Intranets Experiences at Xerox.rtf by censhunay


									                                            *** DRAFT ***

                       Intranets: Experiences at Xerox
                                    Gail L. Rein and Daniel L. McCue
                                            Xerox Corporation

            This paper is an in-depth examination and analysis of one Fortune 500
            company’s experiences with the intranet (use of the Internet internally to
            support organizational work). Through an analysis of a sampling of the
            company’s successful and less successful intranet applications, the authors
            explore why certain types of applications are a natural fit for the
            technology while others are more challenging. Their findings suggest that,
            for the less successful applications, insufficient attention has been paid to
            work processes and that important functions provided in paper-based
            environments are missing from current Web-based environments.
            Document management functions are found to be the most significant
            element missing in off-the-shelf Internet software.

Internet technologies are growing at a rate unprecedented for computer-based information
technology. In 1996, Zona Research projected Internet-related revenue would double every year
through 1997 and then grow about 30-50% through 1999. Company intranets—the use of the
Internet to support an organization’s internal information needs—are expected to be the fastest
growing component of that market.1 Yet, despite its impressive growth and apparent success,
immaturity of Internet technology may be the most significant factor preventing even wider
spread usage and adoption of the technology.2 This paper is an in-depth look at the use of the
intranet in one Fortune 500 company, Xerox Corporation, to better understand the technology’s

    1 Consistent with this expectation, Netscape reported 1996 revenue that was more than quadruple a
      year earlier ($100 million compared to $23.3 million). Furthermore, the company attributed this
      growth to their line of intranet products. Source: The Wall Street Journal, 10/23/96, p. B7.
    2 This finding was reported in a 1996 survey of advanced technology conducted by Systems
      Development Incorporated. Of the 76 information technology executives surveyed, 37.5% said
      technology immaturity was the major factor preventing use of the Internet in their companies, 18.7%
      said technology benefits were not demonstrated, 12.5% cited lack of awareness of the technology in
      their organizations, 6.3% said their organizations were not prepared, and 25% cited other reasons.
      Source: PC Week, 10/7/96, p. ?.

Draft 8                                       8/9/2012 6:11 PM                                     Page 1 of 15
Rein & McCue                                                   Intranets: Experiences at Xerox
                                          *** DRAFT ***

Xerox, like so many other companies, is rapidly expanding its internal Web. The Xerox Intranet,
with more than 200 Internet servers, supports over 20,000 employees with Web browsers. These
numbers have grown exponentially over the past year and are expected to continue to grow
rapidly for the next several years. What is of particular interest in this paper is that Xerox’s
internal Web, as an environment for supporting organizational work, is falling short of
expectations and hopes in significant ways. We suspect that our experiences are not unique, and
therefore, in the interest of accelerating others’ learning, we share these experiences and our
analysis of them.
This paper is not a proper case study [9], but rather a reflective account by two Xerox Intranet
users (the authors). We address the following questions by examining and analyzing a sampling
of successful and less successful uses of the Xerox Intranet:
             What types of applications are a natural fit?
             What types of applications are more challenging? Why?
             What is missing in today’s off-the-shelf Internet software?
The paper is organized into five major sections: Terminology explains our use of the terms Web
and intranet; About Xerox provides relevant background on the company; Xerox Intranet
Applications describes and analyzes the successful and less successful applications; Root Cause
Analysis explores the differences in the affordances, or innate opportunities, of paper-based and
Web-based environments to better understand what underlies the problems of the less successful
applications; Conclusion summarizes what we learned, suggests directions for future
development of general-purpose Internet software, and outlines some open questions.

Before proceeding, let us clarify some terminology. The papers in this special issue of the
Communications of the ACM focus on the impact of the World-Wide Web on the way we work.
Since this paper examines intranets, we need to explain our use of the terms Web and intranet.
From a technical perspective, the Internet is the set of protocols and formats defined by the
Internet Engineering Task Force (IETF) and the products and services based on these protocols
and formats. The term intranet refers to the use of the Internet within an organization to support
its internal information needs.
Technically, the Web is a subset of Internet-related standards, products, and services such as
HyperText Transport Protocol (HTTP), HyperText Markup Language (HTML), Common
Gateway Interface (CGI), and Universal Resource Locator (URL). The term Web has three
common connotations: (1) the world-wide physical network of servers and gateways, etc., (2) the
actual information or content contained in the distributed world-wide web of links and nodes,
and (3) the Internet technology—an Internet operating system or computing environment with its
tools—that is applied to (1) and (2). Another articulation of (3) views the Web as an interface,
most often graphical, to the Internet.

