Effective digital library design is not simply a matter of converting
existing information practices and artifacts to a digital world. Digital
libraries (DLs) support cognitive or knowledge work. Designing effective DLs,
then, requires understanding knowledge work and how DLs not only support but
potentially change it. We must look at the work, its tools and practices, the
people who do the work and the institutions that support it, and the interaction
of all these with the DL. To say that we would understand all of this would be
to aspire to too much in a complex and shifting world; but we can inquire with
interest and reach some understanding that will help the continuing enterprise
of knowing.
Although there has been much discussion about user-centered design of
information systems in general, and digital libraries in particular, a
theoretical or conceptual base for such work has been lacking. In this chapter,
I investigate the uses of irreductionist approaches to social theory, with an
emphasis on situated action, science studies and science and technology studies,
and actor network theory as bases for understanding, first, knowledge work, and
then DLs. This chapter emphasizes three critical characteristics of knowledge
work: it is situated, distributed, and social. That it is situated means that
knowledge work is performed by specific people under specific conditions for
specific purposes. It is distributed because it entails cooperation among people
who don’t know one another, as well as among those who do, across space and
time. In fact, some current approaches to learning (Lave and Wenger, 1991) argue
that it is the community and not the individual that “knows.” Finally, it is social because we work
and learn together, and decide what and whom to believe and rely upon, in
community. Much of what we claim to know comes, not from our own direct
experience, but what others tell us is so, including our knowledge of whom to
believe.
This focus on the nature of knowledge work applies to DLs in two ways.
First, documents and other information artifacts, publishing, and libraries are
critical to knowledge work. The digital library, as a new entry into this array
of artifacts and institutions, is affected by, and potentially changes, existing
processes and relations. Second, DL design, construction, and maintenance are
also a form of knowledge work, involving designers, implementers, and content
providers, as well as users.
This analysis is rooted in an empirical study of the UC Berkeley Digital
Library Project, but it is not specific to any one DL. It addresses concerns
raised by participants and non-participants in this DL about actual and
potential changes in information production and use made possible by DL
technology, not about the specific design and content of the Berkeley DL.
Furthermore, I contend that these findings apply to many areas where information
technology is changing the processes of knowledge work. I argue that these
findings reflect, not just issues specific to one DL, or to DLs alone, but to
the social and material practices of knowledge work and the changes brought
about by the electronic distribution of information. The DL challenges existing
practices of knowledge work, the boundaries of knowledge communities, and
practices of trust and credibility, all of which are central to the creation and
use of knowledge.
The first part of the chapter reports on conversations with actual and
potential users, contributors, and creators of the UC Berkeley DL, and discusses
some of their concerns and problems. The second part begins the development of
an analytical perspective from which to better understand these processes. It is
rooted in situated action, science studies and science and technology studies,
and actor-network theory (ANT). The final section considers the implications of
this analysis for digital libraries.
In this chapter, I argue that the DL is not simply a new technology or
organizational form, but a change in the social and material bases of knowledge
work and the relations among people who use and produce information artifacts
and knowledge. I further argue that it crosses boundaries, from private to
public and across work communities, in ways that are becoming increasingly
common with networked information. Finally, I argue that the DL and its
potential interaction with knowledge work, communities, and practices highlights
critical issues of trust and credibility in the networked world. I then examine
the implications for DL design and evaluation, and for other forms of networked
information.
This paper draws on research conducted as part of the
UC Berkeley Digital Library Project (http://elib.cs.berkeley.edu) (UCB DL), part
of the Digital Libraries Initiative funded by NSF and others. This research was
conducted during Phase I, which developed a work-centered DL supporting
environmental planning. Its premise was that work groups require sophisticated
digital library services and collaborative support to effectively utilize
massive, distributed repositories of multimedia information; and that we need to
understand the work and the work group to build DL services to support
them. During Phase II, which began in 1999, the focus of
the project has shifted; the content from Phase I remains, but has been
expanded. This paper refers to the project as it existed under Phase
I.1 This is only one of many digital library initiatives being
undertaken at UC Berkeley and within the UC system.2 The UCB DL has
developed tools for image retrieval, new document models, Web-based geographical
information systems (GIS), distributed search, and natural language processing
for distributed search. A major component of the project was and remains a
substantial and diverse testbed of textual documents, photo images, data sets,
and various kinds of geo-referenced data. It currently3 holds 165,000
photo images, nearly two million records in geographical and biological
datasets, and 2,500 scanned documents comprising 300,000 pages of text for a
total of over one terabyte of data.
Much of the discussion in this paper refers to one major part of the
testbed, called Calflora,4 a linked collection of botanical datasets
containing descriptive information, photos, and diversity mapping for over 8000
vascular plants in California, and nearly 675,000 records of plant observations
from 15 state, federal, and private databases. These records came in a range of
formats in a variety of places. The size of this database of observations and
its coordination with the other parts of the testbed make this DL significant
for the botanical and environmental planning communities as both a resource and
a model.
Interviews
The premise of the user needs assessment and
evaluation component of the UCB DL is that, to design DLs to support work, we
must first understand that work (Schiff, Van House, and Butler, 1997; Van House,
1995; Van House, Butler, Ogle, and Schiff, 1996; Van House, Butler, and Schiff,
1998). We investigated the practices of environmental planning, particularly
those of information use and production; the information artifacts used and
produced; and the possible effects of a Web-based digital library. More
generally, our goal has been to better understand distributed, collaborative
cognitive work, the role of information and information practices and artifacts,
and the potential effects of digital information.
This chapter reports on a particular aspect of this research, issues
related to the sharing and use of digital data that arose during
interviews5 with two different communities. The purpose of these
interviews was to learn about environmental planning and its use and production
of information and information artifacts. The interviews were not about the UCB
DL per se so much as about the potential of DLs in general, including the
UCB DL.
One set of interviews was with people engaged in work related to water
planning in California. Participants included employees of various state, local,
and federal agencies, private consultants, members of the public, and people
from a range of professions, including engineering, hydrology, biology, computer
science, and planning. In earlier papers drawn from these interviews, we
described the social world of water planning (Schiff, Van House, and Butler,
1997; Van House, 1995; Van House, Butler, and Schiff,
1998).
The second set of interviews was with people whose work was associated
with the botanical datasets included in the UCB DL. We interviewed people in
state and federal agencies, nonprofit research institutions, and within the
University of California. With them we discussed a range of issues related to
current efforts to make biological data available at UC and
elsewhere.
