through repeated practice orthey will not be dominant in dangeroussituations.

• Cross-train in roles.

• Value wisdom and openness.

• Initiate respectful face-to-faceencounters between crew membersand between crews.

• Remain curious and observant.

• If things don’t make sense, speak up.

• Avoid overconfidence and overcau-tiousness.

• When situations deteriorate, pay moreattention to leadership, perceptions,and group interactions. Strengthen ties.

• Group dynamics before a crisis affectsurvival during a crisis.

• Expect everyone to work safely, com-municate effectively, and cooperate.

57

Part 4 of 4• Talk to other crew members and crews.

Expect them to talk to you—then listen.

• Be especially wary of accepting incre-ments of worsening conditions. It isdeceptive to accept the incrementsrather than the entire change.

It is apparent from this list that to beadequately prepared requires training,overlearning, and using these skillsroutinely before a crisis strikes. It is alsoclear these skills are a necessaryprerequisite for effective decisionmakingconcerning integrating fire behavior,weather, fuels, equipment, and humanfactors.

A Start

Within the wildland fire agencies, aware-ness is growing about the value ofcockpit resource management typetraining and the need to pay more atten-tion to psychological and sociologicalaspects of fighting fires. Paul Gleason,a seasoned hotshot superintendent,believes that the 10 Fire Orders , 18Watchout Situations , and 9 Downhill/Indirect Line Construction Guidelinescan be information overload for thefirefighter on the line. For this reason hebelieves four of the key factors shouldbe constantly emphasized: Lookouts,Communications, Escape routes, andSafety zones (LCES) as central to safefirefighting (Gleason 1991,1994 ). Weknow from our previous model that 30+warnings are an overload under normalconditions (seven is the practical limit) soLCES, while based on the others, is anexcellent system because it is manage-able in crisis situations. Since LCES iseasy to use, firefighters can constantlyreevaluate their situation. Gleason con-cludes that a change in training contentis not needed and that we need to betterpractice what we already know.

However, I’m arguing that a differentkind of training is needed to be able touse our existing knowledge (including

LCES) in crisis situations. To link thehuman factors involved in firefighting tothe classic Look Up, Look Down, LookAround , we can add Look Inside . Andwe could change LCES to I-LCES,where the “I” means Inside, Inner, andInterpersonal.

Patrick Withen, a smokejumper andsociologist, has discussed firefighterattitudes and has pointed out (Withen,1994) that there is no way to “just sayno” in firefighting that doesn’t carry formalor informal sanctions. The onus is on theindividual firefighter—not management—to justify the decision. Routinely, there isa stigma attached to leaving the fireline.

While looking at the firefighter from psy-chological and sociological perspectivesis encouraging, this idea has not beenwell received by many in the wildlandfire community. When suggested to theSouth Canyon Fire Investigation Teamand the follow-up Review Board as apossible causal factor, the suggestionwas dropped from further consideration.Their strongest recommendations shouldcome as no surprise—improve firebehavior prediction, improve weatherforecasting, develop better fuel invento-ries, and look at our firefighting institutionfrom the external perspective. Thesetried-and-true solutions simply fail todeal with a major cause of the fatalities.

We lost firefighters on Storm KingMountain because decision processesnaturally degraded. At this time we donot have training courses that give fire-fighters the knowledge to counter theseprocesses. Both the Investigation Teamand Review Board recommendedcreating a passion for safety but did notacknowledge that this passion is deter-mined by psychological and sociologicalprocesses. The type and skill level ofinvestigation team members and reviewboards (typically they include IMTpersonnel, a fire weather forecaster, firebehaviorist, fuels specialist, equipmentspecialist, but no psychologist or sociol-ogist) predisposes them to focus on thetraditional inputs, which effectivelyexcludes other types of input, hencepredetermining the outcome. This calls

into question the very process and struc-ture by which we investigate fatalities andcommunicate the results to the fire com-munity. We can and ought to do better.

Discussion

There is no intent here to blame theindividual firefighters and managers forwhat they did or didn’t do related to thefire on Storm King Mountain. The realissue is that we are not preparing ourfirefighters and managers to operatewith maximal effectiveness underknown stressful, risky conditions. Theprocesses and papers cited, whenconsidered in the light of the SouthCanyon Fire Investigation report, clearlydemonstrate that an almost automaticcollapse of decisionmaking and organi-zational structure occurred. It should alsobe clear that we are not unique in oper-ating under stressful, risky conditions.Other organizations have reduced fatali-ties through training using techniqueswith a proven track record. Paying moreattention to the psychological andsociological processes of our people islong overdue.

It is clear that even our best crews arenot adequately trained in escapeprocedures and fire shelter use. This isa reflection of the prevailing attitudeamong managers that if we give fire-fighters more training and betterpredictions for fire behavior, fuels,weather, and tactics, entrapments won’thappen. So why plan for them?Individual firefighters agree with theirmanagers and also have the attitude thatit won’t happen to me, so why practicefor an entrapment. These attitudescaught up with our best and brightestfirefighters on Storm King Mountain andwere a causal factor in the fatalities.

Since 1990, extended droughts andmore severe fire behavior haveshortened the time firefighters have todecide whether to try to escape or todeploy shelters. Some 23 firefighters

58

Part 4 of 4have perished trying to escape uphillcarrying packs and equipment. Estimatesshow most would have lived had theysimply dropped their gear and run forsafety carrying only fire shelters.

This is why mandatory training for shelteruse, escape, decisionmaking understress, and stress-resistant organiza-tional characteristics should becomenational priorities.

Everyone agrees our top priority shouldbe reducing the number of entrapmentsby practicing safety and LCES. But wealso need to face the reality that onaverage 30 firefighters are trapped eachseason, and that we have not taughtthem how to escape, how to use fireshelters effectively, or the conceptsdiscussed here. Clearly, firefighters needthis type of training. Better personal andinterpersonal skills will enable firefightersto use all their training and experienceoptimally under risky, stressful conditions.

Recommendations

➊ Implement recommendations in fireshelter training stemming from theanalysis of protective clothing andequipment and its use on the SouthCanyon Fire (Putnam, 1994).

➋ Convene a task group of firefighters,fire training and safety officers,psychologists, sociologists, and otherswho will recommend specific actionsfor individuals and groups that willmaximize their resistance to decisionand organizational collapse understressful conditions.

➌ Develop a training program to com-municate these new skills to person-nel such as Incident ManagementTeams, Type I and II crews, striketeam leaders, and others at risk orwho make decisions under stress.

➍ Analyze the organizational structureof initial attack and extended initialattack crews and how these crewsinterrelate to form an effectiveorganization with optimal leadershipand decisionmaking capabilities.

➎ Develop professional requirements,best skills mix, and organizationalstructure for fatality investigationteams and review boards. Form IMT-type teams before fatalities occur soinvestigation teams are trained andready for dispatch.

➏ Consider adding a Look Insidecomponent to Look Up, Look Down,Look Around and an “I” to LCES.Incorporate an inner check list into theFireline Safety Reference Notebook.

Literature Cited

Driessen, Jon. 1990. The Supervisorand the Work Crew. USDA ForestService, Missoula Technology andDevelopment Center, Missoula, MT.

Fireline Safety Reference. 1992. NFES2243. National Interagency FireCenter. Boise, ID.

Gleason, Paul. 1991. LCES—A Key ToSafety In the Wildland FireEnvironment. Fire ManagementNotes. 52 (4): 9.

Gleason, Paul. 1994. Unprepared forthe Worst Case Scenario. Wildfire. 3(3): 23-26.

Maclean, Norman. 1992. Young Menand Fire. Chicago: University ofChicago Press.

Putnam, Ted. 1994. Analysis of EscapeEfforts and Personal ProtectiveEquipment on the South Canyon Fire.Missoula, MT: USDA Forest Service,Missoula Technology and Develop-ment Center.

South Canyon Fire Investigation Report.1994. NIFC/NFES.

Taynor, Janie; Gary Klein, and MarvinThordsen. 1987. DistributedDecisionmaking in WildlandFirefighting. Alexandria, VA: U.S.Army Research Institute for theBehavioral and Social Sciences.

Weick, Karl. 1990. The VulnerableSystem: Analysis of the Tenerife AirDisaster. Journal of Management.16: 571-593.

Weick, Karl. 1993. The Collapse ofSensemaking in Organizations: TheMann Gulch Disaster. Ithaca, NY:Cornell University.

Wiener, Earl, Barbara Kanki, and RobertHelmrich. 1993. Cockpit ResourceManagement. San Diego: AcademicPress.

Withen, Patrick. 1994. Fire Culture.Inner Voice. September/October1994: 12-13.

59

Part 4 of 4

The Collapse of Sense-making in Organizations:The Mann Gulch DisasterKarl E. Weick. Reprinted from The Collapseof Sensemaking in Organizations: The MannGulch Disaster by Karl E. Weick published inAdministrative Science Quarterly Volume 38(1993): 628-652 by permission of Administra-tive Science Quarterly. © 1993 by CornellUniversity 0001-8392/93/3804-0628.

This is a revised version of the Katz-Newcomb lecture presented at the Universityof Michigan, April 23-24, 1993. The 1993lecture celebrated the life of Rensis Likert, thefounding director of the Institute for SocialRelations. All three people honored at thelecture—Dan Katz, Ted Newcomb, and RenLikert—were born in 1903, which meant thislecture also celebrated their 90th birthdays.I am grateful to Lance Sandelands, DebraMeyerson, Robert Sutton, Doug Cowherd,and Karen Weick for their help in revisingearly drafts of this material. I also want tothank John Van Maanen, J. RichardHackman, Linda Pike, and the anonymousASQ reviewers for their help with later drafts.

The death of 13 men in the Mann Gulchfire disaster, made famous in NormanMaclean’s Young Men and Fire , isanalyzed as the interactive disintegrationof role structure and sensemaking in aminimal organization. Four potentialsources of resilience that make groupsless vulnerable to disruptions of sense-making are proposed to forestalldisintegration, including improvisation,virtual role systems, the attitude ofwisdom, and norms of respectful inter-action. The analysis is then embeddedin the organizational literature to showthat we need to reexamine our thinkingabout temporary systems, structuration,nondisclosive intimacy, intergroupdynamics, and team building.

