PromotingCommunityAwarenessthroughOpportunistic ...john.stasko/papers/cscw.pdf · ple’s minds. Next, we deployed a detailed written questionnaire to 18 community members of various

Promoting Community Awareness through Opportunistic,

Peripheral Interfaces

Q. Alex Zhao1

John T. Stasko2

1RA3-250, Intel Corp.5200 NE Elam Young PkyHillsboro, OR 97124-6497

2College of Computing/GVU CenterGeorgia Institute of TechnologyAtlanta, GA 30332-0280

September 3, 2002

Abstract. Social capital is the collective value of the relationships between people,including their level of trust and respect for each other. It accumulates throughsocial interactions such as communication and shared activities. In communities withhigher levels of social capital, members often experience better health, better educa-tion and economical outcomes, as well as better collaboration and sharing. One wayto enhance the level of social capital in a community is through increased communityawareness and community information exchange. Unfortunately, as a communitygrows in size and becomes more geographically dispersed, people’s levels of com-munity awareness typically decline. This research introduces “What’s Happening,”a computer system designed to promote and enhance community awareness in alarge, distributed academic organization. The article describes the design evolutionand resulting system functionality of What’s Happening. The article also reviewsuse of the deployed system and evaluates its perceived utility. What’s Happeningappeared to be effective in enhancing community awareness, but it did not achieveas significant a level of use as hoped.

Keywords: Community awareness, CSCW, informal communication, multimedia,opportunistic interfaces, peripheral awareness

1. Introduction

Generally speaking, a community is a group of people associated ina certain way. People can associate by living in the same residentialarea, attending the same school, being interested in the same topic,taking part in the same organization, or engaging in related professionalactivities. The size of a community can be as small as several people,or as large as the global population on this planet. Even though aperson may associate with many different communities under differentcontexts, each particular community helps define a certain aspect ofthat person’s identity. At the same time, members of a communitycollectively shape the identity of the community.

c© 2003 Kluwer Academic Publishers. Printed in the Netherlands.

cscw.tex; 4/02/2003; 17:16; p.1

2 Zhao and Stasko

Community can also describe the level of cohesiveness that a groupof people experience. For example, community awareness often refers tothe degree that people generally know about each other, about issuesthat affect the community, and about social norms as well as people’sdifferent roles within the community. When members know the commu-nity well, orient toward each other and the group as a whole, and feel astrong sense of attachment to the group, a stronger sense of communityis achieved.

The benefits of communities are sometimes discussed through theconcept of social capital. Similar to physical resources, the relationshipbetween one person and other people has value. Social capital is thecollective value of such relationships as well as the level of tolerance andtrust toward each other, and the inclinations that arise through general-ized reciprocity (Coleman, 1988). Social capital accumulates primarilythrough social interactions such as communication and shared activities(Preece, 2002). It is also an enabler for future interactions that in turnmay produce even more social capital. Conversely, social capital maydissipate over time if it is not reinvested or if it is spent in meaninglesssocial interactions.

A level of community awareness is gained and maintained throughinformation exchange, and is a fundamental aspect of the social in-teractions that produce social capital. For example, knowledge aboutpeople’s interests and what may affect them facilitates informationrouting, which in turn facilitates sharing and exchanging resourcesother than information. Knowledge of each other, combined with ad-equate level of trust and expectation, helps people provide emotionalsupport to each other. Knowledge of each other’s on-going activitiesfacilitates coordination of interdependent actions. And finally, know-ing people’s opinions and intentions, as well as community roles andnorms, helps members overcome collective action dilemmas in whichpeople’s individual efforts become significant only if others participateand contribute the same efforts.

Putnam (2000) demonstrated in his recent book Bowling Alone that,in high awareness and high social-capital communities, members oftenexperience better health, better education and economic outcomes, andof particular interest to this work, better collaboration and sharing.These benefits are not limited to those who actively participate in thecommunity. Everyone, including those who never participate and onlymarginally relate to the community, may benefit from being a part ofthe community. Additionally, from an individual person’s point of view,even though one’s relationship with the general community is usuallynot as strong as that with immediate colleagues and family members,maintaining the “weak-ties” with the rest of the community has many

cscw.tex; 4/02/2003; 17:16; p.2

Promoting Community Awareness 3

benefits such as preparing for future “strong-tie” relationships and ac-cess to locally unavailable but important information (Pickering andKing, 1992).

In our research, we focus specifically on a community of academic re-search — a group of people working in the same academic organization,often at the same locality, sharing similar interests or goals in scientificresearch. Since an academic research community is an environment forcreative work, and creative activities in such organizations are mainlycollaborative (Thompson, 1997), a cohesive and healthy communitynourishes collaborative activities and is potentially more productiveand satisfying than otherwise (Rogers, 2000).

There are several problems facing academic research communitiestoday. First, the rapid growth that some communities have encounteredmay bring extra burdens to the maintenance of community awareness.In a large or geographically separated organization, social interactionswith other people are relatively more difficult than in smaller, collo-cated communities, and therefore likely to be less frequent. The qualityof social interactions, and consequently the restocking process of socialcapital, may suffer. Without an adequate level of knowledge aboutthe community and its members that matches the growth, a personmay have difficulties in adapting to the changes in the community andmay not be able to take advantage of the expanding possibilities ofcollaboration.

Second, although an increasing large percentage of information aboutindividual communities is becoming available on-line in electronic formsthat enable faster access, it is becoming more difficult to notice and pro-cess such information due to our inherent attention limits (Thorngate,1997) and the overwhelmingly large amount of unrelated informationthat is available to us (Lyman and Varian). People may not know theexistence of potentially important information or they may not knowhow to find that information. Moreover, people are sometimes pressuredto decide quickly what available information they should pay attentionto and what they should not, potentially ignoring information thathelps maintain awareness and builds social capital.

Finally, the continuous declination of social capital in virtually allaspects of the American society since the mid-1970’s (Putnam, 2000)also casts a shadow on community awareness. Being an importantcomponent of the civic society, academic research communities havethe responsibilities of not only solving their own problems, but alsoinvestigating socio-technical answers to rejuvenate the sense of commu-nity in the greater society. As human-computer interaction researchers,we need to explore how technologies can help create or enhance socialpractices that fit people’s current lifestyles (Resnick, 2001).

cscw.tex; 4/02/2003; 17:16; p.3

4 Zhao and Stasko

As the primary focus of our research, we examine our home aca-demic unit. The College of Computing at Georgia Tech is an educationand research organization consisting of several hundred faculty, staff,and graduate students in different areas of computing. Due to spacelimitations, this community grew out of its original home building toadditionally occupy two satellite buildings. The first satellite buildingis a short walk from the home building (approximately five blocks) andthe second is a short drive away (approximately 3/4 mile).

The general consensus among College of Computing inhabitants wasthat geographic separation negatively affected the sense of communitywithin the College. To more precisely understand and gauge people’sopinions about community issues, we performed a number of initialinquiries, both formal and informal. Our examination began by simplytalking with people in the community about the community. Thesediscussions provided a baseline for more formal, detailed subsequentinquiries and they informed us about some of the key issues in peo-ple’s minds. Next, we deployed a detailed written questionnaire to 18community members of various backgrounds and followed up with 7interviews. Also, to find out on a larger scale how members viewed thecommunity, we deployed an email survey to the faculty and graduatestudents and received 59 responses. In all these survey instruments, weasked people to rate their familiarity with research and social events inthe community, especially with research work in groups other than theirown. We also asked people to rate the sense of community to obtain anestimate of people’s satisfaction with the state of the community.

We found that even though the physical distances among the threebuildings brought inconveniences and extra burdens on formal collab-orations, people consistently worried more about the reduced chancesfor casual interpersonal interactions. The trend to a growing number offocused research labs within the College was worsening the problem aswell. In addition, people wanted better visibility of community mem-bers’ achievements, better involvement in community events, and betterresource sharing. Most people agreed that we needed ways to foster asense of responsibility among community members and improve thesense of community, even though a certain amount of separation wasinevitable given the amount of growth that had occurred.

Given our background and interests, we were curious to learn if tech-nological aids could help to enhance and repair our sense of community.Unfortunately, relatively few technologies have been developed to helppeople maintain community awareness. On the other hand, a large bodyof research has been focusing on supporting people and task awarenessas well as collaborative activities in organized teams with commongoals (Baecker, 1993). These systems shorten the physical distances

cscw.tex; 4/02/2003; 17:16; p.4


between team members by providing virtual presences of people. Theyallow people to contribute to the common goal across time boundaries,both synchronously and asynchronously, through innovative tools thatsupport collaborative activities. However, these technologies often de-mand full attention from the user and participation often requires theuser to spend a considerable amount of effort. Since the benefits ofcommunity awareness are rarely immediate, using these current tech-nologies to maintain awareness at the community level appears to beinappropriate.

