The urgency of time may make it too onerous forthe extra effort of articulating actions as they are beingperformed, yet most emergency response requires somecommunication.

Explicitly articulated narratives mayalso make clearer that multiple sequences of actions maybe occurring simultaneously, thus resolving role conflictsby allowing multiple ways to accomplish a task

In the context of disaster, social media and other ICTare enablingthe manifestation of a “knowledge commons” [11], a shared information space for victims, onlookers, and the convergent digital volunt

Disasters might be one of the few natural events that override the socio-temporal order; indeed, damage to the routines of social life is a defining characteristic that separates disasters from local emergencies and other disruption

In other words, the articulation work [24], which had to be made explicit previously, became in part implicit by being embedded in the document and in the actions taken upon the document.

Under the stress of the situation, with too many peopledoing too many things at once, the socio-technical infrastructure that underliestheirwork practice wasbreaking down. Star and Ruhleder [25] explain that infrastructure becomes visible only at these points of breakdown. Volunteers directedtheir attention to their social configuration as the critical infrastructure here (the technical infrastructure remains take

Giddens’ theory of structuration explains how social structures, defined as rules and resources or transformational relations, are both the products and the pathways of human action [10]. Employing the concept of duality of structure, Giddens contends that social action both shapes and is shaped by these structures. Orlikowski [20] provides a duality of technology framework for applying structuration theory to research on the role of information technology (IT) within organizational change, whereIT is both the product of human action and a medium of human action, functioning to enable and constrain it. The communication constitutes organization perspective again extends structuration to communicative processes, claiming that communication and the organization co-produce and co-adapt[23], and provides a helpful approach for examining organizing within the virtual organization though the digital traces of its communication [4].

Additional features of the disaster domain are that action is often fast-paced and social structures are emergent [15].

uited. Response and relief work ischaracterized byconvergence of people, information and resources. It is also characterized by improvisation, withvolunteers and formal responders alike innovatively adapting to unique and changing conditions, using resources in new ways, taking on new tasks and assembling into new organizational forms [7,9,15]

Describing the from-scratch conditions of virtual organizing, Finholt et al. write that the “absence of prior structure means group members must develop new structures for sharing information, for example, norms or rules for reporting progress and division of labor” [8p. 292]

relief. Virtual organizations are “geographically distributed [organizations] whose members are bound by a ... common interest or goal, and who communicate and coordinate their work through information technology”[1]

ces. In their study of the digital volunteers who instituted the “disaster desk” in response to the 2011 Peru earthquake,Starbird and Palen [34]reveal how work was restructured in response to the restructuring of the information environment volunteers were working in—which itself was an exasperated response to a confused division of labor and in the end enabled the group to sustain itself relative to its production functions

The desire to assist in disaster events in some way is broad [12, 18], but the mechanisms for enabling action in the form of on-line work or commitment, as with other causes, can beunclear [22, 28, 30].

METHODS

By examiningwork practices, and tracing how those practices are reified in the social-technical organization of a group that is forming and stabilizing as they do the work, we learn not just what this particular group did, but also how the mechanisms by which collective action in digital environments are organizedbottom-up. We also learn how those lessonsaregraduated into prescriptivetop-down direction to sustain and direct future action

Following Orlikowski [29], this analysis unites both practice-based [35]and structurational-basedinterpretations of coordination and social organization [8] to understand the nature of collective work in large, distributed,and emergent groups—groups that havesome existing common motivation to help but have little prior precedent for how that work might be conducted[21]

In the domain of pet advocacy, the latent potential for crowd interaction comes fromintrinsic and extrinsic motivations—we focus on how that potentialwas transformed intoa viable form of distributed, decentralized cooperative work.

As Rosner [35] explains, this goes beyond the “affordances” of objects [28] and instead goes to what the tools represent to their craft and their expert execution of w

Furthermore, tolink this back to the matter of expertise, we see thatexpertise was displayed through material objects:people wore clothing that was consistent with their identification as equine experts (such asboots and cowboy hats),and the Posse memberswore theiruniforms.At the ranch, onejob was to hand out halters and lead ropesto riders. If riders’preferred materials were not available,their expertise allowed them to adapt to what was at

Calling upon media theory, which considers how mass media frames and focusesthekind ofattention an eventreceives,e.g., [4,8], social media can do the same.

priori. Such is the situation with disaster.We easily dismisshow uncertainsituations of disaster areor can become, and how a goalin safety-critical work is to avert situations beforethey become problems. Much of the work in safety-and time-critical matters in CSCW appreciates the implications of this goalon vigilance, mutual awareness, and, of course, error, especially propagated error. It is all too easy to blame “pilot error” when a sequence of preceding systemic conditions took place to set a pilot up for perceiving the problem as he or she did [34,48], including one that warns of hazard. Indeed, disaster can magnifyproblems, not necessarily out of proportion, though that can happen, but rather too so that wefocusonspecific detailswhen many things are happening.