The purpose of this article is to reanalyzethe Mann Gulch fire disaster in Montanadescribed in Norman Maclean’s (1992)award-winning book Young Men andFire to illustrate a gap in our currentunderstanding of organizations. l wantto focus on two questions: Why doorganizations unravel? And how canorganizations be made more resilient?Before doing so, however, l want to stripMaclean’s elegant prose away from the

events in Mann Gulch and simply reviewthem to provide a context for theanalysis.

The Incident

As Maclean puts it, at its heart, the MannGulch disaster is a story of a race (p.224). The smokejumpers in the race(excluding foreman “Wag” WagnerDodge and ranger Jim Harrison) wereages 17-28, unmarried, seven of themwere forestry students (p. 27), and 12 ofthem had seen military service (p. 220).They were a highly select group (p. 27)and often described themselves asprofessional adventurers (p. 26). Alightning storm passed over the MannGulch area at 4 p.m. on August 4, 1949and is believed to have set a small firein a dead tree. The next day, August 5,1949, the temperature was 97 degreesand the fire danger rating was 74 out ofa possible 100 (p. 42), which means“explosive potential” (p. 79). When thefire was spotted by a forest ranger, thesmokejumpers were dispatched to fightit. Sixteen of them flew out of Missoula,Montana at 2:30 p.m. in a C-47 transport.Wind conditions that day were turbulent,and one smokejumper got sick on theairplane, didn’t jump, returned to thebase with the plane, and resigned fromthe smokejumpers as soon as he landed(“his repressions had caught up withhim,” p. 51). The smokejumpers andtheir cargo were dropped on the southside of Mann Gulch at 4:10 p.m. from2000 feet rather than the normal 1200feet, due to the turbulence (p. 48). Theparachute that was connected to theirradio failed to open, and the radio waspulverized when it hit the ground. Thecrew met ranger Jim Harrison who hadbeen fighting the fire alone for four hours(p. 62), collected their supplies, and atesupper. About 5:10 p.m. (p. 57) theystarted to move along the south side ofthe gulch to surround the fire (p. 62).Dodge and Harrison, however, havingscouted ahead, were worried that thethick forest near which they had landed

might be a “death trap” (p. 64). Theytold the second in command, WilliamHellman, to take the crew across to thenorth side of the gulch and march themtoward the river along the side of the hill.While Hellman did this, Dodge and Harri-son ate a quick meal. Dodge rejoinedthe crew at 5:40 p.m. and took hisposition at the head of the line movingtoward the river. He could see flamesflapping back and forth on the southslope as he looked to his left (p. 69).

At this point the reader hits the mostchilling sentence in the entire book:“Then Dodge saw it!” (p. 70). What hesaw was that the fire had crossed thegulch just 200 yards ahead and wasmoving toward them (p. 70). Dodgeturned the crew around and had themangle up the 76-percent hill toward theridge at the top (p. 175). They were soonmoving through bunch grass that wastwo and a half feet tall and were quicklylosing ground to the 30-foot-high flamesthat were soon moving toward them at610 feet per minute (p. 274). Dodgeyelled at the crew to drop their tools, andthen, to everyone’s astonishment, he lita fire in front of them and ordered themto lie down in the area it had burned. Noone did, and they all ran for the ridge.Two people, Sallee and Rumsey, madeit through a crevice in the ridge unburned,Hellman made it over the ridge burnedhorribly and died at noon the next day,Dodge lived by lying down in the ashesof his escape fire, and one other person,Joseph Sylvia, lived for a short while andthen died. The hands on Harrison’swatch melted at 5:56 (p. 90), which hasbeen treated officially as the time the 13people died.

After the fire passed, Dodge foundSallee and Rumsey, and Rumsey stayedto care for Hellman while Sallee andDodge hiked out for help. They walkedinto the Meriwether ranger station at8:50 p.m. (p. 113), and rescue partiesimmediately set out to recover the deadand dying. All the dead were found in anarea of 100 yards by 300 yards (p. 111).It took 450 men five more days to getthe 4,500-acre Mann Gulch fire undercontrol (pp. 24, 33). At the time the crew

60

Part 4 of 4jumped on the fire, it was classified asa Class C fire, meaning its scope wasbetween 10 and 99 acres.

The Forest Service inquiry held after thefire, judged by many to be inadequate,concluded that “there is no evidence ofdisregard by those responsible for thejumper crew of the elements of riskwhich they are expected to take intoaccount in placing jumper crews onfires.” The board also felt that the menwould have been saved had they“heeded Dodge’s efforts to get them togo into the escape fire area with him”(quoted in Maclean, p. 151). Severalparents brought suit against the ForestService, claiming that people should nothave been jumped in the first place (p.149), but these claims were dismissedby the Ninth Circuit U.S. Court ofAppeals, where Warren E. Burger arg-ued the Forest Service’s case (p. 151).

Since Mann Gulch, there have been nodeaths by burning among Forest Servicefirefighters, and people are nowequipped with backup radios (p. 219),better physical conditioning, the tactic ofbuilding an escape fire, knowledge thatfires in timber west of the ContinentalDivide burn differently than do fires ingrass east of the Divide, and the insist-ence that crew safety take precedenceover fire suppression.

The Methodology

Among the sources of evidence Macleanused to construct this case study wereinterviews, trace records, archivalrecords, direct observation, personalexperience, and mathematical models.

Since Maclean did not begin to gatherdocuments on Mann Gulch until 1976(p. 156) and did not start to work inearnest on this project until his seventy-fourth birthday in 1977, the lapse ofalmost 28 years since the disaster madeinterviewing difficult, especially sinceDodge had died of Hodgkin’s disease

five years after the fire (p. 106). Macleanlocated and interviewed both living wit-nesses of the blaze, Sallee and Rumsey,and persuaded both to accompany himand Laird Robinson, a guide at thesmokejumper base, on a visit back tothe site on July 1, 1978. Maclean alsoknew Dodge’s wife and had talked to herinformally (p. 40). He attempted to inter-view relatives of some who lost theirlives but found them too distraught 27years later to be of much help (p. 154).He also attempted to interview (p. 239)a member of the Forest Service inquiryteam, A. J. Cramer who, in 1951, hadpersuaded Sallee, Rumsey, and rangerRobert Jansson to alter their testimonyabout the timing of key incidents.Cramer was the custodian of seven oreight watches that had been removedfrom victims (p. 233), only one of which(Harrison’s) was released and used asthe official time of the disaster (5:56p.m.). To this day it remains unclearwhy the Forest Service made such astrong effort to locate the disaster closerto 6:00 p.m. than to 5:30, which wassuggested by testimony from Jansson,who was near the river when the fireblew up, and from a recovered watchthat read 5:42. Maclean had continuingaccess to two Forest Service insiders,Bud Moore and Laird Robinson (p. 162).He also interviewed experts on prece-dents for the escape fire (p. 104) andon the nature of death by fire (p. 213).

The use of trace records, or physicalevidence of past behaviors, is illustratedby the location during a 1979 trip to thegulch, of the wooden cross that hadbeen placed in 1949 to mark the spotwhere Dodge lit his escape fire (p. 206).The year before, 1978, during the tripinto the gulch with Sallee and Rumsey,Maclean located the rusty can ofpotatoes that had been discarded afterHellman drank its salty water throughtwo knife slits Rumsey had made in thecan (p. 173). He also located the flatrocks on which Hellman and Sylvia hadrested while awaiting rescue, the junipertree that was just beyond the creviceSallee and Rumsey squeezed throughon the ridge (p. 207), and Henry Thol,Jr.’s flashlight (p. 183). Considering the

lapse of time, the destructive forces ofnature over 28 years, and the power ofa blowup fire to melt and displace every-thing in its path, discovery of thesetraces is surprising as well as helpful inreconstructing events.

Archival records are crucial to the devel-opment of the case, although the ForestService made a considerable effort afterits inquiry to scatter the documents (p.153) and to classify most of them “Confi-dential” (p. 158), perhaps fearing it wouldbe charged with negligence. Recordsused by Maclean included statisticalreports of fire suppression by smoke-jumpers in Forest Service Region 1(e.g., p. 24); the report of the ForestService Board of Review issued shortlyafter the incident (dated September 29,1949, which many felt was too soon forthe board to do an adequate job); state-ments made to the board by people suchas the C-47 pilot, parents of the deadcrew (p. 150), and the spotter on theaircraft (p. 42); court reports of litigationbrought by parents of smokejumpersagainst the Forest Service; photographs,virtually all of which were retrieved forhim by women in the Forest Service whowere eager to help him tell the story (p.160); early records of the smokejumpersorganization, which was nine years oldat the time of the disaster; reports of the1957 task force on crew safety (p. 221);and contemporary reports of the disasterin the media, such as the report in theAugust 22, 1949 issue of Life magazine.

Direct observation occurred duringMaclean’s three visits to Mann Gulch in1976, 1977, and 1978 (p. 189), tripsmade much more difficult because ofthe inaccessibility of the area (pp. 191-192). The most important of these threevisits is the trip to the gulch with Salleeand Rumsey, during which the latter pairreenacted what they did and what theysaw intermittently through the densesmoke. When their accounts werematched against subsequent hard data(e.g., their estimation of where Dodgelit his escape fire compared againstdiscovery of the actual cross planted in1949 to mark the spot), it was found thattheir reconstruction of events prior to

61

Part 4 of 4the time they made it to safety throughthe crevice is less accurate than theirmemory for events and locations afterthey made it to safety. This suggests toMaclean that “we don’t remember asexactly the desperate moments whenour lives are in the balance as weremember the moments after, when thebalance has tipped in our favor” (p. 212).Direct observation also occurred whenMaclean and Robinson themselveshiked the steep slopes of Mann Gulchunder summer conditions of heat andslippery, tall grass that resembled theconditions present in the disaster of1949. The two men repeatedly comparedphotos and maps from 1949 with physi-cal outcroppings in front of them to seemore clearly what they were looking at(e.g., photos misrepresent the steep-ness of the slope, p. 175). There werealso informal experiments, as when RodNorum, an athlete and specialist on firebehavior, retraced Dodge’s route fromthe point at which he rejoined the crew,moved as fast as possible over the routeDodge covered, and was unable toreach the grave markers as fast as thecrew did (p. 67). During these trips,Maclean took special note of prevailingwinds by observing their effect on thedirection in which rotted timber fell.These observations were used to builda theory of how wind currents in thegulch could have produced the blowup(p. 133).