Draft 8                                     8/9/2012 6:11 PM                          Page 2 of 15
Rein & McCue                                                          Intranets: Experiences at Xerox
                                            *** DRAFT ***

This paper relies heavily on the second and third connotations. To us, the distinctions between
the Internet/intranet and the Web are blurring; consequently, we use these terms more or less
interchangeably throughout the paper.

About Xerox
Xerox, headquartered in Stamford, Connecticut, USA, is a global company with offices in every
continent except Antarctica. It provides products and services for production publishing, color
copying and printing, electronic printing, and office network systems to the global document
processing market. Of its nearly 87,000 employees, more than half (50,714) are located in the
United States where the major facilities are in New York (Rochester) and California (El Segundo
and Palo Alto). The U.S. work force is 13.3% managers, 22.4% professionals, 24.8%
technicians, 7.6% sales workers, 22.6% office and clerical workers, and 9.3% union workers.
Xerox’s organizational structure is typical of many large corporations. The major high-level
organizational units include: three large business groups, each one directed towards a specific
market segment; a central customer operations unit; a central manufacturing services unit; and
corporate units for research, services, and finance. Xerox operates as a team-based organization.
Much work gets done by geographically distributed teams—both short-lived task forces and
more permanent, long-lived research and development teams—with team members often coming
from many organizational units.
In the mid-1990’s, Xerox moved to a commercial, off-the-shelf PC-based computing
environment for its internal office computing platform. UNIX and Sun workstations are still
heavily used by some research and development groups, but almost all UNIX users also have a
PC running Microsoft Windows in their offices. The standard set of software used for office
computing consists of Microsoft Office, e-mail (MS Mail or Eudora), and a Web browser
(Netscape Navigator or Windows Explorer).3 This move to a PC-based computing environment
is a dramatic change for Xerox, where a state-of-the-art proprietary computing system was used
with great success in the past.
The graphs in Figure 1 give a reasonable approximation to the growth rate of the Xerox Intranet.
The actual numbers in the graphs are slight under-estimates of the total intranet traffic volume
because data was collected only for the major internal gateways. Nonetheless, the graphs do
show the Xerox Intranet is undergoing a continuous rapid expansion. In the six-month period
January 1996 to June 1996, there was a 33% increase in the number of clients, a 150% increase
in the number of intranet files, and a 100% increase in gigabytes.

    3 UNIX is a trademark of UNIX Systems Laboratories. Sun is a trademark of Sun Microsystems
      Incorporated. Microsoft Windows, Microsoft Office, MS Mail, and Windows Explorer are
      trademarks of Microsoft Corporation. Eudora is a trademark of the University of Illinois Board of
      Trustees, licensed to QUALCOMM Incorporated. Netscape Navigator is a trademark of Netscape
      Communications Corporation.

Draft 8                                       8/9/2012 6:11 PM                                     Page 3 of 15
Rein & McCue                                                                           Intranets: Experiences at Xerox
                                                       *** DRAFT ***


                           Clients     10000
                                                 SE    OC    No     DE       JA   FE   MA   AP MY
                                                 95    95    95     95       96   96   96   96 96



                                                      SE OC No DE JA FE MA AP MY
                                                      95 95 95 95 96 96 96 96 96



                                             SE       OC    No     DE       JA    FE   MA   AP   MY
                                             95       95    95     95       96    96   96   96   96

                                Figure 1                   Xerox Intranet traffic volume.

The impressive growth of the Xerox Intranet would normally be interpreted as a sign of success.
Contrary to intuition, the research and development units are finding the intranet has serious
limitations when it is used to directly support organizational work. We seek to understand why in
this paper.

Xerox Intranet Applications
In this section of the paper, we examine a sampling of successful and less successful Xerox
Intranet applications with the goal of understanding why certain types of applications are a

Draft 8                                                    8/9/2012 6:11 PM                                Page 4 of 15
Rein & McCue                                                   Intranets: Experiences at Xerox
                                         *** DRAFT ***

natural fit for the technology while others are more challenging. The successful and less
successful applications are discussed, in turn, in the following two subsections.