Expertise is a type of embedded knowledgedeveloped within a cultural, social and cognitive environment[6].Expertiseistheability to apply knowledge in different contexts[6], including in emergent situations that require experts to improvise, as Normark and Randall note [29]

Mendonça, et al.[26] and Kendra and Wachtendorf [20] have characterized this as improvisation, whichhas strong parallels to the conversations in CSCW about the nature of situated cognition or situated work [14,44], as well as the relationship between informal as well as formal aspects of work [30,44]

Threaded throughout thesearguments is the idea of distributed cognition particularly as it materializes in the on-the-ground work, but also through prior online preparation.Through this lens, we see how ideation ofsolutions sprung from uncertainexpressions ofproblem statementswhich were quickly forwardedto the local (or local enough) domain experts—horsepeople in Colora

Ethnographic Investigation

Finally, a maincontribution of this research lies in the examination of the solicitation of expertise in a digitally-connected world, where widely distributed and diverse expertise must nevertheless be realized under highly localized conditions.

We see how problem definition, work articulation [37], and the materiality of work[27,35] come together to make the work happen in asocially-, spatially-, and temporally-distributed matter[14].

We see howperformances aroundpaperwork intended to connect the online to the offlineare once again superficial[44], and that the offline work is refiguredat the very endprimarily to communicate its successful completion back to a waiting, online crow

In this online-meets-offline account of cooperative work, we see connections to the classic literature in CSCWaround matters of mutual awareness in safety-critical systems[12]that is partially achieved online andonly “satisficingly”[38] achieved

Disconnection Between Offline & Online ResponseIndeed, amajor criticism of much current crisis social mediaresearch is that it does not consider the relationship between online work and offline or on-the-ground activities(Wulf, et al.[49]is a notable published location, and it isadiscussion often brought up at conferencesand in paper reviews). It is an important conce

By ignoring the diversity and discord of the ‘goals’ of theparticipants involved, the differentiation of strategies, and the incongruence of theconceptual frames of reference within a cooperating ensemble, much of the currentCSCW research evades the problem of how to provide computer support for peoplecooperating through the establishment of a common information space.

On the one hand, the visibility requirement is amplified by this divergence. Thatis, knowledge of the identity of the originator and the situational context motivat-ing the production and dissemination of the information is required so as to enableany user of the information to interpret the likely motives of the originator. On theother hand, however, the visibility requirement is moderated by the divergence ofinterests and motives. A certain degree of opaqueness is required for discretionarydecision making to be conducted in an environment charged with colliding inter-ests. Hence,visibility must be bounded.

Thus, a computer-basedsystem supporting cooperative work involving decision making should enhancethe ability of cooperating workers to interrelate their partial and parochial domainknowledge and facilitate the expression and communication of alternative perspec-tives on a given problem. This requires a representation of the problem domainas a whole as well as a representation, in some form, of the mappings betweenperspectives on that problem domain.

If the decision making process (1) involves a large and indefinite number of peo-ple, (2) requires the integration of a number of different perspectives or domains,and (3) continues for a protracted period of time or even indefinitely, the interpreta-tion of the objects in a common database and hence the construction of a commoninformation space is hampered by the fact that the other originators and recipientsare not co-present.

The fact that information produced by discretionary decision making cannotbe conveyed anonymously has important implications for CSCW systems design.Naturally, such information must be accompanied by the identity of the source.But how to represent and present the identity of the source?

In cooperative work settings involving discretionary decision making, the exer-cise of mutual critique of the decisions arrived at by colleagues is required for allparticipants. Therefore, in order to be able to assess information generated by dis-cretionary decision making, each participant must be able to access the identity ofthe originator of a given unit of information.

At the level of the objects themselves, shareabilitymay not be a problem, but in terms of their interpretation, the actors must attempt tojointly construct a common information space which goes beyond their individualpersonal information spaces. A nice example of how this is a problem has been givenby Savage (1987, p. 6): ‘each functional department has its own set of meaningsfor key terms. [...] Key terms such aspart, project, subassembly, toleranceareunderstood differently in different parts of the company.’