Personal experience was part of thecase because, in 1949, Maclean hadvisited the Mann Gulch fire while it wasstill burning (p. 1). Maclean also was aForest Service firefighter (not a smoke-jumper) at age 15 and nearly lost his lifein the Fish Creek fire, a fire much likethe one in Mann Gulch (p. 4). Macleanalso reports using his practical experi-ence as a woodsman to suggest initialhypotheses regarding what happenedat Mann Gulch (e.g., he infers windpatterns in the gulch from observationsof unusual wave action in the adjacentMissouri River, p. 131).

Having collected data using the abovesources, but still feeling gaps in hisunderstanding of precisely how the race

between fire and men unfolded, Macleantaught himself mathematics and turnedto mathematical modeling. He workedwith two mathematicians, Frank Albiniand Richard Rothermel, who had builtmathematical models of how firesspread. The group ran the predictivemodels in reverse to see what the firein Mann Gulch must have been like togenerate the reports on its progress thatwere found in interviews, reports, andactual measurements. It is the combina-tion of output from the model andsubjective reports that provide therevealing time line of the final 16minutes (pp. 267-277).

If these several sources of evidence arecombined and assessed for theadequacy with which they address“sources of invalidity,” it will be foundthat they combat 12 of the 15 sourceslisted by Runkel and McGrath (1972:191) and are only “moderatelyvulnerable” to the other three. Of course,an experienced woodsman andstoryteller who has “always tried to beaccurate with facts” (p. 259) wouldexpect that. The rest of us in organiza-tional studies may be pardoned,however, if we find those numbers agood reason to take these data seriously.

Cosmology Episodes inMann Gulch

Early in the book (p. 65), Maclean asksthe question on which I want to focus:“what the structure of a small outfitshould be when its business is to meetsudden danger and prevent disaster.”This question is timely because the workof organizations is increasingly done insmall temporary outfits in which thestakes are high and where foul-ups canhave serious consequences (Heyde-brand, 1989; Ancona and Caldwell,1992). Thus, if we understand whathappened at Mann Gulch, we may beable to learn some valuable lessons inhow to conceptualize and cope withcontemporary organizations.

Let me first be clear about why I think thecrew of smokejumpers at Mann Gulchwas an organization. First, they have aseries of interlocking routines, which iscrucial in Westley’s (1990: 339) definitionof an organization as “a series of inter-locking routines, habituated actionpatterns that bring the same peopletogether around the same activities inthe same time and places.” The crew atMann Gulch have routine, habituatedaction patterns, they come together froma common pool of people, and whilethis set of individual smokejumpers hadnot come together at the same places ortimes, they did come together aroundthe same episodes of fire. Westley’sdefinition suggests it doesn’t take muchto qualify as an organization. The otherside is, it also may not take much to stopbeing one.

Second, the Mann Gulch crew fits thefive criteria for a simple organizationalstructure proposed by Mintzberg (1983:158). These five include coordination bydirect supervision, strategy planned atthe top, little formalized behavior, organicstructure, and the person in chargetending to formulate plans intuitively,meaning that the plans are generally adirect “extension of his own personality.”Structures like this are found most oftenin entrepreneurial firms.

And third, the Mann Gulch crew has“generic subjectivity” (Wiley, 1988),meaning that roles and rules exist thatenable individuals to be interchangedwith little disruption to the ongoingpattern of interaction. In the crew atMann Gulch there were at least threeroles: leader, second in command, andcrewmember. The person in the leadsizes up the situation, makes decisions,yells orders, picks trails, sets the pace,and identifies escape routes (pp. 65-66).The second in command brings up therear of the crew as it hikes, repeatsorders, sees that the orders are under-stood, helps the individuals coordinatetheir actions, and tends to be closer tothe crew and more of a buddy with themthan does the leader. And finally, thecrew clears a fire line around the fire,cleans up after the fire, and maintains

62

Part 4 of 4trails. Thus, the crew at Mann Gulch is anorganization by virtue of a role structureof interlocking routines.

I want to argue that the tragedy at MannGulch alerts us to an unsuspectedsource of vulnerability in organizations.Minimal organizations, such as we findin the crew at Mann Gulch, are suscep-tible to sudden losses of meaning, whichhave been variously described as funda-mental surprises (Reason, 1990) orevents that are inconceivable (Lanir,1989), hidden (Westrum, 1982), orincomprehensible (Perrow, 1984). Eachof these labels points to the low proba-bility that the event could occur, which iswhy it is meaningless. But these explan-ations say less about the astonishmentof the perceiver and even less aboutthe perceiver’s inability to rebuild somesense of what is happening.

To shift the analytic focus in implausibleevents from probabilities to feelings andsocial construction, l have borrowed theterm “cosmology” from philosophy andstretched it. Cosmology refers to abranch of philosophy often subsumedunder metaphysics that combinesrational speculation and scientific evi-dence to understand the universe as atotality of phenomena. Cosmology is theultimate macro perspective, directed atissues of time, space, change, andcontingency as they relate to the originand structure of the universe. Integra-tions of these issues, however, are notjust the handiwork of philosophers.Others also make their peace with theseissues, as reflected in what they take forgranted. People, including those who aresmokejumpers, act as if events coherein time and space and that changeunfolds in an orderly manner. Theseeveryday cosmologies are subject todisruption. And when they are severelydisrupted, l call this a cosmology episode(Weick, 1985: 51-52). A cosmologyepisode occurs when people suddenlyand deeply feel that the universe is nolonger a rational, orderly system. Whatmakes such an episode so shattering isthat both the sense of what is occurringand the means to rebuild that sensecollapse together.

Stated more informally, a cosmologyepisode feels like vu jàdé—the oppositeof déjà vu: I’ve never been here before,l have no idea where I am, and I have noidea who can help me. This is what thesmokejumpers may have felt increas-ingly as the afternoon wore on and theylost what little organization structure theyhad to start with. As they lost structurethey became more anxious and found itharder to make sense of what washappening, until they finally were unableto make any sense whatsoever of theone thing that would have saved theirlives, an escape fire. The disaster atMann Gulch was produced by the inter-related collapse of sensemaking andstructure. If we can understand thiscollapse, we may be able to forestallsimilar disasters in other organizations.

Sensemaking in MannGulch

Although most organizational analysesbegin and end with decision making,there is growing dissatisfaction with thisorthodoxy. Reed (1991) showed howfar the concept of decision making hasbeen stretched, singling out the patchingthat James G. March has done in recentdiscussions of decision making. March(1989: 14) wrote that “decision makingis a highly contextual, sacred activity,surrounded by myth and ritual, and asmuch concerned with the interpretiveorder as with the specifics of particularchoices.” Reed (1991: 561) summarizedMarch this way: “decision making prefer-ences are often inconsistent, unstable,and externally driven; the linkagesbetween decisions and actions areloosely-coupled and interactive ratherthan linear; the past is notoriously unre-liable as a guide to the present or thefuture; and…political and symbolicconsiderations play a central, perhapsoverriding, role in decision making.”Reed wondered aloud whether, if Marchis right in these descriptions, decisionmaking should continue to set theagenda for organizational studies. Atsome point a retreat from classicprinciples becomes a rout.

There have been at least three distinctresponses to these problems. First, therehas been a shift, reminiscent of Neisserand Winograd’s (1988) work on memory,toward examining naturalistic decisionmaking (Orasanu and Connolly, 1993),with more attention to situationalassessment and sensemaking (Klein,1993). Second, people have replacedan interest in decision making with aninterest in power, noting, for example,that “power is most strategically deployedin the design and implementation ofparadigmatic frameworks within whichthe very meaning of such actions as‘making decisions’ is defined” (Brown,1978: 376). And third, people are replac-ing the less appropriate normativemodels of rationality (e.g., Hirsch,Michaels, and Friedman, 1987) basedon asocial “economic man” (Beach andLipshitz, 1993) with more appropriatemodels of rationality that are moresophisticated about social relations, suchas the model of contextual rationality(White, 1988).

Reed (1991) described contextualrationality as action motivated to createand maintain institutions and traditionsthat express some conception of rightbehavior and a good life with others.Contextual rationality is sensitive to thefact that social actors need to createand maintain intersubjectively bindingnormative structures that sustain andenrich their relationships. Thus, organi-zations become important because theycan provide meaning and order in theface of environments that impose ill-defined, contradictory demands.

One way to shift the focus from decisionmaking to meaning is to look moreclosely at sensemaking in organizations.The basic idea of sensemaking is thatreality is an ongoing accomplishmentthat emerges from efforts to create orderand make retrospective sense of whatoccurs. Recognition-primed decisionmaking, a model based in part on com-mand decisions made by firefighters, hasfeatures of sensemaking in its relianceon past experience, although it remainsgrounded in decision making (Klein,1993). Sensemaking emphasizes that

63

Part 4 of 4people try to make things rationallyaccountable to themselves and others.Thus, in the words of Morgan, Frost, andPondy (1983: 24), “individuals are notseen as living in, and acting out theirlives in relation to, a wider reality, somuch as creating and sustaining imagesof a wider reality, in part to rationalizewhat they are doing. They realize theirreality, by reading into their situationpatterns of significant meaning.”

When the smokejumpers landed atMann Gulch, they expected to find whatthey had come to call a 10:00 fire. A10:00 fire is one that can be surroundedcompletely and isolated by 10:00 thenext morning. The spotters on theaircraft that carried the smokejumpers“figured the crew would have it undercontrol by 10:00 the next morning”(Maclean, p. 43). People rationalizedthis image until it was too late. Andbecause they did, less and less of whatthey saw made sense:

1. The crew expects a 10:00 fire butgrows uneasy when this fire does notact like one.

2. Crewmembers wonder how this firecan be all that serious if Dodge andHarrison eat supper while they hiketoward the river.

3. People are often unclear who is incharge of the crew (p. 65).

4. The flames on the south side of thegulch look intense, yet one of thesmokejumpers, David Navon is takingpictures, so people conclude the firecan’t be that serious, even thoughtheir senses tell them otherwise.

5. Crewmembers know they are movingtoward the river where they will besafe from the fire, only to see Dodgeinexplicably turn them around, awayfrom the river, and start anglingupslope, but not running straight forthe top. Why? (Dodge is the only onewho sees the fire jump the gulchahead of them.)

6. As the fire gains on them, Dodge says,“Drop your tools,” but if the people inthe crew do that, then who are they?Firefighters? With no tools?

7. The foreman lights a fire that seemsto be right in the middle of the onlyescape route people can see.

8. The foreman points to the fire he hasstarted and yells, “Join me,” whateverthat means. But his second in com-mand sounds like he’s saying, “Tohell with that, I’m getting out of here”(p. 95).