Unlike solution-seeking groupware systems, as those technologies areoften referred to, the notion of communityware emphasizes the discov-ery of knowledge and people in a diverse and amorphous community(Ishida, 1998). In particular, our research in communityware focuseson how technologies can help people opportunistically learn about andstay aware of the vast amount of community-related information thatflows through the physical as well as the virtual work space. Thisinformation can be relatively formal and high-level such as “Whatresearch project is Jane working on now?” to more personal detailed in-formation such as “Did Bill’s son win his soccer tournament game?” Inparticular, learning about information such as this should be achievedwithout distracting or interrupting people from their normal tasks andprocesses.

More specifically, it is our belief that it is desirable to positioncommunity information at the periphery first. Opportunistic interfacesrepresent designs that present information at people’s periphery ofattention without pulling their attention away from tasks at hand.A person accesses the information by chance where the conditions forsuch chances to happen are well understood between the human and themachine, so that an instance of such opportunity is minimally intrusiveto other tasks. In addition, the exact piece of information presented toa person at the time of access is not always deterministic.

In the physical world, for example, flyers posted on telephone polesand billboards set up along highways let people who pass by opportunis-tically process the information if they are not attending to anythingelse. A person does not necessarily know what information they will seewhen they pass the flyers or billboards, and while information foundthere can usually be safely ignored, this medium can heighten a person’sawareness about a certain issue when the information is absorbed.

To summarize, opportunistic interfaces support opportunistic pre-sentation of information and opportunistic access to information. Thisresearch seeks to show that, when used properly, computer supportedopportunistic interfaces can provide lightweight alternatives in facili-tating social information gathering and exchange in academic research

cscw.tex; 4/02/2003; 17:16; p.5

6 Zhao and Stasko

communities. This research contributes to the field of human-computerinteraction by identifying opportunistic interfaces as an important de-sign focus and by exploring such systems in promoting communityawareness. Through the iterative process of developing and evaluatingtwo example applications, we will demonstrate the key challenges indesigning opportunistic interfaces. From this experience, we may beginto build a basic understanding of the role of opportunistic interfaces incommunityware applications.

In the next section of the article, we review work related to ourresearch. In Section 3, we describe our objectives in the project andprovide a detailed description of the two system components we devel-oped. A main focus of the section is a review of the design evolutionof the components. In Section 4 we evaluate the effectiveness of thesystem in achieving our goals of promoting community awareness, andthe last section provides some concluding thoughts.

2. Related Work

A number of systems and projects have influenced our research andthe eventual systems that we developed. In this section, we brieflyreview those influential systems to provide a foundation for the designdiscussions that follow.

Buddy lists such as ICQ1 and Lotus Sametime2 are small applica-tions that show simple online status of remote users through changingfont styles or indicators. ActiveBuddy3 is a collection of software agentsacting as intelligent “buddies” that automatically answer instant mes-sages and provide requested information. Buddy list usage, however,tends to be limited to existing relationships rather than supporting theexploration of information about the unfamiliar parts of the community.Furthermore, interruptions caused by instant messages can hinder theuser’s overall productivity (Cutrel, Czerwinski and Horvitz, 2001).

Irwin (McCrickard, 1999), the Awareness Monitor (Cadiz and oth-ers, 1998), and Sideshow (Cadiz and others, 2002) aggregate dynamicinformation from multiple sources and use small graphical tags or viewsto signal the relevance, magnitude, or urgency of the changes. Thesetools still require notable screen real estate and demand a considerableamount of effort in order to specify what information is being moni-tored, decode what is presented, or find particular pieces of informationin the display.

The once-popular PointCast4 and many other commercial “Inter-net Toolbars” show information in the edges or corners of computerdesktops or in the screen-saver mode. We feel that the developers’

cscw.tex; 4/02/2003; 17:16; p.6


profit-driven business needs were in conflict with the users’ need ofmonitoring secondary information in the periphery of attention. Theresult is that these tools often used flashy animation to attract theusers’ attention away from their primary tasks. Furthermore, when co-existing with other applications on the user’s desktop, these systems,e.g., Tickertape (Fitzpatrick and others, 1998), often relied on contin-uously scrolling animation in order to fit long textual messages in oneor two lines, also distracting the user.

Calm technologies explore interfaces that stay in the user’s peripheryof attention while still providing some value to the user and shiftingsmoothly into the focus when necessary (Weiser and Brown, 1996).The water lamps and pinwheels (Dahley, Wisneski, and Ishii, 1998)and ambient displays in ambientROOM (Ishii and others, 1998) con-vey information in the background, using unobtrusive physical objects,reflections, and sound effects. Audio Aura provides auditory cues abouttasks and colleagues while the user moves around in the physical workspace (Mynatt and others, 1998). These technologies do not distractusers from other tasks or use valuable screen real estate. However,they typically provide low information bandwidth and require a certainamount of effort to decode and comprehend the information.

Most related to our research are a number of recent systems that useshared public displays, typically wall projections or plasma displays, tocommunicate information of interest to a particular community. TheApple On-line Newspaper (Leahy, Housde, and Bellamy, 1998), theAware Community Portals (Sawhney, Wheeler and Schmandt, 2001),and the Plasma Poster (Churchill, Nelson, and Denoue, 2002) use largepublic displays to bring on-line content into the physical space wherepeople work or pass by. The information being displayed keeps in-habitants of the shared spaces updated on local community news andevents, external news, and stories contributed by community members.It also gives people something to talk about when they are in the sameplace. These systems are best suited for organizations that have naturalshared spaces.

Continuing in this style, the UniCast/OutCast/GroupCast trio (Mc-Carthy, Costa, and Liongosari, 2001) employ a similar public styledisplay, but also supports personal information presentation on anindividual’s desktop. The Notification Collage (Greenberg and Round-ing, 2001) and the MessyBoard (Fass, Forlizzi, and Pausch, 2002) aresomewhat similar systems that allow members of small communitiessuch as research groups to post information items of interest suchas images, notes, and web pages. The Notification Collage’s primarydisplay platform is a person’s second monitor, while the MessyBoard isprimarily projected onto a wall in the community’s workspace. Semi-

cscw.tex; 4/02/2003; 17:16; p.7

8 Zhao and Stasko

public displays (Huang and Mynatt, 2003) target communication ofinformation in a small co-located group of people already possessingawareness of each other. They use a predefined screen presentationwith components for presence information, activity attendance, andwork plans. Systems such as these also can focus on people withoutpre-existing knowledge of each other. For example, the Opinionizer(Rogers and Brignull, 2002) is a projected wall display designed tostimulate conversations in strangers who are drawn into its vicinity.

Our research and the system we developed differ in that the com-munity they target is larger than the communities primarily addressedby the previous systems, most of which were designed for small groupsof people already quite familiar with each other. We also focused onusing people’s own desktop computers as a delivery platform, ratherthan shared public displays. Our systems are flexible, however, in thattheir displays can be presented on large public displays, and they werealso used in this manner.

The notion of collage-based image layouts, as used in our web imagescreen saver that is described later, is evident throughout a numberof recent systems. Our web image collage service was inspired by theCollageMachine (Kerne, 1997) and Mandala (Helfman, 1999) that usecollages to facilitate browsing large sets of images and correspondingweb sites. PolyNav (Wittenburg and others, 1999) uses an automaticcollage layout technique to allow viewers to browse videos. Keyframesfrom a video are displayed in a rapid, serial manner using a collagelayout algorithm similar to the one we employ. The Notification Collage(Greenberg and Rounding, 2001) and the MessyBoard (Fass, Forlizzi,and Pausch, 2002) exhibit user-created collages in their displays. Lastly,while we show original, related web images to support quick comprehen-sion, the Kandinsky System (Fogarty, Forlizzi, and Hudson, 2001) usesartistic templates with stock images to produce aesthetically pleasingcollages that may evoke some level of understanding of the suggestedinformation.