Successful Intranet Applications
Xerox has several notably successful intranet applications. We briefly describe the applications
developed by two organizations, as they are representative of the set of successful applications.
             The Technical Information Center uses their home page on the Xerox Intranet to
              provide Xerox employees with on-line access to information resources such as the
              corporate library’s collections catalog, subscription databases, and the company
              telephone directory. On-line access is also provided to more specific, targeted
              information such as the current price of Xerox stock.
             The corporate human resources organization uses the intranet to provide Xerox
              employees with on-line access to a variety of private and public information. Private
              information includes personal data (name, home address, birth date, etc.), job/salary
              data (compensation type, skill code, job title and grade, salary history, etc.), and
              benefit data. Public information includes 401k investment fund performance data,
              Xerox policies, van and shuttle schedule, holiday schedules (each Xerox business
              group specifies its own holidays), and stock quotes.
In each of the above applications, the information is factual—it does not involve interpretation.
Also, in each application, the goal is to distribute, or make accessible, this information. The
intranet effectively solves this distribution problem.
Reflecting further on the information in the above applications, there is another important
attribute. The information is dynamic: stock prices change daily, phone numbers change, and so
on. The intranet is so effective at solving the information distribution problem, that for these
applications it also provides a superior way to distribute current information.
For two of these applications, the intranet provides an elegant solution to the information update
problem (ensuring that the information is current). The employee can easily update his/her own
personal information in the telephone directory and the human resources data base. In the
telephone directory example, especially, there is high motivation on the part of the employee to
keep this information current because the Xerox community has come to rely on the on-line
directory when making phone calls.
In summary, for a sampling of Xerox’s successful applications, the intranet solves the
distribution problem for current factual information that does not involve interpretation. These
applications have relatively simple, well understood work processes, and this is another reason
we believe they have been successfully implemented using off-the-shelf Internet software.

Less Successful Intranet Applications
In this section, we discuss some less successful intranet applications. In particular, we examine
Xerox research and technology groups’ experiences with using the intranet to support and

Draft 8                                    8/9/2012 6:11 PM                             Page 5 of 15
Rein & McCue                                                 Intranets: Experiences at Xerox
                                       *** DRAFT ***

maintain repositories of project-related information. “Less successful” here is not meant to imply
that these applications are total failures, but rather that they have been disappointing in one or
more major ways. Our goal in this section is to better understand and explain why the
disappointments arise; hence we focus on the problems, not the successes, with these
applications. In the subsections that follow we discuss the project repository application in
general, email-based implementations, Web-based implementations, and lastly, the major

Project Repository Basics
Given that almost half of the Xerox work force is professionals and technicians and that these
people typically work in project teams, one of the most rapidly expanding intranet application
areas is project repositories. A project repository is a managed archive of “documents” important
to the project, where a document is defined in [8] to be “information stored for human
comprehension.” Both current and past versions of these documents are stored in the repository.
The repository should be secure to protect against corruption and unauthorized access, yet easily
accessible by all who have a need to know.
The set of document types that make up a project repository can be extremely diverse. For
example, the repository for a software development project could include documented software
process procedures (mostly Microsoft Word files), project plans (Word and Microsoft Project
files), requirements documents (Word and PowerPoint files), design documents (Word and
PowerPoint files), system documentation (Word and PowerPoint files), source code (ASCII text
files), test code (ASCII text files), project tracking spreadsheets (Excel files), and more.
Repositories may be organized on an individual project or organizational group basis. Consider
the software development example again. In the former case, each project within the software
development group would have its own repository. In the latter case, the software development
group as a whole would have a repository. With the project-based arrangement, there is typically
a large variation in the ways the repositories are organized since each project does it its own way.
Repositories that are organized at the group level usually have more uniformity at the project
level. Regardless of its organization, the purpose of the repository is to collect and manage the
documents that are critical to the project.