9. Each individual faces the dilemma, lmust be my own boss yet follow ordersunhesitatingly, but I can’t comprehendwhat the orders mean, and I’m losingmy race with the advancing fire (pp.219-220).

As Mann Gulch loses its resemblance toa 10:00 fire, it does so in ways that makeit increasingly hard to socially constructreality. When the noise created by wind,flames, and exploding trees is deafening;when people are strung out in a line andrelative strangers to begin with; whenthey are people who, in Maclean’s words,“love the universe but are not intimidatedby it” (p. 28); and when the temperatureis approaching a lethal 140 degrees (p.220), people can neither validate theirimpressions with a trusted neighbor norpay close attention to a boss who is alsounknown and whose commands makeno sense whatsoever. As if these werenot obstacles enough, it is hard to makecommon sense when each person seessomething different or nothing at allbecause of the smoke.

The crew’s stubborn belief that it faceda 10:00 fire is a powerful reminder thatpositive illusions (Taylor, 1989) can killpeople. But the more general point isthat organizations can be good atdecision making and still falter. Theyfalter because of deficient sensemaking.The world of decision making is aboutstrategic rationality. It is built from clearquestions and clear answers thatattempt to remove ignorance (Daft andMacintosh, 1981). The world of sense-

making is different. Sensemaking isabout contextual rationality. It is built outof vague questions, muddy answers,and negotiated agreements thatattempt to reduce confusion. People inMann Gulch did not face questions likewhere should we go, when do we take astand, or what should our strategy be?Instead, they faced the more basic, themore frightening feeling that their oldlabels were no longer working. Theywere outstripping their past experienceand were not sure either what was up orwho they were. Until they develop somesense of issues like this, there is nothingto decide.

Role Structure in MannGulch

Sensemaking was not the only problemin Mann Gulch. There were alsoproblems of structure. It seems plausibleto argue that a major contributor to thisdisaster was the loss of the only struc-ture that kept these people organized,their role system. There were two keyevents that destroyed the organizationthat tied these people together. First,when Dodge told Hellman to take thecrew to the north side of the gulch andhave it follow a contour down toward theriver, the crew got confused, the spacesbetween members widened appreciably,and Navon—the person taking pictures(p. 71)— made a bid to take over theleadership of the group (p. 65). Noticewhat this does to the role system. Thereis now no one at the end of the linerepeating orders as a check on theaccuracy with which they are under-stood. Furthermore, the person who isleading them, Hellman, is more familiarwith implementing orders than withconstructing them or plotting possibleescape routes. So the crew is left for acrucial period of time with ill-structured,unacknowledged orders shouted bysomeone who is unaccustomed to beingfirm or noticing escape routes. Bothroutines and interlocking are beginningto come apart. The second, and in someway more unsettling threat to the role

64

Part 4 of 4system occurred when Dodge told theretreating crew “throw away your tools!”(p. 226). A fire crew that retreats from afire should find its identity and moralestrained. If the retreating people are thenalso told to discard the very things thatare their reason for being there in thefirst place, then the moment quickly turnsexistential. If I am no longer a firefighter,then who am l? With the fire bearingdown, the only possible answerbecomes, An endangered person in aworld where it is every man for himself.Thus, people who, in Maclean’s words,had perpetually been almost their ownboss (p. 218) suddenly became com-pletely their own boss at the worstpossible moment. As the entity of a crewdissolved, it is not surprising that the finalcommand from the “crew” leader to jumpinto an escape fire was heard not as alegitimate order but as the ravings ofsomeone who had “gone nuts” (p. 75).Dodge’s command lost its basis of legiti-macy when the smokejumpers threwaway their organization along with theirtools.

Panic In Mann Gulch

With these observations as background,we can now look more closely at theprocess of a cosmology episode, aninterlude in which the orderliness of theuniverse is called into question becauseboth understanding and procedures forsensemaking collapse together. Peoplestop thinking and panic. What is inter-esting about this collapse is that it wasdiscussed by Freud (1959: 28) in thecontext of panic in military groups: “Apanic arises if a group of that kind[military group] becomes disintegrated.Its characteristics are that none of theorders given by superiors are any longerlistened to, and that each individual isonly solicitous on his own account, andwithout any consideration for the rest.The mutual ties have ceased to exist,and a gigantic and senseless fear is setfree.” Unlike earlier formulations, such

as McDougall’s (1920), which hadargued that panic leads to group disinte-gration, Freud, reversing this causality,argued that group disintegration precip-itates panic. By group disintegration,Freud meant “the cessation of all thefeelings of consideration which themembers of the group otherwise showone another” (p. 29). He described themechanism involved this way: “If anindividual in panic fear begins to besolicitous only on his own account, hebears witness in so doing to the fact thatthe emotional ties, which have hithertomade the danger seem small to him,have ceased to exist. Now that he is byhimself in facing the danger, he maysurely think it greater.”

It is certainly true in Mann Gulch thatthere is a real, palpable danger that canbe seen, felt, heard, and smelled by thesmokejumpers. But this is not the firsttime they have confronted danger. Itmay, however, be the first time theyhave confronted danger as a member ofa disintegrating organization. As the crewmoved toward the river and becamemore spread out, individuals wereisolated and left without explanations oremotional support for their reactions. Asthe ties weakened, the sense of dangerincreased, and the means to cope be-came more primitive. The world rapidlyshifted from a cosmos to chaos as itbecame emptied of order and rationality.

It is intriguing that the three people whosurvived the disaster did so in ways thatseem to forestall group disintegration.Sallee and Rumsey stuck together,their small group of two people did notdisintegrate, which helped them keeptheir fear under control. As a result, theyescaped through a crack in the ridge thatthe others either didn’t see or thoughtwas too small to squeeze through. WagDodge, as the formal leader of a grouphe presumed still existed, ordered hisfollowers to join him in the escape fire.Dodge continued to see a group and tothink about its well-being, which helpedkeep his own fear under control. The restof the people, however, took less noticeof one another. Consequently, the group,

as they knew it, disintegrated. As theirgroup disintegrated, the smokejumpersbecame more frightened, stoppedthinking sooner, pulled apart even more,and in doing so, lost a leader-followerrelationship as well as access to thenovel ideas of other people who are a lotlike them. As these relationships disap-peared, individuals reverted to primitivetendencies of flight. Unfortunately, thisresponse was too simple to match thecomplexity of the Mann Gulch fire.

What holds organization in place maybe more tenuous than we realize. Therecipe for disorganization in Mann Gulchis not all that rare in everyday life. Therecipe reads, Thrust people intounfamiliar roles, leave some key rolesunfilled, make the task more ambiguous,discredit the role system, and make allof these changes in a context in whichsmall events can combine into somethingmonstrous. Faced with similar conditions,organizations that seem much sturdiermay also come crashing down (Miller,1990; Miles and Snow, 1992), much likeIcarus who overreached his competenceas he flew toward the sun and alsoperished because of fire.

From Vulnerability toResilience

The steady erosion of sense and struc-ture reached its climax in the refusal ofthe crew to escape one fire by walkinginto another one that was intentionallyset. A closer look at that escape fireallows us to move from a discussion ofwhat went wrong at Mann Gulch, to adiscussion of what makes organizationsmore resilient. l want to discuss foursources of resilience: (1) improvisationand bricolage, (2) virtual role systems,(3) the attitude of wisdom, and (4)respectful interaction.

65

Part 4 of 4

Improvisation andBricolage

The escape fire is a good place to startin the search for sources of resiliencesimply because it is clear evidence that,minimal though the organization of thecrew might have been, there still was asolution to the crisis inside the group.The problem was, no one but Dodgerecognized this. The question thenbecomes, How could more people eithersee this escape fire as a solution ordevelop their own solution? This is notan easy question to answer because,from everything we know, Dodge’sinvention of burning a hole in a fireshould not have happened. It should nothave happened because there is goodevidence that when people are put underpressure, they regress to their mosthabituated ways of responding (e.g.,Barthol and Ku, 1959). This is what wesee in the 15 people who reject Dodge’sorder to join him and who resort insteadto flight, a more overlearned tendency.What we do not expect under life-threatening pressure is creativity.

The tactic of lighting a fire to create anarea where people can escape a majorprairie fire is mentioned in James Feni-more Cooper’s 1827 novel The Prairie ,but there is no evidence Dodge knewthis source (Maclean, p. 104). Further-more, most of Dodge’s experience hadbeen in timbered country where such atactic wouldn’t work. In timber, anescape fire is too slow and consumestoo much oxygen (p. 105). And the firethat Dodge built did not burn longenough to clear an area in which peoplecould move around and dodge the fireas they did in the prairie fire. There wasjust room enough to lie down in theashes where the heat was less intense(p. 104).

While no one can say how or why theescape fire was created, there is a lineof argument that is consistent with whatwe know. Bruner (1983: 183) describedcreativity as “figuring out how to usewhat you already know in order to go

beyond what you currently think.” Withthis as background, it now becomesrelevant that Dodge was an experiencedwoodsman, with lots of hands-on exper-ience. He was what we now would calla bricoleur, someone able to createorder out of whatever materials were athand (e.g., Levi-Strauss, 1966; Harper,1987). Dodge would have known atleast two things about fires. He wouldhave known the famous fire triangle—you must have oxygen, flammablematerial, and temperature above thepoint of ignition to create a fire (Maclean,p. 35). A shortage of any one of thesewould prevent a fire. In his case, theescape fire removed flammable material.And since Dodge had been with theForest Service longer than anyone elseon the crew, he would also have knownmore fully their four guidelines at thattime for dealing with fire emergencies(p. 100). These included (1) start abackfire if you can, (2) get to the top ofa ridge where the fuel is thinner, (3) turninto the fire and try to work through it,and (4) don’t allow the fire to pick thespot where it hits you. Dodge’s invention,if we stretch a bit, fits all four. It is abackfire, though not in the conventionalsense of a fire built to stop a fire. Theescape fire is lit near the top of a ridge,Dodge turns into the main fire and worksthrough it by burning a hole in it, and hechooses where the fire hits him. The 15who tried to outrun the fire moved towardthe ridge but by not facing the fire, theyallowed it to pick the spot where it hitthem.

The collapse of role systems need notresult in disaster if people develop skillsin improvisation and bricolage (seeJanowitz, 1959: 481). Bricoleurs remaincreative under pressure, preciselybecause they routinely act in chaoticconditions and pull order out of them.Thus, when situations unravel, this issimply normal, natural trouble for brico-leurs, and they proceed with whatevermaterials are at hand. Knowing thesematerials intimately, they then are able,usually in the company of other similarlyskilled people, to form the materials orinsights into novel combinations.