3. The “What’s Happening” Tools

The primary goal of this project was to allow people to discover infor-mation about their community in a non-distracting, low effort manner.Our specific focus was on moderate-sized communities, on the orderof a few hundred people, some of whom may possess pre-existing fa-miliarity with each others, but many of whom may not. In particular,the following objectives, formulated largely by our initial studies of the

cscw.tex; 4/02/2003; 17:16; p.8


community and our own personal beliefs and hypotheses, guided thework:

− Provide useful and interesting community-related information andcontent. To jump-start participation and minimize user effort, thesystem should provide a significant amount of automatically gen-erated informative content. The system also should support user-submitted content to allow more personalized direct input.

− Encourage open, cross-group communication. (By “group” we meansmall subsets of community members with significant pre-existingknowledge of each other.) It is possible to help expand people’ssocial ties by fostering an open environment for cross-group com-munication (Pickering and King, 1992). In contrast, within-groupcommunication may be better supported by other applicationssuch as e-mail or instant messaging.

− Exhibit a simple and calm interface. We feel that awareness toolsshould remain calmly in the user’s periphery of attention whenthe person is not directly interacting with them. When the user isdirectly addressing the awareness application, the interface mustbe simple enough to allow the user to quickly finish what he or shewants to do so the person can return to their primary tasks.

The system that we have developed is called “What’s Happening”(WH). It consists of two main components, the WH Communication-Bar and the WH Screen-Saver. The Communication-Bar was designedto reside at the edge of a person’s computer display and use very littlereal estate. It deliberately cycles through pieces of community related orpersonally interesting information, as briefly described earlier in (Zhaoand Stasko, 2000). The Screen-Saver is just that, a normal screen-saverthat is activated when no user input has recently occurred. It showsgraphics and text taken from pages on web sites developed by membersof the community. The pages presented are randomly chosen, fosteringa notion of serendipity and informality. Each of these components isdescribed in more detail below. These descriptions are an expandeddiscussion of the systems described in (Zhao and Stasko, 2002) and arethe main focus of the first author’s doctoral dissertation (Zhao, 2001).

3.1. The WH Communication-Bar

One specific focus of this work was to promote awareness of local com-munity events such as seminars, social occasions, and “get-togethers.”Typically, people use email and newsgroups to promote and learn about

cscw.tex; 4/02/2003; 17:16; p.9

10 Zhao and Stasko

such events, but the over-saturation of both of these communicationchannels has led increasing numbers of people to overlook or ignorethem (Huang and others, 2002). We sought to design another com-munication channel that would be more casual, more informal, andthat would require less commitment in its users, that is, overall be“lighter weight.” People should be able to opportunistically stumbleupon interesting items at the down times that frequently occur duringa work day. In addition to promoting community talks and occasions,we wanted to support messages from individuals. For instance, a personshould be able to post that their daughter is selling girl scout cookiesor that they have an extra ticket to the baseball game for sale. Suchmessages are often considered spam in our community when sent viaemail now.

Most importantly, we did not want this communication channel tointerfere with people’s work or established routines. We did not wantit to intrude in their environment, including their computer displayspace. One possible approach would be to employ a type of ambientdisplay (Ishii and others, 1998), but it appeared to be impractical toimplement such displays in numbers significant enough in a large orga-nization to be influential and such displays are not flexible enough toprovide information on many different topics. We chose to implementa peripheral display, one that would reside at the edge of a person’scomputer display. The peripheral display should be small in size, notbe an attention-grabbing interface, and should facilitate intermittentglancing for information acquisition. We call this peripheral displaythe WH Communication-Bar.

3.1.1. User InterfaceThe Communication-Bar’s user interface has a small footprint on a per-son’s computer display, and is often suitably placed in the corner (seeFigure 1). It is designed to remain visible and not be obscured by otherwindows. The system shows short “blurbs” of automatically collectedlocal content such as official announcements of upcoming communityevents, as well as external items such as news reports and weatherforecasts (see Figure 2 for several examples). Blurbs are shown one ata time, in a cyclic manner. After displaying a blurb for a user-adjustableamount of time (default of one minute), the visible blurb is replacedwith the next one in the current set. Once all the blurbs have beenshown in a cycle, the system loops back to the first and continuouslyrepeats in this manner. On a typical usage day, the active set containsroughly 20 to 30 blurbs.

Each automatically-collected blurb typically remains active for aday or until it is removed from the corresponding data source provider.

cscw.tex; 4/02/2003; 17:16; p.10


Figure 1. The Communication-Bar is at the bottom right corner of the desktop. Allother screen items are intentionally blurred. The Bar uses relatively little screen realestate even on this 1024 x 768 display. In its default horizontal layout, the height ofits main window is fixed at 66 pixels while the width is user- adjustable (340 pixelsas shown here). The user can choose to lay out the interface vertically, in which casethe height will become adjustable and the width will be fixed.

In addition, users can contribute their own content blurbs either byposting new stories or by “following up” on existing content in a built-in chat room style capability. Users can make their postings last froman hour to a week. Chat Rooms, on the other hand, are automaticallypurged after being inactive for two hours.

The Communication-Bar uses multiple levels of detail in the displayof the blurbs to facilitate both general and detailed examination. Themost prominent visual feature is an image depicting the topic, source,or status of the current information, so that with a quick glance, aviewer can decide whether to pursue the information further or switchto other tasks (see Figure 3 for example blurb images). A red flag atthe top right corner of the image signals that the blurb is being shownfor the first time. The title of the blurb and a short summary providethe user with progressively more detailed information. In addition, thetitle of a blurb signals the scope of the information: local communitycontent uses a bold font, external content uses the regular font. Finally,

cscw.tex; 4/02/2003; 17:16; p.11

12 Zhao and Stasko

(a) An automatically collected seminar announcement.

(b) The chat room corresponding to the blurb in (a).

(c) A user-contributed blurb.

(d) External content: a Slashdot blurb in this case.

Figure 2. Example Communication-Bar blurbs.

the user can invoke a web browser to see the original full-text article,if an associated one exists, by simply clicking on the title.

Small iconic buttons at the bottom right corner of the content imageprovide quick access to actions that can be performed on the current

Figure 3. Example icons used in interface. Topics include announcements, focuseditems such as stocks or birthdays, particular sources such as Motley Fool or WiredNews, messages from particular individuals, and weather indicators.

cscw.tex; 4/02/2003; 17:16; p.12


blurb. The trash can button removes the blurb so that the program willnot automatically show it again in the future, even though the blurb isstill stored in memory and can be accessed through a list of availableblurbs. The head-and-question-mark button shows the activity levelin the chat room for the displayed blurb: the bigger and darker thehead, the more recent that someone added to the chat room discussion.Clicking on the head-and-question button retrieves the correspondingchat room (as shown in Figure 2-b), which replaces the chat buttonwith a document button that links the chat room with the web pagefor the original blurb.

The “Menu” button to the left of the interface displays a menuof items to pause or resume automatic cycling, list available blurbs,and show the preferences dialog. The preferences dialog allows theuser to customize the Communication-Bar to fit his or her needs (seeFigure 4). A user can choose either horizontal or vertical layout, adjusthow long the program pauses when displaying individual blurbs, selectwhether the transition from one blurb to the next should be animated ornot, and specify whether clicking on the content image should directlybring up the web page. In addition, users can save the geometry ofthe application, thus enabling the main interface to appear at a fixedlocation on the desktop every time the program starts. Finally, theuser can select the external information sources to monitor throughthe interface, including different news feeds and individual stocks tomonitor. Since community-related blurbs should have higher prioritythan those from external sources, local content is always automaticallyincluded.

The “Post” button on the main interface displays the post-a-blurbdialog that lets the user submit a new blurb by specifying a title, abrief summary, an expiration time, and an optional web address for anydetails (see Figure 5). Since the precise moment that a blurb expires inWH is usually not a critical matter, the Communication-Bar presentsa few choices of expiration time instead of requiring the user to typein a precise time, hence reducing the effort involved in posting a blurb.The image shown to the left of the dialog will be displayed along withthe submission. It provides a space for self-expression in similar waysthat the “zsigs” in Zephyr do for its users (Ackerman and Palen, 1996).

Lastly, the two arrow buttons on the main window allow manualjumping to the previous or the next blurb in the cycle.

Even though there are relatively many controls in the application,only a few of these, the most frequently used ones, are located on themain interface, thereby simplifying its appearance. Furthermore, theprimary intended method of use is for a person to initially configurethe Communication-Bar, but thereafter to leave it in a corner of the

cscw.tex; 4/02/2003; 17:16; p.13

14 Zhao and Stasko

Figure 4. The preferences dialog.

Figure 5. Dialog for posting a new blurb.

display and let it calmly run without any interaction. The deliberatecycling of blurbs is meant to provide users with the opportunity ofviewing an item when the person has a free moment or is in-betweentasks.