Their importance lies in the interpretation human actors place on themeaning of the representational object. The distinction between the material carrierof information—the object—and its meaning is crucial. The material representationof information in the common space (e.g., a letter, memo, drawing, file) exists asan objective phenomenon and can be manipulated as an artifact. The semantics ofthe information carried by the artifact, however, is, put crudely, ‘in the mind’ of thebeholder, and the acquisition of information conveyed by the artifacts requires aninterpretive activity on the part of the recipient. Thus, a common information spaceencompasses the artifacts that are accessible to a cooperative ensembleas well asthe meaning attributed to these artifacts by the actors.

Cooperative work is not facilitated simply by the provision of a shared database, butrequires the active construction by the participants of a common information spacewhere the meanings of the shared objects are debated and resolved, at least locallyand temporarily. Objects must thus be interpreted and assigned meaning, meaningsthat are achieved by specific actors on specific occasions of use. Computer supportfor this aspect of cooperative work raises a host of interesting and difficult issues.

These protocols, formal structures, plans, procedures, and schemes can be con-ceived of asmechanismsin the sense that they (1) are objectified in some way(explicitly stated, represented in material form), and (2) are deterministic or at leastgive reasonably predictable results if applied properly. And they aremechanisms ofinteractionin the sense that they reduce the complexity of articulating cooperativework.

Therefore, instead of pursuing the elusive aim of devising organizational modelsthat are not limited abstractions and thus in principle brittle when confronted withthe inexhaustible multiplicity of reality, organizational models in CSCW applica-tions should be conceived of asresourcesfor competent and responsible workers.

In this section of the paper we broach two aspects of this articulation issue, onefocusing on the management of workflow, the other on the construction and manage-ment of what we term a ‘common information space’. The former concept has beenthe subject of discussion for some time, in the guise of such terms as office automa-tion and more recently, workflow automation. The latter concept has, in our view,been somewhat neglected, despite its critical importance for the accomplishmentof many distributed work activities

Articulation work

However, in general cooperative workin real world settings has a number of characteristics that must be taken into accountif CSCW systems are to be acceptable to users and, hence, commercially viable:

A cooperative work arrangement arises simply because there is no omniscientand omnipotent agent. Specifically, a cooperative work arrangement may emerge inresponse to different requirements (Schmidt,1990):

The Rich Diversity of Cooperative Work

As a research effort that involves a large number of established disciplines,research areas, and communities, CSCW is an arena of discordant views, incom-mensurate perspectives, and incompatible agendas. However, in the conception ofCSCW proposed here—as a research area devoted to exploring and meeting the sup-port requirements of cooperative work arrangements, CSCW is basically adesignoriented research area. This is the common ground. Enter, and you must change.

CSCW should be conceived of asan endeavor to understand the nature andrequirements of cooperative work with the objective of designing computer-basedtechnologies for cooperative work arrangements.

Our under-standing of the gap is driven by technological exploration through artifact cre-ation and deployment, but HCI and CSCW systems need to have at their corea fundamental understanding of how people really work and live in groups, or-ganizations, communities, and other forms of collective life. Otherwise, wewill produce unusable systems, badly mechanizing and distorting collabora-tion and other social activity.

The gap is also CSCW’s unique contribution. CSCW exists intellectually atthe boundary and interaction of technology and social settings. Its unique intel-lectual importance is at the confluence of technology and the social, and its

Nonetheless, it has been argued here that theunique problem of CSCW is the social–technical gap. There is a fundamentalmismatch between what is required socially and what we can do technically.Human activity is highly nuanced and contextualized. However, we lack thetechnical mechanisms to fully support the social world uncovered by the socialfindings of CSCW. This social–technical gap is unlikely to go away, although itcertainly can be better understood and perhaps approached.

Nonetheless, several guiding questions are required based on thesocial–technical gap and its role in any CSCW science of the artificial:• When can a computational system successfully ignore the need fornuance and context?• When can a computational system augment human activity withcomputer technologies suitably to make up for the loss in nuance andcontext, as argued in the approximation section earlier?• Can these benefits be systematized so that we know when we are add-ing benefit rather than creating loss?• What types of future research will solve some of the gaps betweentechnical capabilities and what people expect in their full range of so-cial and collaborative activities?

