31
Active Shooter: An Agent-Based Model of Unarmed Resistance Tom Briggs | William G. Kennedy Dec 14, 2016 | Winter Simulation Conference 2016

Active Shooter: An Agent-Based Model (ABM) of Unarmed Resistance

Embed Size (px)

Citation preview

Active Shooter: An Agent-Based Model of Unarmed Resistance

Tom Briggs | William G. KennedyDec 14, 2016 | Winter Simulation Conference 2016

Agenda

Motivation

Background

Model

Results

Conclusions and Future Work

Agenda

Motivation

Background

Model

Results

Conclusions and Future Work

Motivation

Mass shootings rare but increasing

Mass shooting difficult to study; difficult to predict or prevent

Increased active shooter training: Run, Hide, Fight

Systems thinking / complexity science perspective?

Research question

To what degree might the rapid action of a few individuals who physically confront a shooter limit casualties in mass shooting scenarios?

Agenda

Motivation

Background

Model

Results

Conclusions and Future Work

Active shooters & mass

shootings

2000 – 2013: U.S. FBI reported 160 active shooter incidents; 486 killed and 557 wounded

Difficult to predict when or where they occur – shooters generally have informational advantage and element of surprise

Median LEO response time: 3 min

Precedent

Precedent

Source: Blair, J. P., Martaindale, M. H., & Nichols, T. (2014). Active shooter events from 2000 to 2012. FBI Law Enforcement Bulletin.

Prior model: Hayes &

Hayes (2014)

Constructed ABMs investigating details of Senator Feinstein’s proposed weapons bill

Reproduced 2012 Aurora, CO movie theater shooting

Variable that matters most in “number shot” is firearm rate of fire

Prior model: Anklam et al

(2015)

Armed school LEOs and/or staff carrying concealed firearms present

On entering room with CCW staff or LEO, active shooter neutralized

Neutralization assumption may be overly optimistic in light of studies of shooting performance (Lewinski et al., 2015)

No distinction between LEOs and civilians; no possibility of intercept by unarmed individuals

Agenda

Motivation

Background

Model

Results

Conclusions and Future Work

Overall model

Open landscape (concert or outdoor rally)

Randomly-located shooter begins firing

Parsimony: fired shot can hit only one victim and no lethality determination is made

Most agents flee; small proportion of “fighters” attempt to tackle shooter

Parameters control shooter armament & accuracy; percentage of fighters and win probabilities.

Assumptions

Shooter fires one round per second (likely overestimate)

Hit likelihood linear function of range and distance

Rounds keep traveling

Round hit likelihood(accuracy)

Three factors:

-Distance between shooter and target

-Shooter accuracy – human component of shooting performance (user can set at 1.0, if desired)

-Firearm effective range – range at which 100% accurate shooter hits target 50% of the time

Control / baseline condition: No fighters.

Control / baseline condition: Model run ends after 3m40s. Median LEO response time: 3 min (FBI).

Fighters & shooter

Fighters within 1 sec range (running) attempt to tackle the shooter

On reaching shooter, struggle begins –shooter shifts attention from targeting victims to fighter

Likelihood of fighter overcoming shooter depends on multitude of factors, so user sets probabilities

Parameters

Parameter Values Notespopulation 500 1000 5000

7500Agent population

%-who-fight 0.001 0.003 0.005 0.010

Percentage of agent population whoare “fighters” rather than “fleers”

chance-of-overcoming-shooter

0.01 0.05 0.10 Per-tick probability of a fighterovercoming the shooter in a hand-to-hand struggle

shooters 1 Number of shootersshooter-magazine-capacity 10 Rounds that can be fired before a

magazine reload (shooters haveunlimited magazines)

firearm-effective-range 30m 50m 70m Range at which a 100% accurateshooter will hit target 50% of thetime; used in hit probability

shot-accuracy 0.5 0.8 1.0 Human factor in accuracy;combines with firearm-effective-range to determine hit probability ofeach shot

field-of-view 180 degrees shooter’s field of view (see section3.2)

shooter-chance-of-overcoming-fighter

0.5 Per-tick probability of shooterovercoming a fighter in a hand-to-hand struggle

Verification and Validation

Challenging

Hayes & Hayes (2014) compared Aurora model to actual casualties

Possibility of online games, maybe VR, but these lack situational realism

“All models are wrong. Some are useful.”

-George Box

Agenda

Motivation

Background

Model

Results

Conclusions and Future Work

Sample run of experimental condition: ½ of 1% fight. Shooter overcome in 30 sec; 19 casualties.

Overall results

0

20

40

60

0 100 200 300Time (seconds)

Num

ber o

f Cas

ualti

es

Control (No Fighters) Shooter not subdued Shooter subdued

Mean casualties: 30Mean time: 100s

Mean casualties: 63

Mean casualties: 57Mean time: 255s

Results

Casualties concentrated at beginning due to distance and delay

Flee vs. fight – 0.1 vs. 0.4 vs. 0.8

Firearm effective range (30 vs. 50 vs. 70 m) had little effect on casualties

Fighters at a distance at severe disadvantage – implications for ambushing LEO entry teams

Agenda

Motivation

Background

Model

Results

Conclusions and Future Work

Conclusions

Fighters will potentially save lives but increase their own risk

Attention is a scarce commodity

Run / hide helps give LEO more time to arrive and sweep, but historical evidence (VA Tech, Sandy Hook) suggests hardened targets will be bypassed for softer targets

Future work

Model extension / criticism

Rapid collective action / swarm attack

Threshold model for fighters (i.e., only attack once certain number of others do)

Calibrate to SME input

Tom [email protected]: @twbriggswww.twbriggs.com

Thank you

Project details:http://bit.ly/shooterABM

Anklam, Charles, Adam Kirby, Filipo Sharevski, and J. Eric Dietz. 2015. “Mitigating Active Shooter Impact: Analysis for Policy Options Based on Agent/computer-Based Modeling.” Journal of Emergency Management 13 (3): 201–16. doi:10.5055/jem.2015.0234.Blair, John Peterson, M. Hunter Martaindale, and Terry Nichols. 2014. “Active Shooter Events from 2000 to 2012.” FBI Law Enforcement Bulletin. https://leb.fbi.gov/2014/january/active-shooter-events-from-2000-to-2012.Blair, John Peterson, and Katherine W. Schweit. 2013. “A Study of Active Shooter Incidents, 2000-2013.” https://hazdoc.colorado.edu/handle/10590/2712.Hayes, Roy, and Reginald Hayes. 2014. “Agent-Based Simulation of Mass Shootings: Determining How to Limit the Scale of a Tragedy.” Journal of Artificial Societies and Social Simulation 17 (2): 5.Lewinski, William J., Ron Avery, Jennifer Dysterheft, Nathan D. Dicks, and Jacob Bushey. 2015. “The Real Risks during Deadly Police Shootouts Accuracy of the Naïve Shooter.” International Journal of Police Science & Management, 117–27.Police Executive Research Forum. 2014. The Police Response to Active Shooter Incidents.Vickers, Joan N., and William Lewinski. 2012. “Performing under Pressure: Gaze Control, Decision Making and Shooting Performance of Elite and Rookie Police Officers.” Human Movement Science 31 (1): 101–17. doi:10.1016/j.humov.2011.04.004.Wilensky, Uri. 1999. NetLogo. Center for Connected Learning and Computer-Based Modeling. Evanston, IL: Northwestern University. http://ccl.northwestern.edu/netlogo.

References