Email-Based Project Repository
The email-based project repository was the first approach to using the intranet for this
application, and many groups still use this approach. It is a natural extension of the Xerox
practice of using email to “cascade” important information (for example, a change in the
corporate travel policy) throughout the organization. The information is sent as an email
attachment, with or without context-setting verbiage, to all division managers and their
administrative assistants, who then forward the message to their managers’ direct reports, and so
on. The attachment may be an internal memo sent as a Word document, a PowerPoint slide
presentation, or even an HTML file.
The email-based project repository simply uses the cascade approach to share project documents
among the members of a project team. With this approach, the project members rely solely on

Draft 8                                  8/9/2012 6:11 PM                              Page 6 of 15
Rein & McCue                                                    Intranets: Experiences at Xerox
                                          *** DRAFT ***

the intranet’s ability to effectively distribute information, hoping to avoid the overhead of
maintaining and managing an actual project repository. There are some noteworthy attributes of
this email-based project repository:
             notification and content (both are part of the email message) are “pushed” out to those
              who need to know,
             the context for the document (if there is context-setting verbiage in the body of the
              email message) is tightly coupled with the document,
             version control is the receiver’s responsibility (receivers squirrel away attachments, in
              effect creating their own repositories),
             ownership of information is ambiguous (since a receiver can modify the information
              before forwarding it, each receiver has to assess the authenticity of each message by
              considering the original sender’s identity).
One can imagine a project repository implemented as distributed system, but clearly off-the-shelf
email software is totally inadequate for managing such a beast. Difficulties with the e-mail
interface result in duplicate messages, messages sent without their attachments, or “bounced”
messages showing up days after the sender believed that task had been completed. Another
complicating factor is that people use different versions of software (e.g., Word and
PowerPoint), and eventually some sender in the cascade chain has to downgrade the attachment
to the lowest version. It is doubtful that anyone has a complete repository; it is even less likely
that everyone has the same repository (this would be a fully replicated distributed system;
wasteful of resources). Even if the project members worked out conventions or procedures for
who should archive what (a true distributed system), it is inconceivable that a complete
repository could be maintained using email as the primary technology.
Web-Based Project Repository
A transition is underway in Xerox’s research and technology units. With the expanding Web, we
are starting to see more Web-based project repositories. Notification of changes to the repository
are still done using email, but instead of attaching documents, email messages reference URLs
that contain the documents of interest. This approach is a natural progression from the approach
described in the previous section. The primary motivation is the desire to save resources (smaller
email messages) and to gain some control over the contents of the project repository (one “copy”
of it sits on the Web). This Web-based approach to project repositories contrasts with the
email-based approach:
             notification is “pushed” (via the email message); content is “pulled” (from the Web),
             the context-setting explanation (i.e., the email message) is separated from the Web
              document (there is no longer an attachment),
             version control and configuration management are centralized at the Web site,
             ownership of the information is intact (managed at the Web site).

Draft 8                                     8/9/2012 6:11 PM                              Page 7 of 15
Rein & McCue                                                       Intranets: Experiences at Xerox
                                           *** DRAFT ***

When using off-the-shelf Internet software, these Web-based project repository applications
experience significant problems that arise as a consequence of the dynamic nature of the Web.
There are unexpected changes in content (i.e., the documents stored within a given node change
without notice), in the format of Web pages (structural changes are made without explanation or
notification, making it difficult to find something specific again), and in location (what was one
place one day may be elsewhere the next). Links as they relate to “topics” may also change
Off-the-shelf Web software has limited mechanisms for version control,4 a function that is
important for this application. Not only is version control missing, but notification is a manual
process. Thus, when there are changes to the repository—either its content or its structure—these
changes must be communicated to the user community; otherwise, they appear unexpected.
Document currency is a related problem: how does one ensure that the documents that are out
there are the current versions without violating the “what one expects” rule?
Furthermore, since it is impossible to determine whether or not a Web document is referred to by
a link somewhere, Web site managers are loath to make structural changes to a site (i.e., moving
or deleting a document). This results in unbounded growth of storage requirements which
ultimately must be resolved by deleting “probably unreferenced” documents and old or
intermediate versions of current documents.