The final first-order approximation is the creation of technical architecturesthat do not invoke the social–technical gap; these architectures neither requireaction nor delegate it. Instead, these architectures provide supportive oraugmentative facilities, such as advice, to users.

Another approximation incorporates new computational mechanisms tosubstitute adequately for social mechanisms or to provide for new social issues(Hollan & Stornetta, 1992).

First-order approximations, to adopt a metaphor from fluid dynamics, aretractable solutions that partially solve specific problems with knowntrade-offs.

Similarly, an educational perspective would argue that programmers andusers should understand the fundamental nature of the social requirements.

CSCW’s science, however, must centralize the necessary gap between whatwe would prefer to construct and what we can construct. To do this as a practi-cal program of action requires several steps—palliatives to ameliorate the cur-rent social conditions, first-order approximations to explore the design space,and fundamental lines of inquiry to create the science. These steps should de-velop into a new science of the artificial. In any case, the steps are necessary tomove forward intellectually within CSCW, given the nature of the social–tech-nical gap.

Ideological initiatives include those that prioritize the needs of the peopleusing the systems.

Simon’s (1969/1981) book does not address the inevitable gaps betweenthe desired outcome and the means of producing that outcome for anylarge-scale design process, but CSCW researchers see these gaps as unavoid-able. The social–technical gap should not have been ignored by Simon.Yet, CSCW is exactly the type of science Simon envisioned, and CSCW couldserve as a reconstruction and renewal of Simon’s viewpoint, suitably revised. Asmuch as was AI, CSCW is inherently a science of the artificial,

At a simple level,CSCW’s intellectual context is framed by social constructionism andethnomethodology (e.g., Berger & Luckmann, 1966; Garfinkel, 1967), systemstheories (e.g., Hutchins, 1995a), and many large-scale system experiences (e.g.,American urban renewal, nuclear power, and Vietnam). All of these pointed tothe complexities underlying any social activity, even those felt to be straightfor-ward.

Yet,The Sciences of the Artificialbecame an an-them call for artificial intelligence and computer science. In the book he ar-gued for a path between the idea for a new science (such as economics orartificial intelligence) and the construction of that new science (perhaps withsome backtracking in the creation process). This argument was both charac-teristically logical and psychologically appealing for the time.

The HCI and CSCW research communitiesneed to ask what one might do to ameliorate the effects of the gap and to fur-ther understand the gap. I believe an answer—and a future HCI challenge—is toreconceptualize CSCW as a science of the artificial. This echoes Simon (1981)but properly updates his work for CSCW’s time and intellectual task.2

As Heilbroner (1994) and other researchers have argued, technological tra-jectories are responsive to social direction. I make the case that they may alsobe responsive to intellectual direction.1Indeed, a central premise of HCI isthat we should not force users to adapt.

The coevolutionary form of this argument is that we adapt resources in theenvironment to our needs. If the resources are capable of only partial satisfac-tion, then we slowly create new technical resources to better fit the need.

A second argument against the significance of the gap is historically based.There are several variants: that we should adapt ourselves to the technology orthat we will coevolve with the technology.

A logically similar argument is that the problem is with the entire vonNeumann machine as classically developed, and new architectures will ame-liorate the gap. As Hutchins (1995a) and others (Clark, 1997) noted, the stan-dard model of the computer over the last 30 years was disembodied, separatedfrom the physical world by ill-defined (if defined) input and output devices.

First, it could be that CSCW researchers merely have not found the properkey to solve this social–technical gap, and that such a solution, using existingtechnologies, will shortly exist.

Theproblem, then, was centered by social scientists in the process of design. Cer-tainly, many studies in CSCW, HCI, information technology, and informa-tion science at least indirectly have emphasized a dichotomy betweendesigners, programmers, and implementers on one hand and the social ana-lyst on the other.

In particular, concurrency control problems arise when the software, data,and interface are distributed over several computers. Time delays when ex-changing potentially conflicting actions are especially worrisome. ... Ifconcurrency control is not established, people may invoke conflicting ac-tions. As a result, the group may become confused because displays are incon-sistent, and the groupware document corrupted due to events being handledout of order. (p. 207)

Moreover,one of the CSCW findings was that such categorization (and especially howcategories are collapsed into meta-categories) is inherently political. The pre-ferred categories and categorization will differ from individual to individual.

Because some of the idealization must be ignored to pro-vide a working solution, this trade-off provides much of the tension in anygiven implementation between “technically working” and “organizationallyworkable” systems. CSCW as a field is notable for its attention and concern tomanaging this tension.