While improvised fire fighting may soundimprobable, in fact, Park Service fire-fighters like those stationed at the GrandCanyon approximate just such a style.Stephen Pyne (1989), a Park Servicefirefighter, observed that people like himtypically have discretion to dispatchthemselves, which is unfathomable tothe Forest Service crews that rely ondispatchers, specialization, regimenta-tion, rules, and a conscious preferencefor the strength of the whole rather thanthe versatility and resourcefulness of theparts. Forest Service people marvel atthe freedom of movement among thePark people. Park Service people marvelat how much power the Forest Serviceis able to mobilize on a fire. Pyne (1989:122) described the Park Service fireoperations as a nonstandard “eclecticassembly of compromises” built ofdiscretion and mobility. In contrast to theForest Service, where people do every-thing by the book, “The Park Servicehas no books; it puts a premium on theindividual. Its collective behavior is tribal,and it protects its permanent ranks.” Ifimprovisation were given more attentionin the job description of a crew person,that person’s receptiveness to andgeneration of role improvisations mightbe enhanced. As a result, when oneorganizational order collapses, a substi-tute might be invented immediately. Swiftreplacement of a traditional order withan improvised order would forestall theparalysis that can follow a command to“drop your tools.”

Virtual Role Systems

Social construction of reality is next toimpossible amidst the chaos of a fire,unless social construction takes placeinside one person’s head, where the rolesystem is reconstituted and run. Eventhough the role system at Mann Gulchcollapsed, this kind of collapse need notresult in disaster if the system remainsintact in the individual’s mind. If eachindividual in the crew mentally takes all

66

Part 4 of 4roles and therefore can then registerescape routes and acknowledgecommands and facilitate coordination,then each person literally becomes agroup (Schutz, 1961). And, in the mannerof a holograph, each person can recon-stitute the group and assume whateverrole is vacated, pick up the activities, andrun a credible version of the role.Furthermore, people can run the groupin their head and use it for continuedguidance of their own individual action.

It makes just as much sense to talkabout a virtual role system as it does totalk about a virtual anything else (e.g.,Bruner, 1986: 36-37). An organizationcan continue to function in the imagina-tion long after it has ceased to functionin tangible distributed activities. For theMann Gulch fire, this issue has bearingon the question of escape routes. In ourresearch on accidents in flight operationsoff nuclear carriers (Weick and Roberts,1993), Karlene Roberts and I found thatpeople who avoid accidents live by thecredo, “never get into anything withoutmaking sure you have a way out.” At thevery last moment in the Mann Gulchtragedy, Dodge discovered a way out.The point is that if other people had beenable to simulate Dodge and/or his rolein their imagination, they too might havebeen less puzzled by his solution orbetter able to invent a different sensiblesolution for themselves.

The Attitude of Wisdom

To understand the role of wisdom (Bige-low, 1992) as a source of resilience, weneed to return to the crew’s belief thatall fires are 10:00 fires. This belief wasconsistent with members’ experience.As Maclean put it, if the major purposeof your group is to “put out fires so fastthey don’t have time to become big ones”(p. 31), then you won’t learn much aboutfighting big fires. Nor will you learn whatMaclean calls the first principle of reality:“little things suddenly and literally can

become big as hell, the ordinary cansuddenly become monstrous, and theupgulch breezes can suddenly turn tomurder” (p. 217). To state the point moregenerally, what most organizations miss,and what explains why most organiza-tions fail to learn (Scott, 1987: 282), isthat “Reality backs up while it isapproached by the subject who tries tounderstand it. Ignorance and knowledgegrow together” (Meacham, 1983: 130).To put it a different way, “Each newdomain of knowledge appears simplefrom the distance of ignorance. The morewe learn about a particular domain, thegreater the number of uncertainties,doubts, questions and complexities.Each bit of knowledge serves as thethesis from which additional questions orantithesis arise” (Meacham, 1983: 120).

The role system best able to accept thereality that ignorance and knowledgegrow together may be one in which theorganizational culture values wisdom.Meacham (1983: 187) argued thatwisdom is an attitude rather than a skillor a body of information:

To be wise is not to know particular factsbut to know without excessiveconfidence or excessive cautiousness.Wisdom is thus not a belief, a value, aset of facts, a corpus of knowledge orinformation in some specialized area, ora set of special abilities or skills. Wisdomis an attitude taken by persons towardthe beliefs, values, knowledge, informa-tion, abilities, and skills that are held, atendency to doubt that these are neces-sarily true or valid and to doubt that theyare an exhaustive set of those thingsthat could be known.

In a fluid world, wise people know thatthey don’t fully understand what ishappening right now, because they havenever seen precisely this event before.Extreme confidence and extremecaution both can destroy what organiza-tions most need in changing times,namely, curiosity, openness, andcomplex sensing. The overconfidentshun curiosity because they feel theyknow most of what there is to know. The

overcautious shun curiosity for fear itwill only deepen their uncertainties.Both the cautious and the confident areclosed-minded, which means neithermakes good judgments. It is this sensein which wisdom, which avoids extremes,improves adaptability.

A good example of wisdom in groups isthe Naskapi Indians’ use of cariboushoulder bones to locate game (Weick,1979). They hold bones over a fire untilthey crack and then hunt in the direc-tions to which the cracks point. Thisritual is effective because the decisionis not influenced by the outcomes of pasthunts, which means the stock of animalsis not depleted. More important, the finaldecision is not influenced by the inevit-able patterning in human choice, whichenables hunted animals to becomesensitized to humans and take evasiveaction. The wisdom inherent in thispractice derives from its ambivalencetoward the past. Any attempt to hunt forcaribou is both a new experience andan old experience. It is new in the sensethat time has elapsed, the compositionof the hunter band has changed, thecaribou have learned new things, and soforth. But the hunt is also old in the sensethat if you’ve seen one hunt, you’ve seenthem all: There are always hunters,weapons, stealth, decoys, tracks, odors,and winds. The practice of divinationincorporates the attitude of wisdombecause past experience is discountedwhen a new set of cracks forms a crudemap for the hunt. But past experienceis also given some weight, because aseasoned hunter “reads” the cracks andinjects some of his own past experienceinto an interpretation of what the cracksmean. The reader is crucial. If thereader’s hunches dominate, randomiza-tion is lost. If the cracks dominate, thenthe experience base is discarded. Thecracks are a lot like the four guidelinesfor fire emergencies that Dodge mayhave relied on when he invented theescape fire. They embody experience,but they invite doubt, reassembly, andshaping to fit novelties in the present.

67

Part 4 of 4

Respectful InteractionThe final suggestion about how tocounteract vulnerability makes explicitthe preceding focus on the individual andsocial interaction. Respectful interactiondepends on intersubjectivity (Wiley,1988: 258), which has two definingcharacteristics: (1) intersubjectivityemerges from the interchange andsynthesis of meanings among two ormore communicating selves, and (2) theself or subject gets transformed duringinteraction such that a joint or mergedsubjectivity develops. It is possible thatmany role systems do not change fastenough to keep up with a rapidlychanging environment. The only formthat can keep up is one based on face-to-face interaction. And it is here, ratherthan in routines, that we are best able tosee the core of organizing. This may bewhy interaction in airline cockpit crews,such as discussed by Foushee (1984),strikes us so often as a plausible micro-cosm of what happens in much largersystems. In a cockpit under crisis, theonly unit that makes sense (pun intend-ed) is face-to-face synthesis of meaning.

Intersubjectivity was lost on everyone atMann Gulch, everyone, that is, but Salleeand Rumsey. They stuck together andlived. Dodge went his own individual waywith a burst of improvisation, and he toolived. Perhaps it’s more important thatyou have a partner than an organizationwhen you fight fires. A partner makessocial construction easier. A partner isa second source of ideas. A partnerstrengthens independent judgment in theface of a majority. And a partner enlargesthe pool of data that are considered.Partnerships that endure are likely to bethose that adhere to Campbell’s threeimperatives for social life, based on areanalysis of Asch’s (1952) conformityexperiment: (1) Respect the reports ofothers and be willing to base beliefsand actions on them (trust); (2) Reporthonestly so that others may use yourobservations in coming to valid beliefs(honesty); and, (3) Respect your ownperceptions and beliefs and seek tointegrate them with the reports of others

without deprecating them or yourselves(self-respect) (adapted from Campbell,1990: 45-46).

Earlier I noted a growing interest incontextual rationality, understood asactions that create and maintain institu-tions and traditions that express someconception of right behavior and a goodlife with others (Reed, 1991). Campbell’smaxims operationalize this good life withothers as trust, honesty, and self-respectin moment-to-moment interaction. Thistriangle of trust, honesty, and self-respectis conspicuously missing (e.g., King,1989: 46-48) in several well-documenteddisasters in which faulty interactionprocesses led to increased fear, dimin-ished communication, and death. Forexample, in the Tenerife air disaster(Weick, 1990), the copilot of the KLMaircraft had a strong hunch that another747 airplane was on the takeoff runwaydirectly in front of them when his owncaptain began takeoff without clearance.But the copilot said nothing about eitherthe suspicions or the illegal departure.Transient cockpit crews, tied together bynarrow definitions of formal responsibili-ties, and headed by captains whomistakenly assume that their decision-making ability is unaffected by increasesin stress (Helmreich et al., 1985), havefew protections against a sudden lossof meaning such as the preposterouspossibility that a captain is taking offwithout clearance, directly into the pathof another 747.

