This paper presents primarily the conceptual basis for our work, social theory and in situated action (SA). Information activity is embedded in the process of the creation and communication of meaning across time and place. Effective design of DLS must be rooted in this larger context. SA emphasizes artifacts, practices, and the community of practice. Studying information practices and artifacts grounds our inquiry in people's actual activities. Environmental planning is both cooperative and competitive, so we supplement SA with Pierre Bourdieu's framework for understanding social relations and resources in fields of competition.
To illustrate the uses of this approach, we report preliminary findings for two kinds of artifacts critical to environmental planning: maps and datasets of monitoring data. This is an on-going project, so all findings remain subject to further refinement.
The UC Berkeley Digital Library project is creating a technologically-innovative DL in support of environmental planning. It is work-based, that is, it is intended to be useful to real people doing actual work. The purpose of the User Needs Assessment and Evaluation part of the project is to work with actual and potential users to understand their work in ways that can then inform the design of the DL. More generally, our goal is to understand cognitive work, the role of information in it, and information practices and artifacts. An effective DL neither reproduces nor supplants existing practices and artifacts, but is mutually-constituted with them: technology changes practice and artifacts, practice and artifacts change technology.
This paper presents primarily the conceptual basis for our work. Our work is grounded in social theory and in situated action. It is only by understanding the information user as a person-acting, the larger trajectory of action of which information action is only a part, current practices and artifacts, and the larger field of social relations in which all this is situated that we can begin to understand how DLS can be effective tools and not simply examples of technology in search of a purpose.
The paper concludes with some examples from our on-going empirical work to illustrate the uses of this approach.
A powerful thread of research that is closely allied to our project describes activity is situated (Chaiklin and Lave, 1993; Hutchins, 1995; Lave, 1993; Suchman, 1987; Suchman, 1993): it is performed by specific individuals in specific socio-cultural contexts. This does not simply mean that action derives meaning from its context: the person-acting, activities and practices, context, and artifacts are continually interacting and affecting one another.
With this emphasis on situated activity comes an emphasis on praxis, the key role of day-to-day, routine, habitual activity in constructing meaning and in producing and reproducing social order (Giddens, 1984; Lave, 1993; Lave and Wenger, 1991; Suchman and Trigg, 1993). Socialization is most powerfully a product of daily activity, not pedagogy. Practice both reflects and shapes understanding.
Work too is situated. It is done by particular people in particular times and places, for particular reasons, under specific social and technological circumstances. The person, community, technology, artifacts, and practices of work dynamically constitute one another. Available technology affords certain resources and constraints that affect practice; people's work practices affect how technology is designed and used.
Learning is integral to situated activity (Chaiklin and Lave, 1993). Learning is not simply the acquisition of a body of knowledge, but an on-going process that affects the individual, the community, the practice, and the knowledge. Newcomers learn by participating in practice. The learner, the "master" or teacher, the community, the practice, and the shared understanding are all affected by the process of learning. The community of practice is perpetuated and transformed by the learning process.
Much of the research on communities of practice has assumed, if not homogeneity, at least a congruence among the participants (e.g., Hutchins, 1995; Lave and Wenger, 1991). Environmental planning, however, is a delicate negotiation of limited agreement among parties with conflicting goals from differing communities of practice who come together around the planning task in both cooperation and competition: landowners and environmentalists, scientists and farmers, biologists and engineers, government agencies and citizens.
Pierre Bourdieu (Bourdieu, 1990; Bourdieu and Wacquant, 1992) provides a framework for understanding social relations and resources in fields of competition. We treat this topic in more detail elsewhere (Schiff, Van House, and Butler, 1997). In brief, Bourdieu uses the metaphor of the game. The field to be studied, in our case, watershed planning, is like the playing field. But the competition is not just to win but to define the rules, the players, the currency of competition, and the boundaries of the playing field. Different groups value different capitals. They have different ways of seeing the world and recognize a different set of options for action under given circumstances. Habitus is an individual's particular outlook and orientation, most strongly influenced by socio-economic background, education, and profession--and community of practice. It can be thought of as the prism through which we view the world, creating a tendency for each of us to see and not see certain things. Biologists, for instance, may think of water as a habitat for fish, while farmers think of it as a resource for their crops. It affects one's interpretations and even the set of actions that seem possible. Habitus is both structuring in that its permanent imprint colors our vision, and structured, in that we are constantly adding to it through the accumulation of our experiences.