Major Problems
In summary, for the less successful applications, it is as we saw for the successful applications:
the intranet solves the information distribution problem. Many of the problems we discussed
above can be seen as second order problems that arise as a result of the electronic media solving
the (first order) problem of information distribution too well. These applications are complex
(especially compared to the simpler successful applications). They have data and work processes
that are not well understood by designers. Off-the-shelf Internet technology does not support
collaborative activity very well. What success we have seen with intranet technology for these
applications occurs because work groups have evolved effective social protocols that compensate
for the technology’s deficiencies.

    4 Interleaf and Documentum now offer document management Web interfaces providing some form of
      version control, configuration management, and security. Lotus Domino and Netscape Communicator
      are newer Web-based products that may soon address document management functions, although
      they don’t appear to yet.
      Interleaf is a trademark of Interleaf Incorporated. Documentum is a trademark of Documentum
      Incorporated. Lotus Domino is a trademark of Lotus Development Corporation. Netscape
      Communicator is a trademark of Netscape Communications Corporation.

Draft 8                                      8/9/2012 6:11 PM                                   Page 8 of 15
Rein & McCue                                                 Intranets: Experiences at Xerox
                                       *** DRAFT ***

Root Cause Analysis
Each of the applications we have examined was performed by some other means before the Web
or Internet technology was applied. Each application involved documents, frequently manually
administered on paper. Part of the evolved work processes associated with these applications rely
on the nature of paper. Gibson defined the term affordances as the possibilities or opportunities
for human interaction offered by the nature of an object or medium [2]. The affordances of paper
are different than the affordances of the intranet/Web. This difference in affordances may
account in part for the runaway success of some applications using Web technology and the
disappointing lack of success for others.
The Web is a medium with certain affordances relating to its dynamic nature, world-wide reach,
and multiple media (e.g., text, graphics, photographs, color, sound, animation). Many
applications of Web technology intercept, augment, or subsume work processes that previously
involved paper documents. What is it that paper offers as possibilities or opportunities for human
interaction? How do those properties translate to the medium of the Web? In the section that
follows, we examine some of these properties of paper and relate them to properties of the Web.

Affordances of the Web Medium
Paper is a “localized medium.” That is, a paper document can be read only if the reader and the
paper are in the same place at the same time [1]. Compare this with a Web document. The
ubiquity of Web access and its world-wide reach make it unnecessary (or, more accurately,
trivially easy) for the document to be wherever the reader is.
Paper is a physical medium bound by limits of physical space. A limited number of people can
read a paper document at the same time. For some applications, this means paper documents are
best used by one person at a time. For example, it would be difficult for two or more people,
each person working alone, to use the same reference book at the same time [1]. In cases where
multiple people are collaborating on a common task and those people are co-located, paper can
be more effective [6]. Paper can be physically manipulated and laid out in space [6]; navigation
and manipulation are easier with paper [6]. For Web documents, multiple people can read them
at the same time. The limited display size of most computers, however, can be a serious
constraint to natural work processes in many situations.
Paper is better for solid, systematic, consecutive reading [1]. Studies of reading have identified a
number of factors affecting reading including reading styles and strategies, familiarity with text
or subject matter, and the purpose of reading [5]. Reading to learn, reading to search/answer
questions, reading for research, reading for problem solving and decision making, proof-reading,
reading to summarize, reading to write and revise documents, reading for enjoyment, and reading
for critical review are all different purposes for reading with different requirements for document
access, annotation, indexing, highlighting, manipulation, navigation, and note-taking [5]. Current
Web technology is strong in some areas such as access, scanning, indexing, and (certain forms
of) navigation, but weaker than paper in other areas such as annotation, highlighting,
manipulation, and note-taking.

Draft 8                                  8/9/2012 6:11 PM                              Page 9 of 15
Rein & McCue                                                  Intranets: Experiences at Xerox
                                        *** DRAFT ***