Even when people try to act withhonesty, trust, and self respect, if theydo so with little social support, theirefforts are compromised. For example,linguists who analyzed the conversationsat Tenerife and in the crash of Air Floridaflight 90 in Washington concluded thatthe copilots in both cases used “devicesof mitigation” to soften the effects of theirrequests and suggestions:

A mitigated instruction might be phrasedas a question or hedged with qualifica-tions such as “would” or “could.” …(I)twas found that the speech of subordi-nate crew members was much more

likely to be mitigated than the speech ofcaptains. It was also found that topicsintroduced in mitigated speech were lesslikely to be followed-up by other crewmembers and less likely to be ratified bythe captain. Both of these effects relatedirectly to the situation in which a subor-dinate crew member makes a correctsolution that is ignored… The value oftraining in unmitigated speech is stronglysuggested by these results. (O’Hare andRoscoe, 1990: 219)

If a role system collapses among peoplefor whom trust, honesty, and self-respectare underdeveloped, then they are ontheir own. And fear often swamps theirresourcefulness. If, however, a rolesystem collapses among people wheretrust, honesty, and self-respect are morefully developed, then new options, suchas mutual adaptation, blind imitation ofcreative solutions, and trusting compli-ance, are created. When a formalstructure collapses, there is no leader,no roles, no routines, no sense. That iswhat we may be seeing in Mann Gulch.Dodge can’t lead because the rolesystem in which he is a leader disap-pears. But what is worse, Dodge can’trely on his crew members to trust him,question him, or pay attention to him,because they don’t know him and thereis no time to change this. The key ques-tion is, When formal structure collapses,what, if anything, is left? The answer tothat question may well be one of life ordeath.

Structures ForResilience

While the answer to that question is nota matter of life or death for organiza-tional theorists, they do have an interestin how it comes out. A theorist who hearsMaclean’s question, “what the structureof a small outfit should be when itsbusiness is to meet sudden danger andprevent disaster,” might come back with

68

Part 4 of 4a series of follow-up questions based onthinking in organizational studies. l lookbriefly at four such questions to linkMann Gulch with other concepts and tosuggest how these linkages might guidefurther research.

First, there is the follow-up question, Is“small” necessarily a key dimension,since this group is also young andtransient? Maclean calls the 16-personsmokejumper crew “small,” except thatit is conventional in the group literatureto treat any group of more than 10people as large (Bass, 1990: 604).Because there is so little communicationwithin the crew and because it operateslargely through obtrusive controls likerules and supervision (Perrow, 1986),it acts more like a large formal groupwith mediated communication than asmall informal group with direct commu-nication.

It is striking how little communicationoccurred during the three and a halfhours of this episode. There was littlediscussion during the noisy, bumpyplane ride, and even less as individualsretrieved equipment scattered on thenorth slope. After a quick meal together,people began hiking toward the river butquickly got separated from one another.Then they were suddenly turned around,told to run for the ridge, and quickly ranout of breath as they scaled the steepsouth slope. The minimal communicationis potentially important because of thegrowing evidence (e.g., Eisenhardt,1993: 132) that nonstop talk, both vocaland nonverbal, is a crucial source ofcoordination in complex systems that aresusceptible to catastrophic disasters.

The lack of communication, coupled withthe fact that this is a temporary group inthe early stages of its history, shouldheighten the group’s vulnerability todisruption. As Bass (1990: 637) put it,“Groups that are unable to interact easilyor that do not have the formal or informalstructure that enables quick reactionsare likely to experience stress (Bass,1960). Panic ensues when members ofa group lack superordinate goals—goalsthat transcend the self-interests of each

participant.” While the smokejumpershave the obvious superordinate goal ofcontaining fires, their group ties may notbe sufficiently developed for this to be agroup goal that overrides self-interest. OrBass’s proposition itself may be incom-plete, failing to acknowledge that unlesssuperordinate goals are overlearned,they will be discarded in situations ofdanger.

Second, there is the follow-up question,Is “structure” what we need to under-stand in Mann Gulch, or might structuringalso be important? By structure, l mean“a complex medium of control which iscontinually produced and recreated ininteraction and yet shapes that interac-tion: structures are constituted andconstitutive…of interpersonal cognitiveprocesses, power dependencies, andcontextual constraints” (Ranson, Hinings,and Greenwood, 1980: 1, 3). Structuring,then, consists of two patterns and therelationships between them. The firstpattern, which Ranson et al. variouslydescribed as informal structure, agency,or social construction, consists of interac-tion patterns that stabilize meaning bycreating shared interpretive schemes. lrefer to this pattern as shared provincesof meaning, or meaning. The secondpattern, variously described as configur-ation, contextual constraints, or a vehiclethat embodies dominant meanings,refers to a framework of roles, rules,procedures, configured activities, andauthority relations that reflect and facili-tate meanings. l refer to this secondpattern as structural frameworks ofconstraint, or frameworks.

Meanings affect frameworks, whichaffect meaning. This is the basic point ofthe growing body of work on structura-tion (e.g., Riley, 1983; Poole, Seibold,and McPhee, 1985), understood as themutual constitution of frameworks andmeanings (Ranson, Hinings, and Green-wood, 1980) or relations and typifications(DiMaggio, 1991) or structures andstructuring (Barley, 1986). Missing inthis work is attention to reversals ofstructuration (Giddens, 1984). The useof descriptive words in structurationtheory such as “continually produced,”

“recreated in interaction,” “constituted,”and “constitutive” directs attention awayfrom losses of frameworks and losses ofmeaning. For example, Ranson, Hinings,and Greenwood (1980: 5) asserted thatthe “deep structure of schema which aretaken for granted by members enablesthem to recognize, interpret, and negoti-ate even strange and unanticipatedsituations, and thus continuously tocreate and reenact the sense andmeaning of structural forms during thecourse of interaction.” The Mann Gulchdisaster is a case in which people wereunable to negotiate strangeness.Frameworks and meanings destroyedrather than constructed one another.

This fugitive quality of meaning andframeworks in Mann Gulch suggests thatthe process of structuring itself may bemore unstable than we realized. Struc-turing, understood as constitutiverelations between meaning and frame-works, may be a deviation-amplifyingcause loop (Maruyama, 1963; Weick,1979) capable of intensifying either anincrease or decrease in either of the twoconnected elements. Typically, we seeinstances of increase in which moreshared meanings lead to less elaborateframeworks of roles, which lead to furtherdevelopments of shared meaning, etc.What we fail to realize is that, whenelements are tied together in this directmanner, once one of them declines, thisdecline can also spread and becomeamplified as it does so. Fewer sharedmeanings lead to less elaborate frame-works, less meaning, less elaborateframeworks, and so on. Processes thatmutually constitute also have the capabil-ity to mutually destroy one another.

If structuration is treated as a deviation-amplifying process, then this suggeststhe kind of structure that could haveprevented the Mann Gulch disaster.What people needed was a structure inwhich there was both an inverse and adirect relationship between role systemsand meaning. This is the only patternthat can maintain resilience in the faceof crisis. The resilience can take one oftwo forms. Assume that we start with anamplifying system like the one in Mann

69

Part 4 of 4Gulch. The role system lost its structure,which led to a loss of meaning, which ledto a further loss of structure, and so on.This is the pattern associated with adeviation-amplifying feedback loop inwhich an initial change unfoldsunchecked in the same direction. Oneway to prevent this amplification is toretain the direct relation between struc-ture and meaning (less role structureleads to less meaning, more structureleads to more meaning) but create aninverse relation between meaning andstructure (less meaning, more structure,and vice versa). This inverse relationshipcan be understood as follows: Whenmeaning becomes problematic anddecreases, this is a signal for people topay more attention to their formal andinformal social ties and to reaffirm and/orreconstruct them. These actions producemore structure, which then increasesmeaning, which then decreases theattention directed at structure. Puzzle-ment intensifies attentiveness to thesocial, which reduces puzzlement.

The other form of control arises when achange in structure, rather than a changein meaning, is responsible for counter-acting the fluctuations in sensibleness.In this variation, less structure leads tomore meaning, and more meaning thenproduces more structure. The inverserelationship between structure andmeaning can be understood this way:When social ties deteriorate, people tryharder to make their own individualsense of what is happening, both sociallyand in the world. These operationsincrease meaning, and they increase thetendency to reshape structure consistentwith heightened meaning. Alienationintensifies attentiveness to meaning,which reduces alienation.

What is common to both of thesecontrolled forms is an alternationbetween attention to frameworks andattention to meanings. More attention toone leads to more ignorance of theother, followed by efforts to correct thisimbalance, which then creates a newimbalance. In the first scenario, whenmeaning declines, people pay moreattention to frameworks, they ignore

meaning temporarily, and as socialrelations become clearer, their attentionshifts back to meanings. In the secondscenario, when social relations decline,people pay more attention to meaning,they ignore frameworks temporarily, andas meanings become clearer, attentionshifts back to frameworks. Both scen-arios illustrate operations of wisdom: InMeacham’s words, ignorance and know-ledge grow together. Either of these twocontrolled patterns should reduce thelikelihood of disaster in Mann Gulch. Asthe smokejumpers begin to lose struc-ture they either also lose meaning, whichalerts them to be more attentive to thestructure they are losing, or they gainindividual meaning, which leads them torealign structure. The second alternativemay be visible in the actions taken byDodge and Rumsey and Sallee.

This may seem like a great deal of fret-ting about one single word in Maclean’squestion, “structure.” What I have triedto show is that when we transform thisword from a static image into a process,we spot what looks like a potential forcollapse in any process of socialsensemaking that is tied together byconstitutive relations. And we find thatsocial sensemaking may be most stablewhen it is simultaneously constitutiveand destructive, when it is capable ofincreasing both ignorance and know-ledge at the same time. That seemslike a fair return for reflecting on asingle word.

Third, there is the follow-up question, Is“outfit” the best way to describe thesmokejumpers? An outfit is normallydefined as “a group associated in anundertaking requiring close cooperation,as a military unit” (Random House,1987: 1374). The smokejumpers are tiedtogether largely by pooled interdepend-ence, since the job of each one is toclear adjacent portions of a perimeterarea around a blaze so that the fire stopsfor lack of fuel. Individual efforts to clearaway debris are pooled and form a fireline. What is significant about pooledinterdependence is that it can functionwithout much cohesion (Bass, 1990:622). And this is what may have trapped

the crew. Given the constantly changingcomposition of the smokejumping crews,the task largely structured their relations.Simply acting in concert was enough,and there was no need to know eachother well in addition. This social formresembles what Eisenberg (1990: 160)called nondisclosive intimacy, by whichhe meant relationships rooted in collec-tive action that stress “coordination ofaction over the alignment of cognitions,mutual respect over agreement, trustover empathy, diversity over homoge-neity, loose over tight coupling, andstrategic communication over unrestric-ted candor.” Nondisclosive intimacy is asufficient ground for relating as long asthe task stays constant and the environ-ment remains stable.