3.1.2. Animated TransitionsThe manner in which the application transitions from displaying oneblurb to the next can affect the level of distraction incurred by the

cscw.tex; 4/02/2003; 17:16; p.14


user. We suspected that directly replacing the current display with thenew one or using the popular “tickering” animation as seen on TVprogramming would be distracting and therefore inappropriate. Sinceempirical studies of animations in awareness applications did not existin the literature at that time, we prototyped several transition anima-tions and presented them to trial users. We found that a slow wipinganimation, which gave the illusion of the old blurb being “wiped” awayto reveal the new one underneath, to be the least distracting and thusimplemented wiping in the application.

Interestingly, two separate research projects confirmed our hypothe-ses on the use of transition animation (Maglio and Campbell, 2000; Mc-Crickard, Catrambone, and Stasko, 2001). They found that directlyreplacing old text with new can be disruptive. Instead, using appropri-ate animation can help the application to blend into the periphery withminimal distraction. The studies also found that moving text is moredifficult to read and more distracting than static text, which helped toexplain why a ticker animation was considered obtrusive. Finally, theyfound that discrete animation, which stops the motion for a momentwhen the information is in the view and resumes the motion at a latertime, is less distracting than continuous motion. This supports ourdecision to use animation only on transitions, then stopping any motionwhile a complete blurb is in view. Our wiping animation takes aboutfive seconds, followed by a (default) blurb display time of one minute.

3.1.3. ImplementationThe Communication-Bar is primarily implemented in the cross-platformscripting language Tcl and its associated user interface toolkit Tk(Ousterhout, 1994). The awareness system consists of a server programrunning on a dedicated Unix machine and client programs running onusers’ computers. The server handles collecting data and extractingcontent from local and external sources, broadcasting information toclients, relaying chat messages among the Communication-Bars, andcleaning up old content. The client program is responsible for receivingcontent from the server, filtering the blurbs according to user prefer-ences, and presenting them in an order that interleaves instances fromdifferent sources. The client also submits new blurbs and chat messagesentered by the user to the server.

3.1.4. Design EvolutionThe development of the Communication-Bar was an iterative process(Zhao, 2001). Early versions were deployed and evaluated by a numberof trial users. Based on their feedback, we reduced the number andcomplexity of interactive components on the main interface. For exam-

cscw.tex; 4/02/2003; 17:16; p.15

16 Zhao and Stasko

ple, one of the early prototypes had a number of buttons for selectingfrom several categories of content to show. Another prototype used atree-browser as the main display area for “expanding” or “collapsing”multiple blurbs. Trial users encouraged us to reduce clutter and sup-port quick, peripheral access, so these features were later removed orsimplified. See Figure 6 for a more detailed account of the interfacedesign evolution.

At the beginning of the development process, the Communication-Bar had a small set of automatically generated content items includingthe community calendar of events, local weather forecasts, and Slashdotnews5. Based on user feedback during initial deployment, we graduallyadded more information sources to the application, such as Salon.comnews6 and user-customizable stock quotes from Yahoo!7. The contentserver uses a common XML parser to periodically import content fromweb sites that have XML backends. For those without XML, we devel-oped ad hoc algorithms to extract necessary information from the webpages.

To avoid community-related information being overwhelmed by moregeneral information, it became increasingly important to add more localcontent. Some types of local information were relatively easy to collectautomatically while other types were not. For example, a group of grad-uate students used a special mailing list to announce “Friday HappyHours”, a weekly off-campus social event. We set up an e-mail filteringrule to automatically forward those messages to the WH system whichthen created a blurb for them.

Conversely, although community-related messages often appear ine-mail and newsgroup messages, developing a program to examine anarbitrary message, extract the required information from its text, anddecide its appropriateness for showing in WH is exceptionally challeng-ing (essentially a natural language understanding problem). Instead ofthis automated approach, we use the notion of a content editor. This isa person who chooses and forwards messages such as community relatednews, events, and discussions not generally available through othercommunication methods. The coverage and quality of this manuallyforwarded content will inevitably depend on how well the editor knowsthe community.

3.2. The WH Screen-Saver

The Communication-Bar presents one form of local information (an-nouncement and discussion-focused items), but there also are othersources of information about a community and its members. One inparticular that intrigued us is the set of local web pages about peo-

cscw.tex; 4/02/2003; 17:16; p.16


(a) The first design. The title bar showed the initial name of the application.The small iconic buttons on the left allowed the user to quickly select thetypes of information that they might be interested in. The buttons on thefar right allowed the user to hide the application window or bring it to thefront of the display. The text would automatically scroll up when the mousepointer entered the application window. This interface was considered rather“busy.”

(b) The “What’s Happening” name was adopted in the second design andthe number of content categories was reduced to two. The “minus” buttoncollapsed the current article to show a list of all available articles. This designwas still viewed as unnecessarily complex.

(c) Customizing the application, an infrequent activity, was supportedthrough menu items and a preference dialog in the third design. The jump-back and -forward buttons were made horizontal to mimic those on homeaudio devices. Hint messages displayed at application start up replaced helpdocuments. One drawback of this design is that the “go to web page” and“forget this article” buttons at the end of each blurb caused excessive scrolling.

(d) Text buttons were replaced by iconic ones in the fourth design and movedto the blurb title line. Since they still caused unnecessary scrolling, some ofthem were moved to the blurb image area in the current design.

Figure 6. Interface design evolution of the Communication-Bar.

cscw.tex; 4/02/2003; 17:16; p.17

18 Zhao and Stasko

ple’s research interests, hobbies, travels, family, etc. This set of pagescontains a vast amount of information about people in the community,but its magnitude serves as a negative as well. With the proliferation ofpersonal web pages, it has become increasingly unlikely that people willbrowse the home pages of other community members—too many pagessimply exist. Thus, showing some of the information opportunisticallywithout requiring much access effort from the user may provide morechances for people to learn about each other.

While information in a textual form may not be comprehensible topeople in a short glance, images usually provide good representationsof the content of web pages and they are easier to grasp than text,especially within a short time limit (Helfman, 1999). Therefore, we feltthat it would be more effective to show the images on the web pagesinstead of providing details of the text content, as a way of helpingpeople to learn about other’s activities.

One fundamental problem, however, is that images generally do notfit easily in a small space such as that provided by the Communication-Bar. As an alternative, we decided to utilize a screen-saver as a com-munity awareness tool. First, we designed a server program to generatecollages of images. Each collage is a large JPEG image composed fromsmaller images that the server has collected from a particular localweb page (see Figures 7 and 8 for examples). The program repeat-edly creates different collages and copies them sequentially to a fixedlocation on the network. We also developed a screen-saver client thatperiodically retrieves those collages from the network and displays themon a screen. Since the images embedded in or directly referenced froma single web page are more likely to be related, a collage based on theseimages is more likely to form a consistent message, allowing a casualviewer to get a rough understanding of what the collage is about witha short glance.

3.2.1. Image CollectionTo gather the locations of web pages and images on all web serversin the local domain, the WH system schedules a weekly web crawler8.Based on the results, the collage server then creates a list of unique webpages that have been modified in the previous 18 months, arranges thelist in a random order, and processes each page sequentially. For eachweb page, the collage server randomly selects items from the set ofimages associated with that page, ignoring decorative elements suchas thin separators and small buttons to reduce clutter, and adds theselected images to a blank canvas to form a collage. If the web page hasless than five images, the collage server inserts the first several lines oftext on that page into the background of the collage to help convey the

cscw.tex; 4/02/2003; 17:16; p.18


Figure 7. Example web image collages.

cscw.tex; 4/02/2003; 17:16; p.19

20 Zhao and Stasko

Figure 8. Example local web image collages.

cscw.tex; 4/02/2003; 17:16; p.20


topic of the page quickly. If the canvas appears to be full, i.e., the sumof the areas of all the images added to the canvas exceeds a threshold,or if the total processing time for this collage exceeds its correspondingthreshold, the server will stop adding images to the current collage.Lastly, the server draws the title of the page at the top of the collageand draws the URL of the page at the bottom. It uses a shadow styleto draw the annotations so that they are recognizable on both darkand light backgrounds.

After making the new collage available on the network, the serversleeps for a short period of time to allow the Screen-Saver clients toupdate their displays and give the users a chance to see the collagebefore it is replaced by the next one. In practice, individual collagesremain on a user display for approximately 30 seconds to a minute.