Incentives are critical.

People not only adapt to their systems, they adapt their systems to theirneeds

There appears to be a critical mass problem for CSCW systems

The norms for using a CSCW system are often actively negotiatedamong users.

Visibility of communication exchanges and of information enableslearning and greater efficiencies

eople prefer to know who else is present in a shared space, and they usethis awareness to guide their work

Exceptions are normal in work processes.

Members of organizations sometimes have differing (and multiple)goals, and conflict may be as important as cooperation in obtaining is-sue resolutions (Kling, 1991). Groups and organizations may not haveshared goals, knowledge, meanings, and histories (Heath & Luff,1996; Star & Ruhleder, 1994).

One finding of CSCW is that it is sometimes easier and better toaugment technical mechanisms with social mechanisms to control,regulate, or encourage behavior (Sproull & Kiesler, 1991)

because people of-ten lack shared histories and meanings (especially when they are indiffering groups or organizations), information must berecontextualized to reuse experience or knowledge. Systems often as-sume a shared understanding of information.

Yet, systems often have considerable difficulty han-dling this detail and flexibility.

Social activity is fluid and nuanced, and this makes systems techni-cally difficult to construct properly and often awkward to use.

March and Simon’s (1958; Simon, 1957) limited rational actormodel underlies CSCW

I also arguelater that the challenge of the social–technical gap creates an opportunity to re-focus CSCW as a Simonian science of the artificial (where a science of the arti-ficial is suitably revised from Simon’s strictly empiricist grounds).

In summary, they argue that human activity is highly flexible,nuanced, and contextualized and that computational entities such as informa-tion sharing, roles, and social norms need to be similarly flexible, nuanced, andcontextualized.

Thesocial–technical gapis the divide between what we know we must support sociallyand what we can support technically. Exploring, understanding, and hopefullyameliorating this social–technical gap is the central challenge for CSCW as afield and one of the central problems for human–computer interaction.

The term ‘enactment’ is used to preserve the central point that when people act, they bring events and structures into existence and set them in motion. People who act in organizations often produce structures, constraints, and opportunities that were not there before they took action.

The way to counteract catastrophes, therefore, is to reduce tight coupling and interactive complexity. To do this, it seems important not to blame technology, but rather to look for and exag- gerate all possible human contributions to crises in the hope that we can spot some previously unnoticed contributions where we can exert leverage.

Perhaps the most important implication of enactment is that it might serve as the basis for an ideology of crisis prevention and management. By ideology, we mean a ‘relatively coherent set of beliefs that bind people together and explain their worlds in terms of cause-and-effect relations’ (Beyer, 1981, p. 166).

Not only does action simplify tasks, it also often slows down the effects of one variable on another.

Enactment affects crisis management through several means such as the psychology of control, effects of action on stress levels, speed of interactions, and ideology

As people see more, they are more likely to notice things they can do something about, which confirms the perception of control and also reduces crisis intensity to lower levels by virtue of early intervention in its development

As forcefulness and ambiguity increase, enactment is more con- sequential, and more of the unfolding crisis is under the direct control of human action. Conversely, as action becomes more tentative and situations become more clearly structured, enactment processes will play a smaller role in crisis development and managment. Enactment, therefore, will have most effect on those portions of a crisis which are loosely coupled.

These possibili- ties are more likely to be seen if we think of large crises as the outcome of smaller scale enactments. When the enactment perspective is applied to crisis situations, several aspects stand out that are normally overlooked. To look for enactment themes in crises, for example, is to listen for verbs of enactment, words like manual control, intervene, cope, probe, alter, design, solve, decouple, try, peek and poke (Perrow, 1984, p. 333), talk, disregard, and improvise. These verbs may signify actions that have the potential to construct or limit later stages in an unfolding crisis

The assumptions that top management make about components within the firm often influence enactment in a manner similar to the mechanism of self-hlfdling prophecy. Many of these assumptions can increase or decrease the likelihood that small errors will escalate into major crises. Thus, assumptions are an important source of crisis prevention.

Capacity can also affect crisis potential through staffing decisions that affect the diversity of acts that are available. Enactment is labour-intensive, which means understaffing has serious effects.

Capacity and response repertoire affect crisis perception, because people see those events they feel they have the capacity to do something about. As capacities change, so too do perceptions and actions. This relationship is one of the crucial leverage points to improve crisis management.