What the Mann Gulch disaster suggestsis that nondisclosive intimacy may limitthe development of emotional ties thatkeep panic under control in the face ofobstacles. Closer ties permit clearerthinking, which enables people to findpaths around obstacles. For example,when Rumsey squeezed through acrevice in the ridge just ahead of the fire,he collapsed “half hysterically” into ajuniper bush, where he would have soonburned to death. His partner Salleestopped next to him, looked at him coldly,never said a word, and just stood thereuntil Rumsey roused himself, and thetwo then ran together over the ridge anddown to a rock slide where they werebetter able to move around and duck theworst flames (Maclean, p. 107). Sallee’ssurprisingly nuanced prodding of hispartner suggests the power of close tiesto moderate panic.

One might expect that the less threaten-ing the environment, the less importantare relational issues in transient groups,but as Perrow (1984) emphasized in hisnormal accident theory, there are fewsafe environments. If events are increas-ingly interdependent, then smallunrelated flaws can interact to producesomething monstrous. Maclean saw thisclearly at Mann Gulch: The colossal fireblowup in Mann Gulch was “shaped bylittle screwups that fitted together tighterand tighter until all became one and the

70

Part 4 of 4same thing—the fateful blowup. Such ismuch of tragedy in modern times andprobably always has been except thatpast tragedy refrained from speaking ofits association with screwups and blow-ups” (Maclean, 1992: 92).

Nondisclosive intimacy is not the onlyalternative to “outfit” as a way to describethe smokejumpers. Smith (1983) arguedthat individual behaviors, perceptions ofreality, identities, and acts of leadershipare influenced by intergroup processes.Of special relevance to Mann Gulch isSmith’s reanalysis of the many groupsthat formed among the 16 members ofthe Uruguayan soccer team whosurvived for 10 weeks in an inaccessibleregion of the Chilean Andes mountainsafter their aircraft, carrying 43 people,crashed (see Read, 1974 for the originalaccount of this event). Aside from theeerie coincidence that both disastersinvolved 16 young males, Smith’sanalysis makes the important point that16 people are not just an outfit, they area social system within which multiplegroups emerge and relate to oneanother. It is these intergroup relation-ships that determine what will be seenas acts of leadership and which peoplemay be capable of supplying those acts.In the Andes crash, demands shiftedfrom caring for the wounded, in whichtwo medical students took the lead, toacquiring food and water, where the teamcaptain became leader, to articulatingthat the group would not be rescuedand could sustain life only if peopleconsumed the flesh of the dead, to exe-cuting and resymbolizing this survivaltactic, to selecting and equipping anexpeditionary group to hike out and lookfor help, and finally to finding someoneable to explain and rationalize theirdecisions to the world once they hadbeen rescued.

What Smith shows is that this group of16 forms and reforms in many differentdirections during its history, each timewith a different coherent structure ofpeople at the top, middle, and bottom,each with different roles. What alsobecomes clear is that any attempt topinpoint the leader or to explain survival

by looking at a single set of actions isdoomed to failure because it does notreflect how needs change as a crisisunfolds, nor does it reflect how differentcoherent groupings form to meet thenew needs.

The team in the Andes had 10 weeksand changing threats of bleeding,hygiene, starvation, avalanche, expedi-tion, rescue, and accounting, whereasthe team in Mann Gulch had more like10 minutes and the increasingly singularthreat of being engulfed in fire. Part ofthe problem in Mann Gulch is the veryinability for intergroup structures to form.The inability to form subgroups withinthe system may be due to such things astime pressure, the relative unfamiliarityof the smokejumpers with one anothercompared with the interdependentmembers of a visible sports team, theinability to communicate, the articulationof a common threat very late in thesmokejumpers’ exposure to MannGulch, and ambiguity about means thatwould clearly remove the threat,compared with the relative clarity of themeans needed by the soccer players todeal with each of their threats.

The point is, whatever chance thesmokejumpers might have had to sur-vive Mann Gulch is not seen as clearlyif we view them as a single group ratherthan as a social system capable ofdifferentiating into many different setsof subgroups. The earlier discussion ofvirtual role systems suggested that anintergroup perspective could be simu-lated in the head and that this shouldheighten resilience. Smith makes it clearthat, virtual or not, intergroup dynamicsaffect survival, even if we overlook themin our efforts to understand the group orthe “outfit.”

As a fourth and final follow-up question,If there is a structure that enables peopleto meet sudden danger, who builds andmaintains it? A partial answer is KenSmith’s intergroup analysis, suggestingthat the needed structure consists ofmany structures, built and maintainedby a shifting configuration of the samepeople. As I said, this perspective makes

sense when time is extended, demandschange, and there is no formal leader atthe beginning of the episode. But thereis a leader in Mann Gulch, the foreman.There is also a second in command andthe remaining crew, which means thereis a top (foreman), middle (second incommand), and bottom (remaining crew).If we take this a priori structure seriously,then the Mann Gulch disaster can beunderstood as a dramatic failure of lead-ership, reminiscent of those lapses inleadership increasingly well documentedby people who study cockpit/crewresource management in aircraftaccidents (e.g., Wiener, Kanki, andHelmreich, 1993).

The captain of an aircrew, who is analo-gous to a player-coach on a basketballteam (Hackman, 1993: 55) can oftenhave his or her greatest impact on teamfunctioning before people get into a tight,time-critical situation. Ginnett (1993) hasshown that aircraft captains identified bycheck airmen as excellent team leadersspent more time team building when theteam first formed than did leaders judgedas less expert. Leaders of highly effec-tive teams briefed their crewmembers onfour issues: the task, crew boundaries,standards and expected behaviors(norms), and authority dynamics.Captains spent most time on those ofthe four that were not predefined by theorganizational context within which thecrew worked. Typically, this meant thatexcellent captains did not spend muchtime on routine tasks, but less-excellentcaptains did. Crew boundaries wereenlarged and made more permeable byexcellent captains when, for example,they regarded the flight attendants, gatepersonnel, and air traffic controllers asmembers of the total flight crew. Thiscontrasts with less-excellent captains,who drew a boundary around the peoplein the cockpit and separated them fromeveryone else.

Excellent captains modeled norms thatmade it clear that safety, effectivecommunication, and cooperation wereexpected from everyone. Of specialinterest, because so little communica-tion occurred at Mann Gulch, is how the

71

Part 4 of 4norm, “communication is important,”was expressed. Excellent crews expectone another to enact any of these fourexchanges: “(1) I need to talk to you;(2) l listen to you; (3) I need you to talkto me; or even (4) I expect you to talk tome” (Ginnett, 1993: 88). These fourcomplement and operationalize the spiritof Campbell’s social imperatives of trust,honesty, and self-respect. But they alsoshow the importance of inquiry, advo-cacy, and assertion when people do notunderstand the reasons why other peopleare doing something or ignoring some-thing (Helmreich and Foushee, 1993: 21).

Issues of authority are handled differ-ently by excellent captains. They shifttheir behaviors between completedemocracy and complete autocracyduring the briefing and thereafter, whichmakes it clear that they are capable ofa range of styles. They establish compe-tence and their capability to assumelegitimate authority by doing the briefingin a rational manner, comfortably, withappropriate technical language, all ofwhich suggests that they have givensome thought to the upcoming flight andhave constructed a framework withinwhich the crew will work.

Less autocratic than this enactment oftheir legitimate authority is their willing-ness to disavow perfection. A goodexample of a statement that tells crew-members they too must take responsi-bility for one another is this: “I just wantyou guys to understand that they assignthe seats in this airplane based onseniority, not on the basis of compe-tence. So anything you can see or dothat will help out, I’d sure appreciatehearing about it” (Ginnett, 1993: 90).Notice that the captain is not saying, lam not competent to be the captain.Instead, the captain is saying, we’re allfallible. We all make mistakes. Let’s keepan eye on one another and speak upwhen we think a mistake is being made.

Most democratic and participative is thecaptain’s behavior to engage the crew.Briefings held by excellent captains lastno longer than do those of the less-excellent captains, but excellent captains

talk less, listen more, and resort less to“canned presentations.”

Taken together, all of these team-building activities increase the probabilitythat constructive, informed interactionscan still occur among relative strangerseven when they get in a jam. If we com-pare the leadership of aircraft captainsto leadership in Mann Gulch, it is clearthat Wag Dodge did not build his team ofsmokejumpers in advance. Furthermore,members of the smokejumper crew didnot keep each other informed of whatthey were doing or the reasons for theiractions or the situational model theywere using to generate these reasons.These multiple failures of leadership maybe the result of inadequate training,inadequate understanding of leadershipprocesses in the late ‘40s, or may beattributable to a culture emphasizingindividual work rather than group work.Or these failures of leadership mayreflect the fact that even the best leadersand the most team-conscious memberscan still suffer when structures begin topull apart, leaving in their wake sense-lessness, panic, and cosmologicalquestions. If people are lucky, andinterpersonally adept, their exposure toquestions of cosmology is confined to anepisode. If they are not, that exposurestretches much further. Which is justabout where Maclean would want usto end.

References

Ancona, Deborah G., and David F.Caldwell. 1992. “Bridging theboundary: External activity and perfor-mance in organizational teams.”Administrative Science Quarterly, 37:634-665.

Asch, Solomon. 1952. Social Psychol-ogy. Englewood Cliffs, NJ: Prentice-Hall.

Barley, Stephen R. 1986. “Technologyas an occasion for structuring:Evidence from observations of CTscanners and the social order ofradiology departments.” Administra-tive Science Quarterly, 31: 78-108.

Barthol, R. P., and N. D. Ku. 1959.“Regression under stress to firstlearned behavior.” Journal of Abnormaland Social Psychology, 59: 134-136.

Bass, Bernard M. 1960. Leadership,Psychology, and OrganizationalBehavior. New York: Harper.

Bass, Bernard M. 1990. Bass andStogdill’s Handbook of Leadership.New York: Free Press.

Beach, Lee R., and Raanan Lipshitz.1993. “Why classical decision theoryis an inappropriate standard for evalu-ation and aiding most human decisionmaking.” In Gary A Klein, JudithOrasanu, Roberta Calderwood, andCaroline E. Zsambok (eds.), DecisionMaking in Action: Models andMethods: 21-35. Norwood, NJ: Ablex.

Bigelow, John. 1992. “Developingmanagerial wisdom.” Journal ofManagement Inquiry, 1: 143-153.

Brown, Richard Harvey. 1978. “Bureauc-racy as praxis: Toward a politicalphenomenology of formal organiza-tions.” Administrative ScienceQuarterly, 23: 365-382.

Bruner, Jerome. 1983. In Search ofMind New York: Harper.

Bruner, Jerome. 1986. Actual Minds,Possible Worlds. Cambridge, MA:Harvard University Press.