3.2.2. Image LayoutThe initial collage server simply used the system built-in random num-ber generator to calculate where to put an image into the collage canvas.Since the system random number generator only outputs deterministicpseudo-random numbers, the images tended to overlap and cluster onthe collages. Trial users felt that the resulting collages did not utilizeavailable space efficiently and often were not aesthetically pleasing.

A revision of the collage server used a simple layout algorithm tofurther spread out the images on the collages. The algorithm dividedthe collage canvas into four quadrants and added a fifth region of thesame size to the center of the canvas. The regions were labeled from0 to 4 as shown in the left side of Figure 9. The collage server addedimages to the five regions 0, 1, 2, 3, and 4 sequentially, iterating backto 0 after region 4. For each image to be added, the collage serverscaled the image down to no more than 90% of the size of the region,if necessary, and placed the image at a random location within theselected region.

This division method reduced the amount of overlapping and theimages in the generated collages appeared more spread out. Trial users,however, commented that collages including four or five large imageswere not visually pleasing. This occurred because the gaps betweenthe images tended to become similar in size when laying out largeimages with similar aspect ratios. While the images were not laid out inan exactly symmetrical fashion, viewers felt that the resulting collagesuggested symmetry and was too “regular.” Other users commentedthat they did not like those collages because of their monotonous ap-pearance.

The current collage server implements a modification to the spread-ing algorithm that expands each of the five regions by 10% to allow a

cscw.tex; 4/02/2003; 17:16; p.21

22 Zhao and Stasko

0 3

1 2

4

0 3

1 2

4

Figure 9. Dividing the collage canvas into five regions. The original regions areshown on the left. The new technique is shown on the right, where the regions areexpanded to provide a little overlap. Region 4 always overlaps with the other regions.

small amount of overlap across regions (see Figure 9, right-side view).Reaction to this collage layout technique has been generally favorable,and people have noted that the new algorithm produces more attractivecollages.

3.2.3. Value-added CollagesIn addition to showing image collages that help people discover partsof the local community web space, the Screen-Saver gave us a chance toprovide value-added services that may opportunistically help viewersin certain aspects of their daily lives. For example, every 40 minutes orso, the collage server generates a weather collage. This collage containsimages from The Weather Channel9 that depict current weather con-ditions and temperatures, weather radar scans, as well as the currentair quality index (Figure 10).

In the Atlanta metropolitan region, congested traffic has becomean increasingly annoying fact of life. Consequently, between 4:30pmand 7:00pm on weekdays, the collage server builds a traffic collageevery 3 minutes (Figure 11) that helps viewers learn the current roadconditions. The higher frequency provides users an increased possibilityto observe traffic conditions, which is a more vital task during thattime period. The traffic collage contains a traffic map in which roadsare colored to indicate average highway speeds. It also includes imagesfrom highway cameras so that a viewer can assess traffic conditionsvisually. These images are extracted from the Georgia Navigator website10.

Finally, the collage server is also a client of the WH content server.Thus, every ten minutes the collage server generates a news collage

cscw.tex; 4/02/2003; 17:16; p.22


Figure 10. A typical weather collage.

Figure 11. An example traffic collage.

cscw.tex; 4/02/2003; 17:16; p.23

24 Zhao and Stasko

Figure 12. An example collage generated from a Communication-Bar blurb.

based on a Communication-Bar blurb. An example is shown in Fig-ure 12.

4. Evaluation

Evaluating the effectiveness of a tool for promoting community aware-ness is challenging for a variety of reasons. First, measuring communityawareness is simply a difficult task. Community awareness is a complexconcept, a composite notion made up of any number of factors. Judgingthe level of community awareness of one individual over another is achallenging objective task. Furthermore, the presence of interveningfactors makes attributing differences in community awareness to a par-ticular system or tool problematic. Because WH is used day-to-dayby people in their own environments, not a controlled lab setting, at-tributing measured differences in awareness to any one factor is simplyimpossible.

Because of these challenges, we decided to evaluate WH using avariety of measures. Even with the restrictions noted above, we didattempt to quantitatively assess community awareness in WH usersversus non-users simply because we were curious if any differences could

cscw.tex; 4/02/2003; 17:16; p.24


be observed (we hypothesized that they would not). The foundation ofour evaluation, however, was a qualitative observational analysis of thesystem’s use for a period of about one and a half years. We noted howmany people used it, how they used it, when they used it, and why theyused it. We made it easy for people to provide us with feedback aboutthe system, and we interviewed many users and non-users to learnabout their experiences with and impressions of the system. Finally,we felt that one key evaluation measure would simply be the amountof use of the system. If people try and continue to use the system, thenat some primitive level it is successful. Short of adequate assessmentsof community awareness influence, simple quantitative use levels thatare easily recorded and unambiguously interpreted appear to be anappropriate evaluation measure.

The remainder of this section describes how the system was deployedand used, as well as our observations of people’s interactions with it.

4.1. Deployment Status

We announced the availability of an initial trial version of the WHCommunication-Bar to a set of people from the local graduate student,faculty, and staff community through an e-mail message in Novem-ber 1999. We discovered soon afterwards that, although the Tcl/Tklibrary was available on Unix, MS Windows, and MacOS platforms,problems in the MacOS Tcl/Tk implementation made running WHthere problematic.

Based on suggestions that we received after the initial trial deploy-ment, we revised the Communication-Bar and initiated a full-scaledeployment in March 2000 by sending a second notification e-mailmessage and posting newsgroup articles. In addition, we advertisedthe system through fliers posted around the community.

The Unix and MS Windows WH Screen-Saver was announced inmid-September, 2000. People on those platforms were able to install itas their primary screen-saver. In addition, we set up four Windows NTmachines in different labs to use the WH Screen-Saver as the defaultlogon screen-saver, which was active when no one was logged on to theconsole. Used in this manner, the Screen-Saver was a form of informa-tion billboard, visible to people walking by. At the same time, we beganmanually selecting interesting e-mail messages, newsgroup articles, andlocal web content, etc. that were difficult to automatically collect, andsubmitting them to WH.

cscw.tex; 4/02/2003; 17:16; p.25

26 Zhao and Stasko

Figure 13. Number of Communication-Bar users graphed over time.

4.2. Communication-Bar Observations

We built logging facilities in the WH content server to record the Inter-net host addresses of the clients that connected or disconnected to theserver and these respective times. We recorded several user activitiessuch as jumping to the previous or next blurb, visiting web pages,posting new blurbs or adding to chat-room discussions.

Figure 13 shows the history of use of the Communication-Bar fromNovember 1999 to June 2001. By analyzing server logs, we estimate thatat the beginning of the initial deployment of the tool, approximately60 people tried the application in the first several weeks with about25 becoming regular users who kept the application running on theirdesktops at all times. Usage usually declined during final exams andschool holidays when fewer people were present.

The many suggestions about the Communication-Bar that we re-ceived seemed to signal that people were interested in using this tooland helping it to improve. However, the overall user population fellshort of what we hoped and it generally declined over time. Systemusage peaked at about 38 regular users when we re-announced theCommunication-Bar and started its full-scale deployment in early 2000.The introduction of the WH Screen-Saver in September 2000 did notsignificantly affect Communication-Bar usage.

In the summer of 2000, we conducted an initial series of interviewswith people about their use of the Communication-Bar. Thereafter, wecontinued to observe usage and discuss the system with its users. In

cscw.tex; 4/02/2003; 17:16; p.26


April 2001, we deployed a concluding survey with detailed quantitativeand qualitative questions, and we interviewed users again. Below, wediscuss our findings.

People indicated that they used the multiple levels of detail in theinterface to quickly gauge what was shown, and they used the web linkfeature to seek related information when interested. Some users evenshrunk the application window to only display the content image andthe first few words of the blurb title. These people wanted an evensmaller screen footprint in the tool.

Reading a long text blurb in the small Communication-Bar windowrequired scrolling. Although most people thought that the screen-sizeof the application was appropriate, a few users wanted the tool toautomatically use more space when they interacted with the systemand read the current blurb.

We observed that there were a number of different styles of inter-acting with the Communication-Bar. Most people used the tool in theanticipated passive manner: they looked at it opportunistically whenthey were waiting for a lengthy computing task to finish, when theywanted to take a break, or when they were generally not very busyand their eyes came across an interesting blurb by chance. None of theusers that we spoke with permanently disabled the automatic cycling ofthe blurbs. As people became accustomed to the application, they alsobecame less inclined to remove unwanted blurbs in the presentationcycle.