Action in the form of capacity can affect crisis management through perception, distribution of competence and control within a hierarchy, and number and diversity of actors.

When action is irrevocable, public and volitional, the search for explanations becomes less casual because more is at stake. Explanations that are developed retrospec- tively to justify committed actions are often stronger than beliefs developed under other, less involving, conditions. A tenacious justification can produce selective attention, confident action, and self-confirmation. Tenacious justifications prefigure both perception and action, which means they are often self-confirming.

From the standpoint of enactment, initial responses do more than set the tone; they determine the trajectory of the crisis. Since people know what they have done only after they do it, people and their actions rapidly become part of the crisis. That is unavoidable. To become part of the problem means that people enact some of the environment they face. Had they not acted or had they acted differently, they would face a different set of problems, opportunities and constraints.

Thus, an enacted environment has both a public and a private face. Publicly, it is a construction that is usually visible to observers other than the actor. Privately, it is a map of if-then assertions in which actions are related to out- comes. These assertions serve as expectations about what will happen in the future.

At the heart of enactment is the idea that cognition lies in the path of the action. Action precedes cognition and focuses cognition. The sensemaking sequence implied in the phrase, ‘How can I know what I think until I see what I say?’ involves the action of talking, which lays down traces that are examined, so that cognitions can be inferred. These inferred cognitions then become pre- conceptions which partially affect the next episode of talk, which means the next set of traces deposited by talk are affected partially by previous labels and partially by current context. These earlier inferences also affect how the next episode of talk is examined and what is seen.

An enacted environment is the residuum of changes produced by enactment. The word ‘residuum’ is preferred to the word ‘residue’ because residuum emphasizes that what is left after a process cannot be ignored or left out of account because it has potential significance (Webster‘s Dictionary of Synonyms, 1951, p. 694). The product of enactment is not an accident, an afterthought, or a byproduct. Instead, it is an orderly, material, social construction that is subject to multiple interpreta- tions. Enacted environments contain real objects such as reactors, pipes and valves. The existence of these objects is not questioned, but their significance, meaning, and content is. These objects are inconsequential until they are acted upon and then incorporated retrospectively into events, situations, and explanations.

Enactment is the social process by which a ‘material and symbolic record of action’ (Smircich and Stubbart, 1985, p. 726) is laid down. The process occurs in two steps. First, portions of the field of experience are bracketed and singled out for closer attention on the basis of preconceptions. Second, people act within the context of these bracketed elements, under the guidance of preconceptions, and often shape these elements in the direction of preconceptions (Powers, 1973). Thus, action tends to confirm preconceptions.

Diverse as HROs may seem, we lump them together because they all operate in an unforgiving social and political environment, an environment rich with the potential for error, where the scale of consequences precludes learning through experimentation, and where to avoid failures in the face of shifting sources of vulnerability, complex processes are used to manage complex technology (Rochlin, 1993).

We will argue that HROs are important because they provide a window on a distinctive set of pro-cesses that foster effectiveness under trying conditions.The processes found in the best HROs provide the cognitive infrastructure that enables simultaneous adaptive learning and reliable performance.

We then move to the heart of the analysis and argue that organizing for high reliability in the more effective HROs, is characterized by a preoccupation with failure, reluctance to simplify interpretations, sensitivity to operations, commitment to resilience, and underspecifi ed structuring. These processes reduce the inertial blind spots that allow failures to cumulate and produce catastrophic outcomes.

We illustrate how the time as change paradigm helps to better capture the dynamic aspects of their data through three types of data temporality: TODO lists, reflective activity (biology research) and project management.

Like Orlikowski and Yates, we are particularly concerned with the rela-tionship between temporality and practice, with the ways in which a nego-tiated temporal order arises within, and lends meaning to, individualactivities coordinated in concert. We believe that a detailed understanding ofthe relationship between temporality and practice provides a basis for both amore detailed analytic understanding of the temporality of collaboration, butalso for technological and representational advances in support of cooper-ative activity.

One of the largest-scalestudies exploring this problem was undertaken at the University of Wash-ington (Fidel et al., 2000), where researchers investigated the information-seeking behavior of teams from two different companies, Boeing andMicrosoft (Poltrock et al., 2003). They found that each team had differentcommunication and information-seeking practices, and that current infor-mation systems are oriented toward individual rather than collaborativeinformation-seeking activities. In practice, though, information seeking isoften embedded in collaboration