Campbell, Donald T. 1990. “Asch’smoral epistemology for socially sharedknowledge.” In Irwin Rock (ed.), TheLegacy of Solomon Asch: Essays inCognition and Social Psychology: 39-52. Hillsdale, NJ: Erlbaum.

72

Part 4 of 4Daft, Richard L., and Norman B.

Macintosh. 1981. “A tentative explor-ation into the amount and equivocalityof information processing in organiza-tional work units.” AdministrativeScience Quarterly, 26: 207-224.

DiMaggio, Paul. 1991. “The micro-macrodilemma in organizational research:Implications of role-system theory.” InJoan Huber (ed.), Micro-macroChanges in Sociology: 76-98.Newbury Park, CA: Sage.

Eisenberg, Eric M. 1990. “Jamming:Transcendence through organizing.”Communication Research, 17: 139-164.

Eisenhardt, Kathleen M. 1993. “Highreliability organizations meet highvelocity environments: Commondilemmas in nuclear power plants,aircraft carriers, and microcomputerfirms. ” In Karlene H. Roberts (ed.),New Challenges to UnderstandingOrganizations: 117-135. New York:Macmillan.

Foushee, H. Clayton. 1984. “Dyads andtriads at 35,000 feet.” AmericanPsychologist, 39: 885-893.

Freud, Sigmund. 1959. Group Psychol-ogy and the Analysis of the Ego. (Firstpublished in 1922.) New York: Norton.

Giddens, Anthony. 1984. The Constitu-tion of Society. Berkeley: University ofCalifornia Press.

Ginnett, Robert C. 1993. “Crews asgroups: Their formation and theirleadership.” In Earl L. Wiener, BarbaraG. Kanki, and Robert L. Helmreich(eds.), Cockpit Resource Management:71-98. San Diego: Academic Press.

Hackman, J. Richard. 1993. “Teams,leaders, and organizations: Newdirections for crew-oriented flighttraining.” In Earl L. Wiener, BarbaraG. Kanki, and Robert L. Helmreich(eds.), Cockpit Resource Management:47-69. San Diego: Academic Press .

Harper, Douglas. 1987. Working Know-ledge: Skill and Community in a SmallShop. Chicago: University of ChicagoPress.

Helmreich, Robert L., and ClaytonFoushee. 1993. “Why crew resourcemanagement? Empirical and theoreti-cal bases of human factors training inaviation.” In Earl L. Wiener, BarbaraG. Kanki, and Robert L. Helmreich(eds.), Cockpit Resource Management:3-45. San Diego: Academic Press.

Helmreich, Robert L., Clayton H.Foushee, R. Benson, and W. Russini.1985. “Cockpit resource management:Exploring the attitude-performancelinkage.” Paper presented at ThirdAviation Psychology Symposium,Ohio State University.

Heydebrand, Wolf V. 1989. “Neworganizational forms.” Work andOccupations, 16: 323-357.

Hirsch, Paul, Stuart Michaels, and RayFriedman. 1987. “ ‘Dirty hands’ vs.‘clean models’. Is sociology in dangerof being seduced by economics?”Theory and Society, 16: 317-336.

Janowitz, Morris. 1959. “Changingpatterns of organizational authority:The military establishment.” Admini-strative Science Quarterly, 3: 473-493.

King, Jonathan B. 1989. “Confrontingchaos.” Journal of Business Ethics,8: 39-50.

Klein, Gary A. 1993. “A recognition-primed decision (RPD) model of rapiddecision making ” In Gary A. Klein,Judith Orasanu, Roberta Calderwood,and Caroline E. Zsambok (eds.),Decision Making in Action: Modelsand Methods: 138-147. Norwood, NJ:Ablex.

Lanir, Zvi. 1989. “The reasonable choiceof disaster: The shooting down ofthe Libyan airliner on 21 February1973.” Journal of Strategic Studies,12: 479-493.

Levi-Strauss, Claude. 1966. The SavageMind. Chicago: University of ChicagoPress.

Maclean, Norman. 1992. Young Menand Fire. Chicago: University ofChicago Press.

March, James G. 1989. Decisions andOrganizations. Oxford: Blackwell.

Maruyama, Magorah. 1963. “Thesecond cybernetics: Deviation-amplifying mutual causal process.”American Scientist, 51: 164-179.

McDougall, William. 1920. The GroupMind. New York: Putnam.

Meacham, John A. 1983. “Wisdom andthe context of knowledge.” In D. Kuhnand J. A. Meacham (eds.), Contribu-tions in Human Development, 8: 111-134. Basel: Karger.

Miles, Ray E., and Charles C. Snow.1992. “Causes of failure in networkorganizations.” California ManagementReview, 34(4): 53-72.

Miller, Danny. 1990. The Icarus Paradox.New York: Harper.

Mintzberg, Henry. 1983. Structure inFives: Designing Effective Organiza-tions. Englewood Cliffs, NJ: Prentice-Hall.

Morgan, Gareth, Peter J. Frost, andLouis R. Pondy. 1983. “Organizationalsymbolism.” In L. R. Pondy, P. J. Frost,G. Morgan, and T. C. Dandridge (eds.),Organizational Symbolism: 3-35.Greenwich, CT: JAI Press.

Neisser, Ulric, and Eugene Winograd.1988. Remembering Reconsidered:Ecological and Traditional Approachesto the Study of Memory. New York:Cambridge University Press.

O’Hare, David, and Stanley Roscoe.1990. Flightdeck Performance: TheHuman Factor. Ames, IA: lowa StateUniversity Press.

73

Part 4 of 4Orasanu, Judith, and Terry Connolly.

1993. “The reinvention of decisionmaking.” In Gary A. Klein, JudithOrasanu, Roberta Calderwood, andCaroline E. Zsambok (eds.), DecisionMaking in Action: Models andMethods: 3-20. Norwood, NJ: Ablex.

Perrow, Charles. 1984. Normal Acci-dents. New York: Basic Books. 1986.Complex Organizations, 3rd ed. NewYork: Random House.

Poole, M. Scott, David R. Seibold, andRobert D. McPhee. 1985. “Groupdecisionmaking as a structurationalprocess.” Quarterly Journal of Speech,71: 74-102.

Pyne, Stephen. 1989. Fire on the Rim.New York: Weidenfeld & Nicolson.

Random House. 1987. Dictionary of theEnglish Language, 2d ed.: Unabridged.New York: Random House.

Ranson, Stewart, Bob Hinings, andRoyston T. Greenwood. 1980. “Thestructuring of organizationalstructures.” Administrative ScienceQuarterly, 25: 1-17.

Read, P. P. 1974. Alive. London: PanBooks.

Reason, James. 1990. Human Error.New York: Cambridge UniversityPress.

Reed, M. 1991. “Organizations andrationality: The odd couple.” Journal ofManagement Studies, 28: 559-567.

Riley, Patricia. 1983. “A structuration-alist account of political culture.”Administrative Science Quarterly, 28:414-437.

Runkel, Phillip J., and Joseph E.McGrath. 1972. Research on HumanBehavior. New York: Holt, Rinehart,and Winston.

Schutz, William C. 1961. “The ego, FIROtheory and the leader as completer.”In Louis Petrullo and Bernard M. Bass(eds.), Leadership and InterpersonalBehavior: 48-65. New York: Holt,Rinehart, and Winston .

Scott, W. Richard. 1987. Organizations:Rational, Natural, and Open Systems.Englewood Cliffs, NJ: Prentice-Hall.

Smith, Ken K. 1983. “An intergroupperspective on individual behavior.”In J. Richard Hackman, Edward E.Lawler, and Lyman M. Porter (eds.),Perspectives on Behavior inOrganizations: 397-408. New York:McGraw-Hill.

Taylor, Shelby E. 1989. Positive Illusions.New York: Basic Books.

Weick, Karl E. 1979. The SocialPsychology of Organizing, 2d ed.Reading, MA: Addison-Wesley.

Weick, Karl E. 1985. “Cosmos vs. chaos:Sense and nonsense in electroniccontexts.” Organizational Dynamics,14(Autumn): 50-64.

Weick, Karl E. 1990. “The vulnerablesystem: Analysis of the Tenerife airdisaster.” Journal of Management,16: 571-593.

Weick, Karl E., and Karlene H. Roberts.1993. “Collective mind in organiza-tions: Heedful interrelating on flightdecks.” Administrative ScienceQuarterly, 38: 357-381.

Westley, Frances R. 1990. “Middle mana-gers and strategy: Microdynamics ofinclusion.” Strategic ManagementJournal, 11: 337-351.

Westrum, Ron. 1982. “Social intelligenceabout hidden events.” Knowledge. 3:381-400.

White, S. K. 1988. The Recent Work ofJurgen Habermas: Reason, Justice,and Modernity. Cambridge:Cambridge University Press.

Wiener, Earl L., Barbara G. Kanki, andRobert L. Helmreich. 1993. CockpitResource Management. San Diego:Academic Press.

Wiley, Norbert. 1988. “The micro-macroproblem in social theory.” SociologicalTheory, 6: 254-261.

74

Part 4 of 4

Notes

75

Part 4 of 4

Notes

76

Part 4 of 4Additional single copies of this document maybe ordered from:

USDA-FS, Missoula Technology & Development CenterBuilding 1, Fort MissoulaMissoula, MT 59804-7294Phone: (406) 329-3900Fax: (406) 329-3719

For further technical information, contactTed Putnam at the above address.

Phone: (406) 329-3965Fax: (406) 329-3719DG: T.Putnam:R01AE-mail: /s=t.putnam/ou1=r01a@mhs-fswa.attmail.com

Library Card

Putnam, Ted. 1995. Findings from the wildland firefightershuman factors workshop; 12-16 June 1995; Missoula, MT.Tech. Rep. 9551-2855-MTDC. Missoula, MT: U.S.Department of Agriculture, Forest Service, MissoulaTechnology and Development Center. 74 p.

This document includes an overview and the findings ofa workshop held to address the human factors involved infirefighter safety. The deaths of 34 firefighters during the1994 fire season, including the 14 firefighters who died onthe South Canyon Fire in Colorado, helped point out theneed for the workshop. Participants explored firefighterpsychology, interactions among firefighters and amongfire crews, and better ways to organize firefighters. Appen-dixes include the four keynote presentations and tworelated papers addressing human factors involved in fire-fighter safety.

Keywords: group behavior, organizational behavior,psychological factors, safety, sociological analysis.

-End of Document-