These groups6 share some important characteristics. Neither is
homogeneous with clear boundaries. Both fields consist of multiple scientific
disciplines engaged in coordinated work and sharing data, but with different
knowledge bases, interests, methods, and terminology. Both include people
working in state, federal, and local government and non-profit and commercial
organizations doing research, planning, and applied work. Both have at least
some involvement in decisions about land and water use that are often highly
political7 with potentially great impact on the environment and
economy (e.g., housing development, agriculture, timber harvesting, and
recreation).
Both rely heavily, at least for some purposes, on large datasets of
records of observations and specimens, and summaries and analyses using these
records. The records come in many forms and formats, from a variety of sources,
collected over a long time by people with varied backgrounds, training, and
interests. For example, to retroactively establish a baseline to determine
changes in an ecological community, researchers may search out and correlate
data about a particular geographic area collected over many years by different
observers for a variety of purposes. Fine-grained data, such as the location of
individual plants in a particular area, is often available only from local,
sometimes nonprofessional, sources, while higher levels of government are often
the source of higher-resolution data, such as aerial photos or statewide
resource surveys.
In both fields, computing and telecommunications enable access to large datasets that were previously difficult to share. These linkages have accelerated recently with developments in digital libraries and the Internet.
This section reviews issues raised in the interviews
about the willingness of individuals, organizations, and knowledge communities
to contribute to and use information from the DL, and to participate in the DL’s
creation and maintenance. The interviews revealed concerns about DLs and their
relationship to the situated, distributed, and social nature of knowledge. The
following section presents an analytical perspective on knowledge work that can
help us understand the issues raised in the interviews and DLs’ role in the
shifting processes of knowledge creation and trust.
DL Contributions. Identifying and acquiring content for a DL like the
UCB DL, which does not begin with a pre-existing collection or even a specific
user community, and includes both unpublished and published information,
requires that some information producers agree to have their content included.
It also requires that information be indexed, synthesized, and correlated. For
example, the Hrusa California Plant Synonyms, a comprehensive table tracking
changes in California plant names created by a single researcher and made
available to the DL, is indispensable for coordinating multiple databases of
botanical specimens and observations.
In the interviews, we found a generally high level of interest in sharing
data. In particular, local environmental and government organizations, who often
have difficulty disseminating their work, saw the DL as a way to reach a wider
audience.8 But we also found many concerns, especially about access
to unpublished data.
Previously, much of the data for which respondents were responsible was
available to others mainly in summary form. Raw data were available directly
from the data owners, on site or remotely, or within a professional community,
e.g., through closed ftp. These social and technical barriers limited the data
to people within a professional community, or those who had dealt directly with
the data owner, who could screen requests and explain the data and their limits.
Now such data can be made available to anyone. However, the possibility of open
access raised several concerns among respondents. I will focus on three: fear of
possible misuse of data, especially by people outside the professional
community; the burden of making data available to others; and the hazards of
making visible previously-invisible work.
The first concern is “inappropriate use” of the data. Data available via
the Internet may be used by people with interests, training, or expertise
different from the data’s producers. A widespread concern in water planning
interviews, in particular, was that uninformed users would download and use a
snapshot of a changing dataset, misinterpret the data, combine datasets
inappropriately, run unsuitable analyses, or otherwise misuse the data. For
example, fish populations are measured by netting a sample. Projecting from the
netted sample to a population is complex. The time of day and year, weather
conditions, type of net, depth, and many other conditions affect the catch; and
various models may be used to project a population from a sample.
The next two sets of concerns relate to what one respondent called
“productizing” one’s work. Sharing data requires work: cleaning the data, and
creating documentation and metadata. One respondent said, “The Web...has made a
lot of our scientists realize that they are [information] providers, and take
much more seriously the notion that they have obligations.” Those obligations
may not be welcome to either the researcher or the institution. Creating and
maintaining data is often considered preliminary to the “real” work of research.
Making it available to others takes time, effort, and sometimes different
skills. It may be inconsistent with the researchers’ or organization’s
priorities. And it may primarily serve the needs of people in other
organizations.
“Productizing” one’s work may make previously invisible work visible
(Star and Strauss, 1998). Sometimes this is desirable; giving workers credit for
workload and skill, for example. But it may make a person accountable in new
ways. For respondents, making available the data underlying their analyses may
allow a critic to redo or refute them: as one respondent said, “use our data
against us.” For example, we learned that water supply projections are derived
from a large number of measures and data points collected over space and time.
The process of incorporating the data into models to generate forecasts is not
mechanical. Forecasters exercise considerable judgment about which observations
are accurate, how to massage the data, and which forecasting models and
assumptions to use. Opening up the datasets to scrutiny opens up the entire
process of generating supply projections.
Two findings are worth emphasizing here. One is the potential problems in
crossing boundaries (Marshall, this volume; Star, 1989; Star and Griesemer,
1989): making data available to and from people in other professional or
practice communities, and making public what was private. The distributed nature
of knowledge work means that people are continually sharing work, whether with
people they know or impersonally through publication. But this sharing generally
takes place within social, organizational, and professional boundaries, helping
to ensure appropriate understanding and use and reciprocity. The DL has the
potential of de-contextualizing information and making it more readily available
outside of the social world within and for which it was produced. The other
finding that I want to stress is the need for at least some DLs to actively
solicit people’s participation, to win their trust and to align with their
interests.
Now we turn
to some factors that affect people’s willingness and ability to use DLs. The
literature contains many discussions of factors influencing people’s adoption of
new technology; a DL is not simply a technological artifact, however. It is an
information resource. Nor is it exactly analogous to a traditional library.
Three possible differences between DLs and physical libraries are
particularly relevant to this discussion. First, traditional libraries are
closely tied to the publishing system, Chartier’s (1994) “circuit of the book.”
The publishing system provides review and quality control, giving users some
imperfect but indicative assurances about the sources, credibility, and
appropriate uses of information. However, some DLs may provide access to
previously-unpublished information, placing them outside these established
institutional safeguards.
Second, traditional libraries are established institutions and
librarianship a respected profession with selection policies and procedures,
standards of performance, and a code of objectivity. Some DLs are affiliated
with traditional libraries and/or staffed by professional librarians; many are
not.
Third, a DL contains content, technology, and functionality, all of which
users must be willing to trust if they are to rely on the DL as a source of
information and possibly an aid in analysis, and incorporate the work of others,
carried by the DL, into their own. The UCB DL provides innovative functionality,
the ability to manipulate and recombine data in a variety of ways. For example,
it includes a GIS function, the ability to select, map, and overlay
geo-referenced data from unrelated sources.