The field variously called information studies, information science, or library and information studies/science has long been interested in information institutions, technology, artifacts, and practices: libraries, information systems, abstracts and indexes, classification schemes, documents, to name a few examples. The practices with which this field has been concerned have been primarily those that (1) take place within the boundaries of the information organization (e.g., the design, operation, and use of libraries and of information retrieval systems), or, (2) consist of use of information representations (e.g., information-seeking behavior of specific groups). With a few exceptions (e.g., Dervin, 1992), however, this research has tended to pick up the trajectory of action at the point at which the individual enters the library door or begins an interaction with the information system or tool.
Information studies research has also been subtly normative; it has been concerned with whether subjects are using tools and representations "correctly" or "well," with whether they "understand" and "make good use" of the functionality available. Some research, for example, has used only librarians as subjects to control for other subjects' lack of expertise in the use of information systems.
Information activity, however, is simply human activity. Information and information systems, practices, and artifacts are embedded in the process of the creation and communication of meaning across time and place. As all work is situated, so is information work, and its context includes the larger project that it supports. Information action is part of a continuous trajectory of action. People seek out information for some purpose. The understanding of information action, tools, and representations must be in relation to the person-acting and the social and technological context and the actor's intentions.
Knowledge exists in a field of action, of social interaction. It is socially mediated and subject to change. Knowledge is constructed in the interplay among the knower, the activity, and the context. It is often contested. Issues of authority and interpretation are mediated by the community of practice. Context and community of practice have been considered, to some degree, in library and information system design. A social science abstracting and indexing service uses different terminology from one in engineering; a university library differs from a public library. But these differences have been for the most part, of material necessity, limited to those that are readily apparent and apply to large classes of individuals. The changes in information technology that are referred to by such terms as "the Digital Library," however, are transcending these limits and opening new possibilities (Butler, Schiff, and Van House, 1997).
Our interest is in how information practices and artifacts interact with human action, social reproduction and transformation, and the creation of meaning; how they constitute, reproduce, and transform practices and communities of practice, and how they are changed by practice and practitioners. By looking at human action through the lens of information, we come to a better understanding of both information-related action and the development of practices and tools to support people's actions.
Our concern is in the end pragmatic: the development of better information systems, practices, tools, and artifacts to support information activity. With the rapid development of information technology, the slow and ad hoc mutual constitution of information practices, tools and artifacts has accelerated and had to become more reflexive and better able to respond to the unique circumstances of various communities of practice. Design of effective information systems requires an understanding of and incorporation of the way in which technology and practices mutually evolve. Effective design, then, requires attention to people's actual and not idealized information practices, the purposes that they serve, and the users' context.
Practices are the product of the field, community of practice, and habitus. To borrow Bourdieu's analogy of the game, players have a certain orientation and understanding of the rules, and certain capabilities which they bring to the game. As they play, they anticipate and respond almost instinctively, based on their habitus. Thus practices can only be understood and actually seen within the boundaries of a specified field and community of practice.
Different communities of practice have different information practices, orientations, understandings, needs, and preferences. Chartier (1994), for example, discusses what he calls communities of interpretation, which have different norms and conventions for the uses of books, different ways to read and use shared materials, and different methods of interpretation. Research on the information needs and uses of specific disciplines, such as scientists and social scientists attempt to elucidate some of these differences, but in a very limited way.
For example: an evolving draft of a report structures the work of the research group engaged in creating it. A map reflects the boundaries of an environmental planning effort and therefore the stakeholders and the problem domain. A map that shows soils types and vegetation but not the distribution of fish and animal species will orient the planning around some factors and not others.
Communities of practice share artifacts. The research report, medical record, map, environmental impact report, or legal form, to name a few examples, are artifacts that structure practice and interpretation. For example, the professional derives from a medical record an understanding of the patient by knowing the meaning of, not only the content, but the sources and format of information in the record (e.g. patient self-reports versus lab tests, nurses' observations or doctors', new and old data) (Berg, 1996).