A few people treated the Communication-Bar in a manner similarto the morning newspaper: at the beginning of the day, they would usethe forward button to manually walk through all available blurbs andcheck if there was anything interesting. After reading the “news”, theirusage pattern usually became more passive, similar to that of otherusers.

Users seemed to feel that the Communication-Bar was fairly un-obtrusive. Once they became familiar with it, they tended to forgetabout its existence and focused on other tasks. One user commented inan e-mail message that it was “the first community support tool thatI’ve seen that is unobtrusive enough for me to actually keep it on myscreen.” Another user said that it was “a less intrusive way to keep upwith information than, say, e-mail.” However, one user commented thatbeing curious about whether there was new information available in theCommunication-Bar could be distracting at times. We speculated thatpeople engaged in more of this curiosity factor use than they indicated.

In a few cases, the Communication-Bar even stimulated real-worldinterpersonal interactions. For example, the daily birthday blurb oftenfacilitated people communicating happy birthday wishes to others.

cscw.tex; 4/02/2003; 17:16; p.27

28 Zhao and Stasko

Although several people posted blurbs or engaged in chat-room dis-cussion on the Communication-Bar initially, participation declined tothe level of about one posting per month, not counting blurbs forwardedby designated editors. We consider this to be a key failure with respectto the system achieving its objectives. People felt reluctant to postinformation because of several reasons: they did not know whethertheir posting would be interesting to others; they did not personallyknow the general audience of WH; they did not know what wouldbe appropriate; and they thought other people might post the sameinformation and make their own posting redundant. In addition, a fewusers commented that they did not know whether their postings werefollowed up or not and did not get the feedback to encourage them topost other information.

People commented that even though all users read approximatelythe same set of blurbs which provided similar frames of reference, theydid not know who else was reading the particular blurb that they werereading at that same time. In other words, a certain amount of oppor-tunity for interaction was lost due to the lack of presence awareness, aswell as differences in choices of content and differences in scheduling ofthe blurbs for different people’s instances of the system.

Our evaluation uncovered that the quality and the type of contentstrongly affect usage and participation. People expressed a desire tosee more local content such as who was on leave or at a conference,recent awards or grants that other community members received, recentpublications, academic and industrial visitors, etc. Other suggestionsof increasing the utility and participation of the application includedadding word games, showing titles of classes currently in session, anddisplaying random excerpts from people’s personal web pages.

4.3. Screen-Saver Observations

Because of the manner of deployment, it was impractical to track theexact number of Screen-Saver users, but observational evidence leadsus to estimate that the WH Screen-Saver had a higher number of usersthan the Communication-Bar. Many individuals installed the Screen-Saver on their personal systems. As noted earlier, WH was also set upas the default screen-saver in a number of labs. In addition, the WHScreen-Saver was shown on a large hallway plasma display in our homebuilding (see Figure 14). This way, the Screen-Saver became a part ofthe physical environment, delivering information to whomever passedby.

Users indicated quite a bit of variety in their evaluations of theScreen-Saver. Many users liked the image display (collage) technique.

cscw.tex; 4/02/2003; 17:16; p.28


Figure 14. WH Screen-Saver used as a public hallway information display.

Others sometimes found the collages unpleasing due to the randomnessof the layout algorithm. Most users were satisfied with the depth ofinformation displayed. Others wanted a convenient way to see moredetails about the last collage shown, such as a way to access the webpage where the images were collected.

All the different types of collages were well-received by the majorityof people queried. Some variability in interest levels did exist, however.Many people liked to see other people’s vacation pictures but one per-son wanted a way to filter out these images. Having a window officedecreased at least one user’s interest level in the weather collage. Inaddition, a few people were not interested in the traffic collages becausethey did not drive on the freeways to get home. Overall, people weregenerally interested in collages of local web images.

A few non-users dismissed the Screen-Saver because they felt theywould rarely see it, i.e., they would work at their computer constantlywhen in the office. People who used the Screen-Saver generally had adifferent work style—they often stopped to read, think, or deal withother activities next to the computers, which gave the Screen-Savermore chance to activate, and consequently, they saw the Screen-Saverdisplay more often.

People sometimes reported that they learned new community infor-mation via the Screen-Saver. For example, one person was not awarethat a certain member of the community was recently married untilhe saw the wedding pictures on the Screen-Saver. People also com-

cscw.tex; 4/02/2003; 17:16; p.29

30 Zhao and Stasko

mented that the Screen-Saver sometimes facilitated social interactions.Community members reported discussing items they saw on the Screen-Saver. Occasionally, external visitors would find a particular collageinteresting or useful, and be intrigued by the Screen-Saver. In addition,several people said that the Screen-Saver provided them a convenientand extremely opportunistic way of checking certain useful information,such as weather and traffic maps.

People did not report that having the Screen-Saver activated wasa distraction to conversations with office guests, but our observationsdid not necessarily concur with this. A few people commented that theScreen-Saver even gave them something to talk about and facilitated in-formation exchange. Even though talking about what was being shownon the Screen-Saver did not seem to directly help with their tasks athand, people expressed the belief that they might indirectly benefitfrom the information in the future.

4.4. Surveying the Community

We deployed a concluding survey about the WH tools in April 2001, to14 people who were using at least one of the applications (13 answeredquestions specific to the Communication-Bar and 9 answered questionsspecific to the Screen-Saver). We also deployed a survey to non-usersin June 2001, more specifically, to 55 people. Both WH users and non-users were asked to rate their awareness of community informationand events. A key part of the non-user survey was a question askingpeople why they did not use WH. Based on the responses to the varioussurveys, Figure 15 summarizes average ratings of people’s estimates oftheir familiarity with community events and research, as well as ratingsof the sense of community.

Comparing the ratings in the surveys, we find that the familiaritywith research events among the WH users is statistically significantlyhigher than that among the non-users (p < 0.01). Familiarity withsocial events among the users also is statistically significantly higherthan that among the non-users (p < 0.05). Conversely, familiarity withresearch in other groups among the WH users is statistically signifi-cantly lower than that among the non-users (p < 0.02). Ratings of thesense of community was virtually identical among users and non-users.

One possible explanation of the higher familiarity with researchevents among WH users is that information about research eventsoften contains only a few key points such as time, place, topic, andthe presenter. This type of detailed factual information is relativelyeasier to comprehend and remember with repeated contact, and theWH applications facilitate such contact.

cscw.tex; 4/02/2003; 17:16; p.30


Question WH Users Non-users

Familiarity with research events 5.6 (1.5) 4.4 (1.6)

Familiarity with social events 4.7 (1.5) 3.7 (1.6)

Familiarity with research in other groups 2.4 (1.1) 3.4 (1.6)

Sense of community in local organization 4.1 (1.4) 4.1 (1.1)

Figure 15. Comparing answers to familiarity and sense of community questions. Rat-ings were given on a 7-point scale, where a rating of 1 represented “very unfamiliar”or “very bad”, and a rating of 7 represented “very familiar” or “very good”. Theneutral rating was 4. The first number in each table cell represents the average rating,while the trailing number in parentheses represents the standard deviation. Ratingsbetween the two groups on the first three questions were statistically significantlydifferent (p < 0.05, two-tailed Student’s t-test). Note that this survey includedpeople from all three main buildings of the organization, and the responses wereconsistent across the different buildings.

Information about social events is often conveyed through infor-mal channels and may be expressed in many different styles. It oftencontains additional key points such as the reasons for the event, thetheme of the event, directions to the location, possibly more than oneorganizer and their humorous messages to make the event appear moreappealing. Interestingly, WH users also had a higher average familiaritywith these types of events.

Conversely, WH users showed a lower overall familiarity with re-search outside their own group. One might speculate that WH makespeople more aware of the variety of activities occurring within ourcommunity, but does not provide depth of knowledge. Thus, WH usersmight rate their familiarity lower.

As noted earlier, these potential explanations are purely speculative.Any number of intervening factors, other reasons of causality, or eventhe self-selection of WH users could contribute to the different ratingsamong the groups. Nonetheless, the mere fact that WH users ratedtheir familiarity as they did was an encouraging note.

4.5. Reflections

The previous sections provide a sense of an objective evaluation ofthe WH system, along with direct feedback from the system’s users.In this section, we discuss our own personal, subjective impressions,focusing on two primary conclusions. First, WH did appear to promotecommunity awareness in its users. People felt that they system helpedthem learn about local community events and members, and this seemsto align with our observations. Second, the system did not achieve thelevels of use that we sought and that would allow us to consider it a

cscw.tex; 4/02/2003; 17:16; p.31

32 Zhao and Stasko

success. The number of regular users noted earlier fell short of whatwe consider significant adoption by our community. With respect tothe two different system components, the Screen-Saver appeared to bemore popular and more widely used. In general, the two componentshad a modest set of devoted users, but that set was small relative tothe size of the local community it involved. Accordingly, one of ourprimary questions about the system was why many people did not useit.