Using published sources, data collected by others, or even technology
designed by others is an act of trust. Shapin (1994) demonstrates that the
distributed nature of knowledge work makes it a collective good. Much of what we
claim to know is rooted in what we have been told by others, including what we
have learned about who or what to trust, and how to evaluate the
trust-worthiness of sources. Even in science, which purports to ground truth
claims in direct experience, he shows that knowledge is grounded in the
community’s prior knowledge, its instruments and methods of observation and
argument, and its assessment of competing knowledge claims. “We rely on
others...[T]he relations in which we have and hold knowledge has a moral
character, and the word I use to indicate that moral relation is trust.”
(Shapin, 1994, p. xxv, emphasis in the original.)
Users of any library have to assess the quality and credibility of its
contents. This is an assessment, partly of the library as an institution or
collection (e.g., a medical school library will have different health
information from a public library), and partly of the information resources
themselves. Users often take for granted the trust-worthiness of a library,
based on their knowledge of the institution: a university library, for example,
is assumed to adhere to collection development standards and procedures that, to
some degree, warrant its contents.
Assessing the credibility of information is not easy. For example, a
steady trickle of email arrives questioning or “correcting” identifications of
the UCB DL’s flower images. No doubt some of the original identifications were
erroneous. But assessing the correctness even of something as factual as plant
identifications is complex. Some species can only be positively identified under
a microscope. Other differences may be due to changes in plant taxonomy over
time: names change; the boundaries between species and subspecies move; new
species or subspecies are defined.
How does a user assess the credibility of information? A forest ecologist
took us through her criteria for plant identifications. A record of a sighting
is less credible than a specimen in hand. A report of a species in an unexpected
place is less plausible than one from a place where it is common. She also
considers the expertise of the person making the identification. She knows that
some people are experts in a geographical area; she’ll trust their
identification of species common to that area, but not necessarily rare species.
Others are experts in a species, and she’ll trust them on that species and
subspecies wherever they may occur. In other words, she judges the professional
skills of data providers--directly or indirectly--to determine the credibility
of the data.
We found that knowledgeable users consider, not just the data providers’
expertise, but their interests. The naive ideal of objectivity in data
collection and analysis quickly broke down in politically-charged discussions
about such issues as water planning with people from, say, resource extraction
industries and environmentalists. Many choices are made in collecting,
analyzing, reporting, and interpreting data that reflect particular groups’
interests and expectations (Wood, 1992). People knowledgeable about (and
suspicious of) information providers’ interests factor those in when assessing
credibility.
How might a DL assist users in assessing data quality? First, a DL
generally exercises informed judgment in the selection process. Collection-level
assessments are made of which botanical data to include in the UCB DL. Some in
the user community have argued, however, that every image or record should be
reviewed by an expert. However, this would require considerable skilled labor
and the UCB DL does not do this.
Second, a DL can provide informed users with information about the
source. The UCB DL links each botanical observation to its source and whatever
metadata are provided. This shifts the burden of assessment to the user–which
may not help the inexpert user, but may be invaluable to the expert like our
forest ecologist.
DLs provide not only content but functionality. This includes search
tools: the DL’s interface may act as an aid or a barrier between the user and
the collection. The UCB DL also performs operations on the data, such as GIS
(geographical information system) overlays, where graphical representations of
multiple datasets are laid over the same geographical substrate. In GIS, which
data can be overlaid, the level of resolution, even the choice of colors affects
the interpretation of the results. Wood (1992) shows how maps, which appear to
be simple representations of a territory, reflect interests by the choices of
what is represented and how. Interestingly, these issues about trusting a DL’s
functionality were raised, not by users, by UCB DL designers with expertise in
the applications area, probably because they knew the most about the
possibilities chosen and those missing.
In sum, in understanding knowledge work, we need to be concerned with how
people decide whether to trust others’ work and to incorporate it into their
own, and how the DL supports or undermines these processes. By removing
information from its original setting, the DL may strip it of the contextual
indicators needed to interpret, assess, and use it–the knowledge of sources,
methods, terminology, the assemblage (Watson-Verran and Turnbull, 1995; Van
House, Butler, and Schiff, 1998) of people and practices that marks the work of
a community.
DLs require contributors and users, but
also designers, builders, and operators of many different kinds. The UCB DL,
like many, is both a research project and an active testbed. Many DLs are
collaborative projects among researchers, user communities, librarians, and
technologists. Participation by at least some of these people may be voluntary.
Two major sets of interests to be reconciled in a DL like the UCB DL are
computer science researchers and staff, and content area specialists, who are
both information providers and users. The DL may be something different to each.
Computer scientists, for example, tend to emphasize technological innovation,
whereas users are more interested in content and stable, reliable functioning.
An interviewee from a natural history museum said that he had seen problems on
similar projects “because the computer scientists are interested in blazing
issues in computer science, because that’s what’s going to make them famous, and
we on the content side are trying to shovel mounds of [data] and it doesn’t flow
down well.” He went on to say, “This gets into the political agendas of what
serves any participant in this project best and how we can devise a work plan
for a project that is going to give everybody the right benefits out of it. What
we need to end up with here in the museum is with more stuff digitized and with
systems that enable us to manage, capture, maintain it after the project has run
its course. We have certain limitations in the toolsets that we will be able to
cope with.”
Such differences are not unique to the UCB DL. Weedman (1998) studied the
Sequoia project, a research project that developed computer-based tools for
earth science research. She found differences between the earth science research
community, who wanted a reliable, working system, an incremental improvement
over existing tools, requiring minimal effort on their part, and computer
science researchers who wanted to be at the cutting edge of their field. These
differences are likely to surface in the choices about development priorities,
content, functionality, the speed with which changes are introduced, and
attention to user-friendliness.
Another source of tension is that participating disciplines may have
different norms and styles for decision-making. Many computer science projects,
and many Web-based projects, are highly democratic and decentralized. Whoever is
willing to contribute is welcome. A member of the DL’s technical staff noted
differences between computer science and the botanical community over how much
effort (and delay) to put into making a “right” decision versus getting
something up and running and dealing with problems later. Of course, it is
dangerous and perhaps unfair to generalize about two research communities from
these comments; what is significant is that some key participants believed that
such differences existed. It seems reasonable to expect differences between
participants focused on the technology and those focused on content and
functionality in their assessment what is good enough to make publicly
available. Problems with content and functionality are likely to be more
distressing to members of the user community.
While the most visible differences may be over major decisions, another
issue is articulation work (Star and Strauss, 1998): tuning, adjusting,
monitoring, and managing the consequences of the distributed nature of work, the
interplay between formal and informal. Articulation work is almost always
invisible (especially when it is done well), and because of this is often
overlooked in technological innovation. In traditional libraries, librarians do
the articulation work, whether it is ensuring the integrity of the collection or
helping users locate information when the tools fail. In research DLs, there may
be judicious discussions about major policy decisions, but then a lack of
attention to articulation work.