Artifacts serve an important role in situated learning. The student of environmental planning may learn much in the classroom, but reading and participating in the writing of environmental impact reports structures the practice by which the student will work with his or her colleagues to actually do the work. In this way artifacts play a key role in the reproduction of communities of practice.
Libraries have long been concerned with public or published representations (e.g., books, maps, images) by which information is transported across space and time. (They are much less often concerned with unpublished representations such as manuscripts.) Human cognition is a social and cultural process in which publications play a role as shared prior experience and thinking (Chartier, 1994), and libraries have played an important role in the preservation and communication of the social and cultural record. Libraries have been less concerned with the practices by which these artifacts were produced and used. Digital information has made information representations more changeable, more responsive to changing practices, and has made permeable the boundary between public and private, published and unpublished, reading and writing, and reading and manipulation.
The medium and the practices mutually constitute one another -- we can do things with electronic texts that radically change the affordances of the text. Content and medium are interdependent but distinct. The affordances of electronic media, paper, and other more "traditional" media have been and will continue to be the subject of much discussion (e.g., Butler, Van House, and Schiff, 1997; Chartier, 1994).
We are interested, then, in the relationship among the text (or content), the medium, the practices, the user, the user's purposes, and the context. Effective design of tools and representations requires attention to this multidirectional, dynamic relationship.
In our work we approach the study of people's information actions through their everyday practices in the creation, reproduction, transformation, and use of information artifacts, technology, and practices. This is for both theoretical and practical reasons. One reason is the emphasis on praxis in situated activity. Furthermore, information is an abstract and vague concept, open to many definitions (Nunberg, 1996). It is not as salient to the users of information as it is to researchers and information professionals. People can talk to us about concrete practices and artifacts, and we can observer them.
Digital libraries have the potential for transforming information artifacts and practices and therefore work and work practices and possibly even communities of practice. It is impossible to fully comprehend the myriad effects of current information artifacts and practices, let alone the impact of innovations. However, it is important to consider the affordances of existing artifacts and practices and how the digital library may, inadvertently or deliberately, alter the work and the communities of practice that it supports.
Their task is to come to a shared understanding of the current state of their watershed, the threats to it, and the likely outcomes of various future actions, and a consensus on the actions that should be taken by various levels of government, individual landowners, and other stakeholders. They do not have to come to an all-embracing agreement, only to a negotiated shared decision space that is likely to include many trade-offs (e.g., timber harvesting will be allowed with certain limits and safeguards, a compromise between those who wish to ban timber harvesting and those who wish for it to be unimpeded). The result is often not a plan in the form of a blueprint but a shared understanding that will guide participants' future actions on more harmonious ways. Participants often speak of "educating" one another. For example, environmentalists and government agency staff speak of educating landowners to think in terms of "upstream" and "downstream," to realize the interdependencies of their actions with those of their neighbors. This orientation becomes embedded in people's interpretations and practices and its impact may become invisible; it is simply part of the shared understanding (Innes, 1987).
Information and information practices and artifacts play key roles in this process. Information is a resource in the decision-making process: not just for making "good" decisions, but for defending points of view. Information artifacts such as maps of soil types and vegetation are used as orienting devices for discussion and decision-making. Decisions may or may not be instantiated in documents. (A useful political device, we have discovered, is the unwritten plan, that is, "everyone knows" what was decided but nothing is written down, deliberately, leaving room for interpretation.)
Digital libraries have the potential to drastically alter the creation and use of information artifacts and the practices and social relations that they represent and support. The purpose of this paper is not so much to report the specifics of one project as to illustrate the implications for research and for DL design and evaluation of the conceptual approach described above.
We report preliminary findings for two kinds of artifacts that are critical to environmental planning: maps and datasets of monitoring data. This is an on-going project, so all findings remain subject to further refinement.
These data are of great interest to many users. Monitoring data are important for assessing current conditions, identifying problem areas, developing forecasts, formulating policies and action plans, and evaluating policies and actions. The data are incorporated into analyses that result in major public and private decisions with significant environmental, economic, and social impact.