When shown pictures of the WH applications and asked why theywere not using the tools, 26 of the 55 non-users whom we surveyed saidthat they were unaware of the Communication-Bar, and 21 indicatedthat they were unaware of the Screen-Saver. In spite of all our promo-tion and advertising of the system, one fundamental problem simplyappeared to be an unfamiliarity with the system.

Another explanation of the lack of wide-spread adoption is thatthe perceived benefits of using the WH applications perhaps did notovercome the associated costs. Installing, running, and setting up theapplications in login scripts were still non-trivial tasks despite of ourefforts in making these processes as easy as possible and writing com-prehensive web pages to help people with the tasks. The flexibility ofusing many different, and incompatible, desktop environments in thecommunity further complicated this problem. Also, even though wemade the Communication-Bar interface small and lightweight, it stillcompeted with other applications for screen space, and the costs ofrunning the application were still relatively high.

Although the administrative staff in our community is the key tra-ditional source of formal, official community information as well asinformal, unofficial information, they largely were unable to use theWH applications because they primarily use the Macintosh operat-ing system on which the Communication-Bar was unstable and theScreen-Saver was unavailable.

Finally, ambiguity of responsibility and low perceived importancewere other issues with the WH deployment. Non-users sometimes re-ported being reluctant to try the applications because they perceivedthe system as a short-term, transient research effort. That view provedto be unfounded, however, as both components still are in use today,over three years after the Communication-Bar was initially deployed.

5. Conclusion

The development of the WH system demonstrates two example designsof community awareness applications, with an emphasis on providing

cscw.tex; 4/02/2003; 17:16; p.32


useful and interesting content through simple and calm interfaces. Thetwo components illustrated the differing implications of two respec-tive design choices—the Communication-Bar that co-existed with otherapplications on the user’s computer desktop versus the Screen-Saverthat had exclusive control of the entire display area. In order to avoiddistracting the user from other tasks and give the user a rough summaryof the information in a short glance, the Communication-Bar utilizeda small on-screen size, a few simple controls, a layout that supportedinformation presentation in progressive detail, and a smooth, gradualanimation that minimized sudden changes in appearance when switch-ing from one instance of information to the next. The Screen-Saver,on the other hand, traded the capabilities of customization and screen-sharing in exchange for the advantage of being able to show large imagesto convey information that was otherwise difficult to communicate.It focused on creating visually pleasing image collages and mixingimage presentations from different sources to let the viewer discoverinteresting aspects of the community.

Our evaluation of trial system use showed that opportunistic inter-faces can convey information in the periphery and unobtrusively enrichone’s information space. In addition, we learned that the effectivenessof community awareness applications depends not only on lightweightinterfaces, but also the information content communicated throughthese interfaces. To foster community awareness, designers should cou-ple awareness technologies with practices in the community culture,such as contributing editors that help create interesting content, setexamples for other users, and encourage participation.

This research presents a number of opportunities for future, follow-on research. First, our observations and the set of suggestions fromusers could be used as the basis for improved designs and a new systemdeployment. It also would be interesting to experiment with similartools in a different style of community. Our academic unit has itsown customs and peculiarities, likely quite different from a corporateorganization or another form of organization. The differing work andsocial customs there might make adoption and use of tools like WH anentirely different proposition.

To improve content, particularly within the Communication-Bar,further research on automated techniques for extracting and postinginformation from other sources is needed. Alternatively, research mightexplore the effects of a much more proactive content editor, a personwho is familiar with community and takes responsibility for posting avariety of information blurbs.

With respect to the Screen-Saver, future research could explore amore customizable version, one that would allow users to select the

cscw.tex; 4/02/2003; 17:16; p.33

34 Zhao and Stasko

types of collages to be shown. In addition to local web pages, collagesof arbitrary web pages could be shown, including those with topical,changing images such as news and sports sites. Extending Screen-Saverfunctionality also appears appropriate, so that a particular user actionwould not terminate the Screen-Saver, but would issue a query foraccessing more information about the collage just shown.

In summary, this research has explored techniques and tools for pe-ripheral awareness and communityware, two new exciting areas withinhuman-computer interaction. By exploring how computing technologiescan complement, not intrude, upon people’s environments and enhancesocial capital within a community, we have learned much to informfuture designs and systems in a variety of areas and domains.

Acknowledgments

This research has been partially supported by Microsoft Corporation.Thanks to the members of the Information Interfaces Research Groupfor their input, assistance, and suggestions.

Notes

1 http://www.icq.com2 http://www.lotus.com/sametime3 http://www.activebuddy.com4 http://www.pointcast.com5 http://slashdot.org6 http://www.salon.com7 http://www.yahoo.com8 http://www.htdig.org9 http://www.weather.com

10 http://www.georgia-navigator.com/traffic

References

Ackerman, Mark S. and Leysia Palen (1996): The Zephyr Help Instance: PromotingOngoing Activity in a CSCW System. In CHI ’96. Proceedings of the Conferenceon Human Factors in Computing Systems, Vancouver, B.C., April 1996. NewYork: ACM Press, pp. 268-275.

Axelrod, Robert (1984): The Evolution of Cooperation. New York: Basic Books.Baecker, Ronald. M. (ed.) (1993): Readings in Groupware and Computer-Supported

Cooperative Work: Assisting Human-Human Collaboration. San Mateo: MorganKaufmann.

cscw.tex; 4/02/2003; 17:16; p.34


Cadiz, J. J., Susan Fussell, Robert Kraut, F. Javier Lerch, and William Scherlis(1998): The Awareness Monitor: A Coordination Tool for Asynchronous, Dis-tributed Work Teams. Demonstrated at CSCW ’98, the ACM Conference onComputer Supported Cooperative Work.

Cadiz, J. J., Gina D. Venolia, Gavin Jancke and Anoop Gupta (2002): Designingand Deploying an Information Awareness Interface. In CSCW ’02. Proceedingsof the Conference on Computer Supported Cooperative Work, New Orleans, LA,November 2002. New York, NY: ACM, pp. 314-323.

Churchill, Elizabeth F., Les Nelson and Laurent Denoue (2002): Designing DigitalBulleting Boards for Social Networking. In CSCW ’02 Conference Workshop onPublic, Community and Situated Displays: Design, use and interaction aroundshared information displays, New Orleans, LA, November 2002.

Coleman, James S. (1988): Social Capital in the Creation of Human Capital.American Journal of Sociology, vol. 94, pp. 95-120.

Cutrell, Edward, Mary Czerwinski and Eric Horvitz (2001): Notification, Disruptionand Memory: Effects of Messaging Interruptions on Memory and Performance. InInteract 2001. Proceedings of the IFIP TC.13 Conference on Human-ComputerInteraction, Tokyo, Japan, July 2001. Amsterdam, Netherlands: IOS Press, pp.263-269.

Dahley, Andrew, Craig Wisneski and Hiroshi Ishii (1998): Water Lamp and Pin-wheels: Ambient Projection of Digital Information into Architecture Space. InCHI ’98 Conference Summary, Los Angeles, CA, April 1998. New York, NY:ACM Press, pp. 269-270.

Fass, Adam M., Jodi Forlizzi and Randy Pausch (2002): MessyDesk and Messy-Board: Two Designs Inspired By the Goal of Improving Human Memory. In DIS’02. Proceedings of Designing Interactive Systems, London, U.K., June 2002.New York, NY: ACM Press, pp. 303-311.

Fitzpatrick, Geraldine, Sara Parsowith, Bill Segall and Simon Kaplan (1998): Ticker-tape: Awareness in a Single Line. In CHI ’98 Conference Summary, Los Angeles,CA, April 1998. New York, NY: ACM Press, pp. 281-282.

Fogarty, James, Jodi Forlizzi and Scott E. Hudson (2001): Aesthetic Informa-tion Collages: Generating Decorative Displays that Contain Information. InUIST ’01. Proceedings of the ACM Symposium on User Interface Software andTechnology, Orlando, FL, November 2001. New York, NY: ACM Press, pp.141-150.

Godz, K. (ed.) (1995): Community Building: Renewing Spirit and Learning inBusiness. San Francisco, CA: New Leader Press.