In the UCB DL document intake process, for example, the work of creating
metadata for documents was initially left to the student assistants doing the
scanning. In time this was seen to be problematic, and a form was created for
document providers to report metadata. However, libraries know well the problems
of inconsistency that come from having metadata assigned by multiple
people.
In sum, then, DLs balance multiple interests. In some DLs, information
providers must be willing to allow their work to be included, to be used,
possibly by people from other knowledge communities whose understanding and
practices may differ from their own. Participants must be willing to use—to
trust--the DL’s contents, which may include documents, data, and other
representations not available through the more traditional channels of
publishing and libraries, and without the warrants that these institutions
provide. The DL provides not simply contents but also functionality: minimally,
to search and filter the DL’s contents, but in some cases to operate on,
recombine, and otherwise manipulate the contents, often in ways that are not
transparent to the users. Finally, the design, implementation, and operation of
the DL often requires the work of different professional groups, such as
computer scientists, librarians, and members of the targeted user communities,
each with different interests and priorities for the DL.
DLs participate in, are affected by, and potentially change the relations
among the people, artifacts, and practices of knowledge work, which is situated,
distributed, and social. The DL itself, furthermore, is an instance of
distributed work that requires balancing the interests and knowledge frameworks
of multiple communities.
To design and evaluate DLs, then, it would be useful to have an
analytical base for understanding issues of knowledge creation, use, and trust;
of crossing of boundaries between public and private, and across knowledge
communities; and of cooperation in the creation of sociotechnical systems such
as DLs.
Up
to this point, this chapter has been concerned with findings from interviews
with participants and potential participants in the UCB DL, with reference to
this DL and to the more general possibilities of sharing digital data. This
section presents a framework for understanding these findings in the context of
knowledge work and DLs as sociotechnical systems. It draws on three related
analytical approaches: situated action/practice theory, based primarily on the
work of Jean Lave and Etienne Wenger, science studies and science and technology
studies; and actor-network theory. In the interests of space, we will
concentrate on the aspects of these approaches, individually and collectively,
that are most relevant to the present discussion. The following section then
applies this approach to the empirical issues described
above.
All three of these approaches are to some degree “irreductionist”
frameworks (Kaghan and Bowker, 2000), according to which social order is not
pre-existing but is continually produced and reproduced through people’s
ongoing, practical action, the concrete, day-to-day activity and interaction.
The emphasis is on the processes by which order is (re)constructed, and on the
practices and artifacts that carry and shape understanding and activity across
time and space.
In the situated action approach of Lave and Wenger (Lave, 1988, 1993;
Lave and Wenger, 1991; Wenger, 1998), learning is a dimension of social
practice. Theories of social practice emphasize the interdependence of the
person and the world. Learning concerns the whole person acting in the world, as
a member of a sociocultural community. “Learning, thinking, and knowing are
relations among people and activity in, with, and arising from the socially and
culturally structured world” (Lave and Wenger, 1991, p. 50). Knowledge is not an
inert substance transferred from teacher to student. It is a complex social and
material phenomenon, a "nexus of relations between the mind at work and the
world in which it works" (Lave, 1988, p. 1).
Practice–concrete, daily activity–is critical to knowing. Information
artifacts, including texts and images, are not simply reflections or carriers of
knowledge. They shape and reflect practice, and are instrumental in creating and
re-creating knowledge, as well as coordinating work across space and time.
Knowledge, practices, and artifacts are tightly bound up with the
knowledge community or community of practice (Lave & Wenger, 1991; Wenger,
1998). People learn and work within groups who share understanding, practices,
technology, artifacts, and language. Communities of practice include
professions, workgroups, and disciplines, as well as less-easily-labeled groups.
The recognition of trustworthy sources is a necessary component of all
systems of knowledge and knowledge communities (Shapin, 1994). Our perceptions
are located within “hierarchies of credibility” (Star, 1995, citing Becker). An
important task of knowledge communities is establishing these hierarchies:
deciding what is known, the processes and principles by which knowledge claims
are evaluated, who is entitled to participate in the discussion, and whom to
believe.
In our society, science is considered the prototype of rational knowledge
construction and validation. Science studies9 and science and
technology studies (STS) investigate the culture and practices of science and
technology, and how it is decided what is known, including how hierarchies of
credibility are created and participants attain their places in them. The
approaches subsumed by science studies and STS vary, but they share a contention
that scientific and technical knowledge is determined, not entirely by nature,
but also by social influences. This does not mean that knowledge is simply a
product of social agreement. But this approach challenges the traditional
boundaries between nature and society, science and politics, and
knowledge-creation practices in science and in other areas of
knowledge.
STS is further concerned with how sociotechnical systems–systems of
people, technology, and practices–are created and maintained. It takes the
stability of such systems as something to be explained rather than taken as
given.
Actor-Network Theory (ANT)10 originated within science studies
and STS with ethnographic studies of science laboratories and the processes by
which reputations were built and resources garnered via the practical, daily
work of the laboratory (e.g., Latour and Woolgar, 1991). ANT sees power and
order as effects to be explained. ANT argues that action at a distance and
mobilization of allies are critical to the development of scientific knowledge,
establishment of credibility, and acquisition of resources, and that
inscriptions (including texts, images, graphics, and the like) play a key role.
It has also been concerned with explaining the stabilization of sociotechnical
systems, which it describes as temporary, changing heterogeneous networks of
resources
ANT has its shortcomings.11 Our purpose here is not to “do”
ANT but to see how it may help us to understand DLs as sociotechnical systems,
and as collections of inscriptions used to act at a distance. To better
understand ANT and how it may apply to DLs, and to our interview findings, we
need to go through some of its elements.
The basic ontological unit of ANT12 is the actor-network,
which is “most simply defined as any collection of human, non-human, and
hybrid human/non-human actors who jointly participate in some organized (and
identifiable) collective activity in some fashion for some period of time”
(Kaghan and Bowker, 2000). ANT is concerned with how these pieces are held
together, as organizations, social institutions, machines, and agents, at least
for a time. Perhaps the most radical contribution of ANT is the inclusion of the
non-human: according to ANT, networks are heterogeneous, composed not only of
people but of machines, animals, texts, money, and other elements.
Translation is a key process by which actor-networks are created
and stabilized, however temporarily. Actors’ disparate interests get translated
into a set of interests that coincide in the network. For example, in Callon’s
(1986b) study of the development of electric vehicles, the producers of vehicle
bodies and of fuel cells both had to see their interests as being served by the
electric vehicle to be willing to research and produce the needed components.