Once collected, such data are often unique and therefore extremely valuable to many people for a variety of purposes. Long time series and data on multiple variables for the same site and/or period of time are particularly useful, so analysts often combine data from multiple sources.
The data go through a complex chain of processing from raw data to summarization, analysis, and interpretation. They may be incorporated into intricate models to describe and forecast complex phenomena. The practices surrounding these data require considerable expertise. For example, water supply forecasts are important in California, which relies on an extensive network of dams, rivers, and canals to capture water during the winter and spring, store it, and redirect it all over the state. One source of water supply data is snow pack, collected from gauges of various sorts in the mountains during the winter. A set of readings from a number of discrete sites throughout the state are corrected, summarized, interpreted, and incorporated into complex models that forecast water supply. The data undergo many transformations that require judgment and expertise. For example, provision must be made for missing data points or errors in data collection. One source of water supply data is "snow pillows," spring-loaded platforms in remote mountain areas that report the weight of accumulated snow. During heavy snow seasons, however, the pillows sometimes freeze into place and report an invariant weight over time. An expert observer can often spot this kind of measurement error and extrapolate a corrected reading based on other data. The process of generating forecasts of water supply from these data requires considerable expertise and judgment, and was described to us as being as much art as science.
Changes in computing and telecommunications have made other distribution methods possible. Many government agencies maintain distribution lists of individuals and agencies who receive more detailed data summaries on a regular basis. Because the names on these lists are known and to some degree controlled, the distributors know what level of expertise recipients are likely to have.
Many agencies will release raw data on request. Some will run custom analyses on demand. Either generally requires that the requestor track down the individual who controls the data. Some data providers prefer this method because they can screen recipients and their needs, explain the limitations of the data, and discuss the possible uses and interpretation of the data.
Agencies share data through such mechanisms as ftp sites. Generally, the individuals and organizations who know where and how to obtain the data are assumed to have the expertise to understand and use the data appropriately.
DL technology could drastically change the means by which monitoring data are distributed. This proves to be a mixed blessing. Concerns for "quality control" were often raised in our interviews. This issue is not unique to this DL, or to DLs in general, of course. But it is illuminating to look more closely at what this topic means in this context.
Our informants were generally concerned about two areas of quality control: misuse of data, and the distribution of poor quality data. Many were concerned that making unanalyzed data readily accessible to a large and uncontrolled user community could result in inappropriate use of the data or incorrect interpretations. Using data appropriately requires an understanding of what the data do and do not represent. This in turn requires an understanding of data collection methods or protocols; the limits of measurement devices; definitions of data elements; summarization and data cleaning methods; and a wide variety of other contextual factors. Combining data from different sources, which is often necessary to, for example, assess the effect of certain actions on water quality, requires even more understanding of the data and the analysis methods.
For example, data on fish supply comes from occasional boat trips that net fish and measure their catch. The time of year and day, the stage in a species' spawning cycle, weather conditions, the type of net, the depth at which the net is trolled, and many many other factors all affect the catch. A comparison with "last year at this time" that did not account for these other contextual factors could be seriously misleading.
Another major concern was about the ease with which data of uncertain quality could be made widely available. Environmental data is collected by many sources: federal, state, and local agencies, researchers, landowners, environmental groups, and citizen action groups, to name a few. Given the continual need for more and more detailed data, a major focus of effort in this community has been locating and sharing data from many sources. Some of our respondents, however, were concerned about the data collection protocols followed by perhaps-untrained individuals, differences in definitions of data elements and measurement methods, possible biases introduced by special interests, the variability in data quality that could result, and uninformed users' inability to evaluate the data or the source. Other respondents saw the democratization of the data supply as a useful addition to the pool of data available and a necessary counter-measure to the biases and limits of "official" sources, and as empowering citizens to take a more active role.
The solution to the quality control problem that is usually proposed is meta-data: attach to the dataset appropriate contextual information. This is a necessary but only partial solution. Knowing the source of the data, for example, is most useful for an "insider" who knows the capabilities and biases of the sources.