Greenberg, Saul and Mike Rounding (2001): The Notification Collage: PostingInformation to Public and Personal Displays. In CHI ’01. Proceedings of theConference on Human Factors in Computing Systems, Seattle, WA, April 2001.New York, NY: ACM Press, pp. 514-521.

Grudin, Jonathan (1999): CSCW and Groupware: Their History and Trajectory.In Y. Matsushita (ed.): Designing Communication and Collaboration SupportSystems. Amsterdam, Netherlands: Gordon and Breach Science, pp. 1-15.

Grudin, Jonathan (2001): Partitioning Digital Worlds: Focal and Peripheral Aware-ness in Multiple Monitor Use. In CHI ’01. Proceedings of the Conference onHuman Factors in Computing Systems, Seattle, WA, April 2001. New York, NY:ACM Press, pp. 458-465.

Helfman, Jonathan I. (1999): Mandala: An Architecture for Using Images to Ac-cess and Organize Web Information. In VISUAL ’99. Proceedings of the Visual

cscw.tex; 4/02/2003; 17:16; p.35

36 Zhao and Stasko

Information and Information Systems Conference, June 1999. Berlin, Germany:Springer, pp. 163-170.

Huang, Elaine M., Joe Tullio, Tony J. Costa and Joseph F. McCarthy (2002): Pro-moting Awareness of Work Activities Through Peripheral Displays. In CHI ’02Extended Abstracts, Minneapolis, MN, April 2002. New York, NY: ACM Press,pp. 648-649.

Huang, Elaine M. and Mynatt, Elizabeth D. (2003): Semi-Public Displays forSmall Co-Located Groups. To appear in CHI 2003 Conference Proceedings, FortLauderdale, FL, April 2003. New York, NY: ACM Press.

Ishida, Toru (1998): Towards Communityware. New Generation Computing, vol.16, no. 1, pp. 5-21.

Ishii, Hiroshi, Craig Wisneski, Scott Brave, Andrew Dahley, Matt Gorbet, BryggUllmer and Paul Yarin (1998): ambientROOM: Integrating Ambient Media withArchitectural Space. In CHI ’98 Conference Summary, Los Angeles, CA, April1998. New York, NY: ACM Press, pp. 173-174.

Kerne, Andreas (1997): CollageMachine: Temporality and Indeterminacy in MediaBrowsing via Interface Ecology. In CHI ’97 Conference Extended Abstracts,Atlanta, GA, March 1997. New York, NY: ACM Press, pp. 238-239.

Leahy, Laureen, Stephanie Houde, and Rachel Bellamy (1998): An On-line News-paper for Improving Communication within Communities. The Apple EducationLearning Technology Review, Apple Computer, Issue 3, pp. 14-23.

Lyman, Peter and Hal R. Varian: How Much Information? .http://www.sims.berkeley.edu/how-much-info/.

Maglio, Paul P. and Christopher S. Campbell (2000): Trade-offs in DisplayingPeripheral Information. In CHI ’00. Proceedings of the Conference on HumanFactors in Computing Systems, The Hague, Netherlands, April 2000. New York,NY: ACM Press, pp. 241-248.

McCarthy, Joseph F., Tony J. Costa and Edy S. Liongosari (2001): UniCast,OutCast & GroupCast: Three Steps Toward Ubiquitous, Peripheral Displays.In Ubicomp ’01. Proceedings of the 2001 Ubicomp Conference, Atlanta, GA,September 2001, Berlin, Germany: Springer-Verlag, pp. 332-345.

McCrickard, D. Scott (1999): Maintaining Information Awareness with Irwin.In ED-MEDIA ’99. Proceedings of the World Conference on Educational Mul-timedia/Hypermedia and Educational Telecommunications, Seattle, WA, June1999. Charlottesville, VA: Association for the Advancement of Computing inEducation, pp. 552-557.

McCrickard, D. Scott, Richard Catrambone and John Stasko (2001): EvaluatingAnimation in the Periphery as a Mechanism for Maintaining Awareness. InInteract 2001. Proceedings of the IFIP TC.13 Conference on Human-ComputerInteraction, Tokyo, Japan, July 2001. Amsterdam, Netherlands: IOS Press, pp.148-156.

Mynatt, Elizabeth D., Maribeth Back, Roy Want, Michael Baer, Jason B. Ellis(1998): Designing Audio Aura. In CHI ’98. Proceedings of the Conference onHuman Factors in Computing Systems, Los Angeles, CA, April 1998. New York,NY: ACM Press, pp. 566-573.

Olson, Gary M., and Judith S. Olson (2001): Distance Matters. In J. Carroll (ed.):HCI in the New Millennium, New York, NY: ACM Press, pp. 397-417.

Ousterhout, John K. (1994): Tcl and the Tk Toolkit. Rading, MA: Addison-Wesley.Pickering, Jeanne M. and John L. King, J. (1992): Hardwiring Weak Ties: Individ-

ual and Institutional Issues in Computer Mediated Communication. In CSCW

cscw.tex; 4/02/2003; 17:16; p.36


’92. Proceedings of the Conference on Computer Supported Cooperative Work,Toronto, Canada, December 1992. New York, NY: ACM, pp. 356-361.

Preece, Jenny (2002): Supporting Community and Building Social Capital.Communications of the ACM, vol. 45, no. 4, pp. 36-39.

Putnam, Robert D. (2000): Bowling Alone: The Collapse and Revival of AmericanCommunity. New York, NY: Simon and Schuster.

Resnick, Paul (2001): Beyond Bowling Together: SocioTechnical Capital. In J.Carroll (ed.): HCI in the New Millennium, New York, NY: ACM Press, pp.647-672.

Rogers, Juan D. (2000): Theoretical Consideration of Collaboration in Scientific Re-search. In J. S. Hauger and C. McEnaney (eds.): Strategies for Competitivenessin Academic Research. Washington, D.C.: AAAS, pp. 151-177.

Rogers, Yvonne and Harry Brignull (2002): Subtle ice-breaking: encouragingsocializing and interaction around a large public display. In CSCW ’02 Confer-ence Workshop on Public, Community and Situated Displays: Design, use andinteraction around shared information displays, New Orleans, LA, November2002.

Sawhney, Nitin, Sean Wheeler and Chris Schmandt (2001): Aware Commu-nity Portals: Shared Information Appliances for Transitional Spaces. In CHI2000 Workshop on Situated Interaction in Ubiquitous Computing, The Hague,Netherlands, April 2000.

Smith, Mark K. Community: An Essay Prepared for the Informal EducationHomepage. http://www.infed.org/community/community.htm.

Thompson, Timothy N. (1997): Organizational Creation and Evolution. In P. Y.Byers (ed.): Organizational Communication: Theory and Behavior, Boston, MA:Allyn and Bacon, pp. 235-256.

Thorngate, Warren (1997): More Than We Can Know: The Attentional Economics ofInternet Use. In S. Kiesler (ed.): Culture of the Internet. Mahwah, NJ: LawrenceErlbaum Associates.

Weiser, Mark and John Seely Brown (1996): Designing Calm Technology. PowerGridJournal, vol. 1, no. 1. Available at http://powergrid.electriciti.com/1.01.

Wittenburg, Kent, John Nicol, Jim Paschetto and Chris Martin (1999): Browsingwith Dynamic Key Frame Collages in Web-Based Entertainment Video Services,In Proceedings of the IEEE International Conference on Multimedia Computingand Systems, Florence, Italy, June 1999, Los Alimitos, CA: IEEE ComputerSociety Press, Vol. II, pp. 913-918.

Zhao, Qiang A. and John T. Stasko (2000): What’s Happening? The CommunityAwareness Application. In CHI ’00 Extended Abstracts, The Hague, Netherlands,April 2000. New York, NY: ACM Press, pp. 253-254.

Zhao, Qiang A. (2001): Opportunistic Interfaces for Promoting Community Aware-ness. Ph.D. dissertation, College of Computing, Georgia Institute of Technology,GA.

Zhao, Qiang A. and John T. Stasko (2002): What’s Happening?: Promoting Com-munity Awareness through Opportunistic, Peripheral Interfaces, In AVI ’02.Proceedings of the Working Conference on Advanced Visual Interfaces, Trento,Italy, May 2002. New York, NY: ACM Press, pp. 69-74.

cscw.tex; 4/02/2003; 17:16; p.37

cscw.tex; 4/02/2003; 17:16; p.38

Documents

PromotingCommunityAwarenessthroughOpportunistic ...john.stasko/papers/cscw.pdf · ple’s minds. Next, we deployed a detailed written questionnaire to 18 community members of various