Without their involvement, the vehicle could not be build.
Black-boxing is a process of closing questions and debates.
Participants agree to accept something as given. Ideas, knowledge, and processes
can all be black-boxed, at least temporarily, at least for some people.
“Standard” research methods, for example, get black-boxed within research
communities. What is black-boxed for some groups may not be for others, e.g.,
different research communities may have different standard
methods.
Finally, “an intermediary is an actor (of any type [i.e. human or
non-human]) that stands at a place in the network between two other actors and
serves to translate between the actors in such a way that their interaction can
be more effectively coordinated, controlled, or otherwise articulated” (Kaghan
and Bowker, 2000). Because networks are never completely stabilized, translation
and black-boxing are continual. This is the work of intermediaries.
Inscriptions are an important kind of intermediary in ANT.
Knowledge, according to ANT, is not abstract and mental. It takes material form:
inscriptions such as journal articles and patents, conference presentations, and
skills embodied in scientists and technicians. “The actor-network approach is
thus a theory of agency, a theory of knowledge, and a theory of machines” (Law,
1992, p. 389). Inscriptions play an important role in scientific and other
regimes of knowledge. Modern information technology, beginning with the printing
press, has made possible the immutability and mobility of inscriptions, allowing
people to send their work and arguments across space and time, and to gather,
compare, combine, contrast, summarize, and refute others’
work.
Another kind of intermediary of interest in the present discussion is the
boundary object. Star and Griesemer (Star, 1989; Star and Greisemer, 1989) noted
that scientific work has always required information that can be used by
multiple users and communities for a variety of purposes, retaining its
integrity across space, time, and local contingencies without losing its
specific meaning in a local setting. To explain how this works, they propose a
model based on symbolic interactionism’s “social worlds” theory. Instead of a
single actor trying to funnel others’ concerns into a narrow passage point, they
argue for multiple translations, entrepreneurs from multiple, intersecting
communities of practice, each trying to map their interests to those of the
other audiences in such a way as to ensure the centrality of their own
interests. One place these interests come together is in a boundary object.
They describe boundary objects as both plastic enough to adapt to local
needs and have different specific identities in different communities, and
robust enough to maintain a common identity across sites, and be a locus of
shared work. They identify several types of boundary objects, one of which is of
particular of interest for the present discussion: repositories, which they
describe as “ordered piles of objects indexed in standardized fashion” (Star and
Greisemer, 1989, p. 408). Their example is the University of California,
Berkeley, Museum of Vertebrate Zoology (MVZ), a collection of specimens of
amphibians, birds, mammals, and reptiles, with extensive, standardized metadata,
including such information as descriptions of the location and conditions under
which specimens were collected and by whom. (The MVZ’s specimen records,
coincidentally, are now accessible through the UCB DL.) Star and Greisemer’s account of the
creation of the MVZ shows how the interests of the director, major funder,
collectors, trappers, and university administrators came together to create an
impressive array of specimens of California flora and fauna with standardized
descriptive information, to be used by a variety of scientific disciplines for
many purposes.
Situated learning has profound implications for the design of information
systems. Our ideas about knowledge creation and use and information transfer are
rooted in our assumptions about learning. The very phrase “information transfer”
implies that information is a thing that can be transferred. If learning is
rooted in practices, artifacts, and communities, then information systems have
to be co-constituted with these, as well. If knowledge is tightly bound up with
a community of practice, then information systems have to be aligned with
communities of practice. At the very least, we have to look more carefully at
the social aspects of knowledge and the role of information artifacts such as
documents in practice and in communities of practice. In particular, we have to
consider how DLs both support the recognition of trustworthy sources, and are
evaluated as trustworthy themselves.
Science studies, STS, and ANT are relevant to the UC Berkeley DL and DLs
in general in a number of ways. First, much of the content of the UCB DL is
scientific; most of the interviewees do work that is to some degree based in
science. The practices of scientific work and of credibility are relevant to
interviewees’ concerns about sharing of data.
Second, as we have said, science is often considered the prototypical
system of knowledge. Furthermore, ethnographic studies of science have concluded
that what scientists do is not fundamentally different from ordinary activity
(Latour, 1987; Latour and Woolgar, 1991). I contend that much of what has been
learned in science studies probably applies to other knowledge communities and
to digital libraries and other digital information generally. In particular, the
recognition of trustworthy sources is a universal issue.
Third, STS and ANT are concerned with the building and maintenance of
sociotechnical systems; DLs are sociotechnical systems. The UC Berkeley DL
consists of technology, texts, images, databases, functionality, users,
contributors, funders, designers, builders, operators, and more. ANT may provide
us with resources for understanding people’s willingness (or unwillingness) to
use and contribute to the DL and its inscriptions, and to participate in the
DL’s creation.
Fourth, ANT addresses the role of inscriptions in translation and action
at a distance, the establishment of credibility and order, and power (e.g.,
Latour, 1986; Law, 1986). DLs substantially affect the creation, circulation,
and combination of inscriptions.
Finally, I contend that DLs are boundary objects. The UCB DL is an
ordered, indexed repository of publications, images, and records of specimens
and observations, of catalogs, indexes, maps, and mapping functionality. It has
different identities in different communities (e.g., a repository of documents
and records; a testbed for technology development), but a common identity across
sites (the UCB DL). And it is the locus of multiple, simultaneous translations,
as the different communities find in it their interests, and seek to enroll
others to help create and maintain the DL in ways that support their
interests.
Now let us look again at participants’ concerns about
contributing content to a DL, using it, and participating in its creation and
maintenance in the light of the discussion so far.
The DL is a locus of shared work of two kinds. First, building and
maintaining a DL involves a variety of groups who contribute content, do the
work of design and operation, and use it. Second, the content of a DL supports
the shared work of information providers and users. Botanical databases, for
example, are used across communities doing botanical and other kinds of
environmental work. The DL is the “host,” so to speak, of other boundary objects
such as databases.
The power of the DL is in the integration of its heterogeneous elements,
that is, its nature as a sociotechnical system, its usefulness to multiple
professional communities. This integration, however, creates difficulties and
conflicts as each participating group places its own concerns at the center and
seeks, in ANT terms, to enroll others – or at least participates only as long as
its interests are met, as long is it is willing to be enrolled. ANT warns us
that network stability must be explained, not assumed; that networks tend to
come apart. Some of the tension that shows up in the interviews--including
differing priorities between technologists and content-area specialists, and
concerns about the effort required of data providers--are a reflection of this
tendency toward instability. They reflect, not problems to be solved once and
for all, but expected and on-going tensions in the DL as a dynamic locus of
multiple translations.