In sum, in the past data providers and recipients generally had a certain level of expertise in data collection and analysis and in the subject area and invested considerable effort in the difficult tasks of acquiring, manipulating, and distributing data. Often, acquisition of the raw data required some form of interaction with the collector or keeper of the data, who could educate and screen (and be screened by) data users. "Authorship"--the sponsoring agency--acted as a proxy for a number of contextual factors that helped the respondent understand the likely orientation of the data providers. Likewise, data providers had to have technical expertise and access to appropriate technology, and expend considerable effort. These technical barriers acted as social barriers, increasing the likelihood that data providers and users would share certain practices and understandings, that is, be members of the same (or at least closely-related) community of practice with shared understanding.
We found that paper fit the existing practices using maps well. Paper is portable and independent of technology, so paper maps can travel to meetings in various sites and go out in the field. Many meetings were focussed on multiple large paper maps that could be compared to one another and readily seen by large numbers of people. Paper maps can be immutable or mutable (or both at the same time): a printed, "official" map can be hand-colored or have a mylar overlay applied to reflect added data. Paper maps of adjoining areas could be taped together to present a larger picture, literally, with no sacrifice in scale.
Maps also have the advantage of being a common artifact, with many common practices. People engaged in environmental planning are generally very familiar with maps, which is not true of computer technology.
We also found differences across participant groups in how maps were used. The extent and scale of the map used varied with the groups' orientation: local people tend to look at high resolution maps covering a small area; state agency employees take a larger perspective. Some groups (e.g. landowners) are strictly map users, others (state agency personnel) are also map creators.
An interesting practice with regard to maps gave rise to what our colleagues Robert Twiss and Howard Foster dubbed a need for "fuzzy GIS". Many of the maps in documents are schematic, without the contextual information needed to determine the exact boundaries of the area represented. This is often deliberate: once the boundaries are drawn, the stake-holders' personal interests become clear. Is my land in or outside the study area? Are there endangered species on my land or only on my neighbor's?
It also illustrates the usefulness of focussing on information practices and artifacts as orienting devices for research on information activity. People can report on, and researchers can observe, practices and artifacts and place them in their context. Digital libraries create or allow, and are supported by, changes in practices and artifacts. In considering how we can design DLs to better support the work, we need to understand the web of social relations, the knowledge base, the set of practices of which work is a part, and the role of information and information artifacts and practices in this larger set of relations. Context is a complex construct and its treatment in research is a thorny problem (Chaiklin and Lave, 1993; Dervin, 1996), but one that must be addressed.
An important context for information activity is the community of practice. If knowledge is socially-mediated, with meaning located in fields of social interaction, information, tools, practices, and representations are most meaningful in relation to a community of practice. Information is interpreted and cognitive authority assigned within communities of practice.
Digital libraries can be customized to communities of practice in ways that are not practical with traditional libraries and information systems. For example, the UC Berkeley Digital Library's MVD architecture views the world in layers which incorporate behaviors. The layers allow for viewing and manipulating documents in new and unique ways. They also allow for the annotation of layers and the sharing of those annotations among a community. Layers could be customized to a community of practice.
Digital libraries are also often able to cross communities of practice. The UC Berkeley Digital Library is available on the World Wide Web to anyone who wanders into it. But in crossing lines of communities of practice, we lose the common understandings that support the communicative activity that is the sharing of information across individuals, space, and time. The concern about "inappropriate" use of raw data and the availability of "poor quality" data reflects the problems raised by unanchoring data from an interpretive context. Maps, on the other hand, seem to more readily cross communities of practice, as common artifacts that are generally more easily interpreted without special skills.
Some social relations are cooperative, but many are competitive. In competition, information is a capital. Some of our subjects maintain a competitive edge by means of proprietary information. Others try to get their information and interpretations adopted as a means of influencing shared understanding. If the removal of vegetation for homebuilding is believed to have contributed to flooding, then timber clear-cutters will bear less of the blame. Any change in information availability affects the constantly changing balance among competitive forces. The digital library is political, regardless of the intentions of its creators.
This is not an argument for the status quo or even for conservatism. It is naive to expect that we could perfectly map out the current situation, correctly project the effects of possible changes in technology, and perfectly match a DL to its context. It is an argument for taking a wider view, seeing DLs as social and well as technological artifacts, and taking an incremental, practice-based approach to design and implementation of information technology. Just because we can build it, that doesn't always mean that we should.