The flexibility created by DLs is a major reason for building them. But
this same flexibility may destabilize the processes and standards of knowledge
creation. In science and technology, as in other areas, knowledge work is
performed within an interdependent network of previous work, methods, tools, and
colleagues – assemblages of people, practices, tools, theories, standards,
publications, social strategies, and the like that are dynamically, mutually
constituted in and by knowledge work (Watson-Verran and Turnbull, 1995; Van
House, Butler, and Schiff, 1998). Inscriptions play a key role in the creation
of knowledge, and the processes by which knowledge communities decide what and
whom to believe, interests are translated, and networks are stabilized.
Hierarchies of credibility are created within communities of practice; there too
inscriptions are most easily understood (see, for example, the discussion of
genres in Agre, 1998). But DLs enhance the ease with which inscriptions cross
community boundaries, making Latour’s “immutable mobiles” even more
mobile–perhaps too mobile, when respondents worry about their data being misused
by people without the needed professional qualifications.
Crossing the boundaries of knowledge communities may call into question
that which is taken for granted within communities, opening black boxes, and
challenging the established cognitive order. Hence the concerns in the
interviews about misuse of data or the inclusion of data from unfamiliar
sources. Shared practices and understandings can no longer be assumed; the
indicators of credibility may be questioned or missing; users may not know how
to assess credibility; inscriptions may be used in ways not envisioned (or
accepted) by their creators. The DL potentially destabilizes the processes and
the hierarchies of credibility and power relations.
Furthermore, the DL can challenge inscriptions’ immutability (Levy, this
volume). The DL is not simply a repository of inscriptions, a passive container
or conduit; it makes them more manipulable and combinable; it even makes new
kinds of inscriptions possible. The UCB DL, for example, allows the user to take
a published table and re-organize it; for example, a table of dams arranged
alphabetically can be re-ordered by size. The concerns about data being misused,
for example, include fears that the “black box” of an inscription such as water
supply forecasts will be opened, the modeling replicated with different
assumptions, and the forecasts challenged.
Of course, the originator never retains control over his or her
inscriptions. But the DL allows them to move even more freely; it may enable
more fluid re-combinations of them; it may even open them up in ways not
possible before.
The DL opens some black boxes but it may close others. For example, its
analytical and search tools may be opaque to the user, making it difficult for
the user to understand, evaluate, and trust what the DL is doing.
In sum, then, the DL is not simply a passive repository of inscriptions.
By potentially changing the mobility and mutability of inscriptions, it may
challenge the established order and de-stabilize the network of relations. It
becomes a participant in the processes of knowledge creation and the
establishment of trust and credibility.
In
this chapter, I have been concerned with better understanding knowledge work,
its social and material practices, and their interaction with the digital
library. I argue that electronic distribution of information potentially
challenges existing practices of knowledge work, boundaries of knowledge
communities, and practices of trust and credibility, all of which are central to
the creation and use of knowledge and the stabilization of the cognitive
order.
I have emphasized three characteristics of knowledge work and their
implications for DLs. Because knowledge work is situated, DLs have to articulate
with the contexts and practices of specific knowledge communities. Because it is
distributed, DLs support the coordination of work across space and time. And
because it is social, they are entangled in the processes by which communities
collectively decide what they know and whom to believe.
This focus on knowledge work applies to DLs in two ways. First, digital
libraries support users’ knowledge work, so understanding knowledge work will
help us to ensure that DLs are useful and used. Second, DL design, construction,
and maintenance are also a form of knowledge work, so this understanding may
help to ensure that DLs are sustainable (Lynch, this
volume).
The DL is not simply a new technology or organizational form, but a
change in the bases of knowledge work and its network of social and material
relations. The DL may cross boundaries, from private to public and across
knowledge communities, in ways that are becoming increasingly common with
networked information. Finally, the DL and its potential interaction with
knowledge work, communities, and practices highlight critical issues of trust
and credibility in the networked world.
A digital library is a heterogeneous network of users, researchers,
funders, operators, and other people; of documents, images, databases, thesauri,
and other information artifacts; of practices and understandings; and of
technology. It is a boundary object, both created and used by different
communities for different purposes. It is the locus of multiple translations as
various participants try to enroll others to ensure that the DL meets their
needs. It is an active participant in the creation and circulation of documents,
images, and other kinds of inscriptions.
This understanding of DLs has implications for DL design, management, and
evaluation, and for our understanding of knowledge work and communities. First,
it suggests that the work of translation is on-going in DLs. The problems and
tensions identified in these interviews are indicative of the dynamic tension
among the DL’s participants. We cannot expect a DL to settle into an easy
equilibrium of interests. We can expect conflicts, compromises, and jockeying
for advantage among various stakeholder groups to be a continual part of
DLs.
Second, it cautions us to be sensitive to the variety of communities’
existing practices of knowledge creation and work and indicators of credibility.
A successful DL has to fit with these practices. In particular, the DL has to
articulate with participants’ hierarchies of credibility and processes of
establishing trustability for people to be willing to use and contribute to the
DL. And these vary across communities of practice.
The
world and the work that the DL serves are continually changing; so too must the
DL. Its design needs to be deliberately fluid and dynamic, to accommodate
emergent work practices, and on-going enrollment and
co-constitution.
There
is currently much discussion about involving users in the design of technology,
but how (and whether) it is done varies. This analysis suggests that high-order
user involvement in DL design is critical. Only members of the knowledge
community fully understand the complexity of their network of practices, tools,
and participants, including their processes and criteria of credibility, and how
the DL might interact with them. However, the question of who is to be involved,
which communities, can become even more uncertain–if both the value and the
threat of DLs is in their ability to cross the boundaries of communities of
practice, which communities are involved and how are their differences
reconciled?
Understanding the DL as a boundary object helps us understand some of the
on-going tension between generality and specificity in DL design. But it also
helps us to see DLs as continuous with past boundary spanning practices in
science and other areas. A DL designed for specific communities and tasks can
mesh with the methods, genres, understandings, and language of a specific
communit –and support existing and emergent practices of trust and credibility.
But a major value of DLs is their ability to cross communities and tasks.
Furthermore, customization is costly. For whom will the DL be customized? Which
group’s interests will dictate design decisions?
The
implications for DL evaluation are several. The DL needs to be evaluated, not
just for how well it performs its intended functions and meets targeted users’
identified needs, but for its interaction with work, practices, artifacts, and
communities. Evaluation needs to be targeted to specific user communities and
tasks, with the understanding that DLs are different things to different groups.
This makes evaluation more complex and uncertain, but also more realistically
aligned with users’ concerns and with the changing array of users and other
stakeholders.