Involving users in DL design is a step in the right direction. The participatory design approach to technology (Greenbaum and Kyng, 1991), which incorporates users into the design process, is most appropriate for limited-domain information systems with clearly-defined, relatively homogeneous user groups and uses. DLs are larger, more complex, and more diffuse.
We have said that learning is critical to situated action; that it is continual; and that it changes the learner and the "teacher," the practice, and the knowledge base. We should take a learning-based approach to the design of digital libraries, recognizing the complex interrelationships among the DL, the work supported, the practices, and its users, and the social and technological context.
Bourdieu, P. (1990). In other words: essays towards a reflexive sociology, trans. Matthew Adamson. Stanford, CA.: Stanford University Press,.
Bourdieu, P. and Wacquant, L.J.D. (1992). An invitation to reflexive sociology. Chicago: University of Chicago Press.
Butler, M. H., Schiff, L.R., and Van House, N.A. (1997). Is a ``digital library'' a ``digital'' ``library'': affordances of the metaphor. Submitted to the ACM Digital Libraries Conference 1997.
Chaiklin, S. and Lave, J., eds. (1993). Understanding practice: perspectives on activity and context. Cambridge, UK: Cambridge University Press.
Chartier, R. (1994). The order of books: readers, authors, and libraries in Europe between the fourteenth and eighteenth centuries. Trans. By Lydia G. Cochrane. Stanford, CA: Stanford University Press.
Dervin, B. (1992). From the mind's eye of the user: the sense-making qualitative-quantitative methodology. In Jack D. Glazier and Ronald R. Powell, eds. Qualitative research in information management. Englewood, CO: Libraries Unlimited.
Dervin, B. (1996). Given a context by any other name: methodological tools for taming the unruly beast. ISIC 96: information seeking in context. August 14-16, Tampere, Finland. To be published in: Reijo Savolainen, Pertti Vakkari, and Brenda Dervin, eds. Information seeking in context. London, UK: Taylor Graham.
Giddens, A. (1984) The Constitution of Society. Berkeley: University of Calif Press.
Greenbaum, J., and Kyng, M. (1991). Design at work: cooperative design of computer systems. Lawrence Erlbaum Assocs.
Hutchins, E. (1993) Learning to navigate. In Chaiklin, S., and Lave, J., understanding practice: perspectives on activity and context, Cambridge, UK: Cambridge Univ Press. P. 35-63.
Hutchins, E. (1995). Cognition in the Wild. Cambridge, MA: MIT Press.
Innes, J. (1987). Introduction to the transaction edition: understanding the alliance of knowledge and policy. In Innes, J. Knowledge and public policy: the search for meaningful indicators. 2nd expanded ed. New Brunswick, NJ: Transaction Books, 1987.
Latour, B. (1987) Science in Action. Cambridge, MA: Harvard University Press.
Lave, J. (1988). Cognition in practice: mind, mathematics, and culture in everyday life. Cambridge, UK: Cambridge University Press.
Lave, J. (1993). The practice of learning. In: Chaiklin, S. and Lave, J. eds. Understanding practice: perspectives on activity and context. Cambridge, UK: Cambridge University Press.
Lave, J. and Wenger, E. (1991). Situated learning: legitimate peripheral participation. Cambridge, UK: Cambridge Univ Press.
Nunberg, G (1996). Farewell to the information age. REST OF REF NEEDED.
Schiff, L.R., Van House, N.A., and Butler, M.H. (1997). Understanding complex information environments: a social analysis of watershed planning. Submitted to the ACM Digital Libraries Conference 1997.
Suchman, L. (1987) Plans and situated actions. Cambridge, UK: Cambridge University Press.
Suchman, L.A. (1993) Reply to vera and simon's situated action: a symbolic interpretation. Cognitive Science 17, 71-75.
Suchman, L.A. and Trigg, R.A. Artificial intelligences as craftwork. In Chaiklin, S. and Lave, J., eds (1993). Understanding practice: perspectives on activity and context. Cambridge, UK: Cambridge University Press; 144-178.