Evaluation needs to address, not just service delivery, but also the
organizational issues around the creation and maintenance of the DL, the
stabilization of the DL as a heterogeneous network, and the on-going enrollment
and coordination of resources and participants. A DL without users is a failure,
however good its technology and its contents.
Common to science studies, STS, and ANT is a reliance on ethnographic
research methods, the study of activities in their natural settings, of what
people actually do as well as their own accounts of their behavior. The
appropriate methods of evaluating DLs are varied (Van House and Bishop, this
volume), but ethnography is certainly central.
I
argue that these findings reflect, not just issues specific to one DL, or to DLs
alone, but to the social and material practices of knowledge work and the
changes brought about by the electronic distribution of information. DLs
challenge existing practices of knowledge work and boundaries of knowledge
communities, both of which support the interpretation and understanding of
information and the assessment its trustability. DLs raise questions about much
that we take for granted in knowledge work. Taking these questions seriously
will help us to better understand these communities and their work,and their
relationship to artifacts and technology, and to design better
DLs.
Two
areas to which this analysis might be extended are knowledge management and the
growing problem of trustability of Internet-based resources. Knowledge
management is concerned with making better use of organizational knowledge
resources. It often tries to de-contextualize knowledge and information
artifacts, “re-purposing” them, to use one of the uglier neologisms. But
potential users need to be able to understand the information’s context to
assess its credibility and appropriateness for their needs. Similary, the
Internet has made a vast amount of information of uncertain provenance
accessible to people who may lack the expertise to evaluate it, medical
information being a prominent example. Both are areas where information is
crossing boundaries in ways that challenge its interpretation and assessment,
and where attention to existing practices of information activity and cognitive
order, and especially of trust and credibility, may help us to design better
information systems.
Agre, Phil. 1998. Designing Genres for New Media:
Social, Economic, and Political Contexts. In Jones, S. Ed, Cybersociety 2.0:
Revisiting CMC and Community. Newberry Park, Ca: Sage.
Biaggioli, M., Ed. 1999. The Science Studies
Reader. New York: Routledge.
Bourdieu. P. 1977. Outline of a Theory of
Practice,(R. Nice, Trans.). Cambridge: Cambridge University
Press.
Callon, M. 1986a. Some Elements of a Sociology of
Translation: Domestication of the Scallops and the Fishermen of St. Brieuc Bay.
In J. Law, Ed., Power, Action, and Belief: a New Sociology of Knowledge?
(pp. 196-233). London: Routledge and Kegan Paul.
Callon , M. 1986b. The Sociology of an Actor-network:
the Case of the Electric Vehicle. In M. Callon, J. Law, and A. Rip, Eds.,
Mapping the Dynamics of Science and Technology: Sociology of Science in the
Real World (pp. 19-34) London: Macmillan Press.
Callon, M., J. Law, and A. Rip, Eds. 1986a.
Mapping the dynamics of science and technology: Sociology of science in the
real world. London: Macmillan Press.
Callon, M., J. Law, and A. Rip. 1986b. How to Study
the Force of Science. In Callon, M., J. Law, and A. Rip, Eds. Mapping the
Dynamics of Science and Technology: Sociology of Science in the Real World
(pp. 3-18). London: Macmillan Press.
Chartier, R. 1994. The Order of Books: Readers,
Authors, and Libraries in Europe Between the Fourteenth and Eighteenth
Centuries( L. G. Cochrane, Trans.). Stanford, CA: Stanford University
Press.
Garfinkel, H. 1967. Studies in
Ethnomethodology. Englewood Cliffs, NJ: Prentice-Hall.
Giddens, A. 1984. The Constitution of Society.
Berkeley: Univ of Calif Press.
Haraway, D. 1997. Modest_witness@second_millenium.
©Femaleman_meets_oncomouse™: Feminism and Technoscience. NY:
Routledge.
Hess, D. J. 1997. Science studies: an advanced
introduction. New York: New York University Press.
Kaghan, W.N., and G. C. Bowker. 2000. Crossing
Boundaries and Building Bridges: Irreductionist ‘Frameworks’ for the Study of
Sociotechnical Systems. The Journal of Engineering and Technology Management,
forthcoming.
Latour, B. 1986. Visualization and Cognition:
Thinking with Eyes and Hands. Knowledge and Society, 6,
1-40.
Latour, B. 1987. Science in Action. Cambridge,
MA: Harvard University Press.
Latour, B. and S. Woolgar. 1991. Laboratory Life:
The Construction of Scientific Facts. Princeton, NJ, Princeton University
Press.
Latour, B. 1999. On Recalling ANT. In J. Law and J.
Hassard, Eds. Actor network theory and after. Oxford: Blackwell
Publishers.
Lave, J. 1988. Cognition in Practice: Mind,
Mathematics, and Culture in Everyday Life. Cambridge: Cambridge University
Press.
Lave, J. 1993. The Practice of Learning. In Seth
Chaiklin and Jean Lave, Eds., Understanding Practice: Perspectives on
Activity and Context. Cambridge: Cambridge University
Press.
Lave, J. and J. Wenger. 1991. Situated Learning:
Legitimate Peripheral Participation. Cambridge: Cambridge University
Press.
Law, J. 1986. The Heterogeneity of Texts. In M.
Callon, J. Law, and A. Rip, Eds., Mapping the Dynamics of Science and
Technology: Sociology of Science in the Real World (pp. 67-83) London:
Macmillan Press.,
Law, J. 1990. Technology and Heterogeneous
Engineering: the Case of Portuguese Expansion. In W. Bijker, and Others,
Eds., The Social Construction of
Technological Systems: New Directions in the Sociology and History of
Technology (pp. 111-134). Cambridge, MA: MIT Press.
Law, J. 1992. Notes on the theory of the
actor-network: ordering, strategy, and heterogenity. Systems Practice, 5,
379-393.
Law , J., and J. Hassard, Eds. 1999. Actor Network
Theory and After. London: Blackwell Publishers.
Livingston, E. 1987. Making Sense of
Ethnomethodology. New York: Routledge & Kegan
Paul.
Lynch, C. Colliding with the Real Wold: Heresies and
Unexplored Questions about Audience, Economics, and Control of Digital
Libraries. This volume.
Ormrod, S. 1995. Feminist Sociology and Methodology:
Leaky Black Boxes in Gender/technology Relations. In K. Grint and R. Gill,
the Gender-technology Relation: Contemporary Theory and Research (pp.
31-78.) Taylor and Francis
Pickering, A. 1992. From Science as Knowledge to
Science as Culture. In A. Pickering, Ed., Science as Practice and Culture
(pp. 1-26). Chicago: University of Chicago Press.
Schiff, L., N. Van House, and M. Butler. 1997.
Understanding Complex Information Environments: a Social Analysis of Watershed
Planning. Digital Libraries ‘97: Proceedings of the ACM Digital Libraries
Conference (pp. 161-186). New
YorK: ACM Press.
Shapin, S. 1995. Here and Everywhere: Sociology of
Scientific Knowledge. Annual Review of Sociology, 21,
289-321.
Shapin, S. 1994. A Social History of Truth.
Chicago: University of Chicago Press.
Star, S. L. 1989. The Structure of Ill-structured
Solutions: Boundary Objects and Heterogeneous Distributed Problem Solving. In L.
Gasser and M. Huhns, Eds., Distributed Artificial Intelligence, 2 (pp.
37-54). Pitman Publishing, .
Star, S. L., and J. Griesemer. 1989. Institutional
Ecology, Translations, and Boundary Objects: Amateurs and Professionals in
Berkeley’s Museum of Vertebrate Zoology, 1907-39. Social Studies of Science,
1, 387-420.
Star, S.L. and K. Ruhleder. 1996. Steps Toward an
Ecology of Infrastructure: Borderlands of Design and Access for Large
Information Spaces. Information Systems Research, 7, 111-134.
Star, S. L. 1995. Introduction. In S. L. Star, Ed.,
Ecologies of Knowledge: Work and Politics in Science and Technology (pp.
1-35). State University of New York Press.
Star , S.L. and A. Strauss. 1999. Layers of Silence,
Arenas of Voice: The Ecology of Visible and Invisible Work. Computer
Supported Cooperative Work 8, 9–30.
Van House, N. and A. P. Bishop. (2001). Introduction:
Digital Libraries as Sociotechnical Systems. This volume. INTRO TITLE MAY
CHANGE.
Van House, N., M. Butler, and L. Schiff. 1998.
Cooperative Knowledge Work and Practices of Trust: Sharing Environmental
Planning Data Sets. CSCW '98: The ACM Conference On Computer Supported
Cooperative Work (pp. 335-343).
Van House, N., M. Butler, V. Ogle, and L. Schiff.
1996. User-centered Iterative Design for Digital Libraries: The Cypress
Experience. DLIB Magazine.
http://www.dlib.org/dlib/february96/02vanhouse.html
Van House, N. 1995. User Needs Assessment and
Evaluation for the UC Berkeley Electronic Environmental Library Project.
Digital Libraries '95: The Second International Conference on the Theory and
Practice of Digital Libraries.
http://csdl.tamu.edu/DL95/papers/vanhouse/vanhouse.html
Watson-Verran, H. and D. Turnbull. 1995. Science and
Other Indigenous Knowledge Systems. In S. Jasanoff, et al, Eds., Handbook of
Science and Technology Studies (pp.115-139). Newbery Park, CA:
Sage.
Weedman, J. 1998. The Structure of Incentive: Design
and Client Roles in Application-oriented Research. Science, Technology, and
Human Values, 23, 315-354.
Wood, D. 1992. The Power of Maps. New York:
Guilford Press.
NOTES
1. Phase I funding for the UCB DL came from the
NSF/NASA,/ARPA Digital Libraries Initiative. Currently, the UC Berkeley Digital
Library Project is part of the Digital Libraries Initiative sponsored by the
National Science Foundation and many others. Additional funding at Berkeley currently
comes from the CNRI-sponsored D-Lib Test Suite, and the NSF-sponsored National
Partnership for Advanced Computational Infrastructure (NPACI) . The work described in this paper was
supported in part as part of the
NSF/NASA/ARPA Digital Library
Initiative under NSF IRI 94-11334.
2. In
particular, this project should not be confused with the California Digital
Library (http://www.cdlib.org), an initiative of the University of California
Office of the President.
3. As of
August, 2000.
4. This collection now has a quasi-separate identity
as Calfora: http://www/calflora.org.
It consists of:
·
CalFlora Species
Database: information about the 8375 currently recognized vascular plants in
California, including scientific and common names, synonymy, distribution from
literature sources, legal status, wetland codes, habitat info, and more;
includes a photo of the plant, if available, from the CalPhotos California
Plants & Habitats Collection.
·
California Plant Synonymy Table
(Experimental): translates older scientific names to those currently recognized,
and finds synonyms for names in current use.
·
CalPhotos California
Plants & Habitats Photographs: 20,000 images of California plants and
fungi linked by taxonomic name to
the CalFlora Names database.
·
CalFlora Occurrence Database: Over
660,000 records of plant observations from 15 state, federal, and private
databases.
·
GIS
Viewer: displays observation point data in conjunction with USGS topographical
maps, geopolitical maps, relief maps, false-color satellite images, political
borders, and aerial ortho-photographs.
5. Interviews were conducted by the author, Mark
Butler, and Lisa Schiff.
6. For
purposes of discussion, I will call these the “water” and “botanical”
groups.
7. E.g., a recent front-page newspaper article
headlined “New Clash Surfaces in Water War,” began as follows: “In the same
‘water’s for fightin’ ‘ spirit that has defined power and politics in
California, East Bay Municipal Utility district directors rejected a proposal
Tuesday that might have ended a long, bitter water war between Sacramento and
the Bay Area.” (Oakland
Tribune, June 23, 1999, p. 1)
8. Others are interested in using the DL for
publishing, as well. A recent
email, for example, offered the DL the sender’s unpublished paper which s/he
described as “the most important and precious document in the world today. I am sure that my paper bring a
revolution in the field of water and soil management on the Earth which is the
fundamental requirement of life on the planet.”
9. Authors like Biaggioli (1999) and Hess (1997)
carefully sidestep the definition of science studies. Even the choice of a phrase with which
to refer to this broad and diverse area of research is problematic, with
important distinctions made among science studies, science and technology
studies (STS), sociology of scientific knowledge (SSKP), technoscience studies,
and other variants.
10. ANT is most identified with the work of Latour,
Callon, and Law (Callon, 1986a,
1986b; Callon, Law, and Rip, 1986; Latour, 1987; Latour and Woolgar, 1991; Law,
1986, 1990, 1992; Law and Hassard, 1999).
11. ANT has been criticized for, among other things,
being too agnostic about social formations such as power and gender, paying too
much attention to design and development of sociotechnical systems and leaving
out use and users, failing to consider why some networks are more enduring than
others, and for its Machiavellian view of the world and often-warlike language (
Ormrod, 1995; Haraway, 1997).
12. The discussion that follows draws heavily on
Kaghan and Bowker’s (1999) cogent presentation