Paper is mobile and easy to access, permitting document-related activities to be easily
interwoven with other activities without disrupting them [4]. It is lightweight, sturdy, easy to put
away rapidly, and inexpensive. Paper can be used in poor lighting conditions and does not
require batteries or other forms of electrical power. Although it is possible to acquire a mobile
Web device, using cellular modems and portable computing devices, such access is currently
insignificant compared to the percentage of Web users who use fixed location access.
Access to paper documents can be physically controlled and monitored [6]. The very physical
limitations of paper that make it difficult for multiple independent users to operate from the same
copy of a paper document are a positive strength where limited access is a goal. Very sensitive
information in business and government is still frequently managed using numbered copies of
physical paper documents. Current Web security technology lacks the confidence of its users for
such sensitive documents. In fact, many computer users do not trust even their local computers
with sensitive information, much less computers connected to the world-wide Internet.
Paper documents are stable in both form and content [3]. For applications requiring “the latest”
information available, such as stock share prices, currency exchange rates or the current
temperature, the dynamic capability of the Web is a clear advantage over paper. However, for
more permanent information, or information whose value extends over a longer period of time,
stability of form and content is important. Web technology by itself does not prohibit stable
document form or content, but, in practice, Web documents change often in structure, layout and
content, links that used to work no longer refer to the same documents or perhaps do not refer to
anything at all.
Annotated paper is the living document [6]. In the work of “knowledge workers,” documents are
the artifacts that record the current state of activities in progress. These documents, even if
electronic in origin, are frequently rendered to paper for working sessions involving one or more
individuals. Annotations, notes, corrections, and highlights are captured on paper, later to be
incorporated back into the electronic rendition of the document.
These differences in affordances between paper documents and Web documents offer clues as to
the suitability of applying Web technology to work processes that currently employ paper.
Differences in affordances are observed in both the nature of the document, its content, stability,
and security requirements, and in the nature of the work process in which it is used, such as the
requirements for mobility, collaboration, manipulation, navigation, and dissemination. For Web
technology to succeed in cases where the affordances of paper seem better matched to current
work processes, either the Web technology must change to acquire those affordances (e.g.,
trustworthy security mechanisms) or the work processes must change to accommodate the
affordances of the Web.
Affordances of the medium or technology, however, do not tell the whole story for most work
processes interact with more than one document. Intranets, when viewed as environments for
supporting organizational work, are conceptually electronic collections of organizational work
artifacts. The affordances or properties of document collections are yet another feature around
which work processes have evolved. Unfortunately, consideration for collections is virtually
non-existent in current Web offerings. Fortunately, though, library science has produced a

Draft 8                                  8/9/2012 6:11 PM                             Page 10 of 15
Rein & McCue                                                    Intranets: Experiences at Xerox
                                          *** DRAFT ***

considerable body of knowledge on the nature and use of document collections with respect to

This section summarizes some of the key ideas about collections from the library science
perspective and relates these ideas to current Web technology. Traditional (non-electronic)
collections play four important roles: preservation, dispensing, bibliographic, and symbolic [1].
             Preservation role: “any document that is not collected and preserved is likely to be
              lost, unavailable to readers both now and in the future. It is difficult to predict what
              might be of interest to someone in the future. When in doubt, it is prudent to preserve
              nonrenewable resources.” [1]
              In electronic media, preservation means that the “document” exists, that it is easy to
              access, and that its access path is preserved even when the path changes.
             Dispensing role: “the principal reason for most investment in collection development
              is not preservation but the need to provide convenient access to materials that people
              want to see where they want to see them. If someone wants to see a book, it is not
              entirely satisfactory to answer that a copy exists and is being carefully preserved in
              some foreign national library. The need is for a copy here and now.” [1]
              Electronic documents promise to be particularly advantageous in this
              role—promising, once the problems with the preservation role are resolved. On the
              Web, the difficulty of obtaining actual content (difficult to repeat access because its
              content changes quickly as does its location) causes us to make our own copy and
              store it where we know where it is (i.e., to preserve for our own use).
             Bibliographic role: “the use of materials depends on identifying and locating what
              exists.” [1]
              On the Web, we don't do a good job of identifying. We rarely identify and when we
              do, we allow the content to change or transform—this invalidates identity if there was
              one. In other words, we fail to preserve the old and to identify and catalog the new.
              With respect to the locating function, the Web offers powerful search engines—in
              effect, dynamic bibliographies.
             Symbolic role: “large collections, particularly of special materials, bring status and
              prestige to the collector whether materials are used or not. The symbolic value of
              collections and of buildings to house them is, perhaps, more marked in the case of
              museums but should not be ignored in the case of libraries.” [1]
              On the Web, symbolic value might also account for why people make their own
              copies/collections. It certainly explains part of the attraction of Web advertisers to
              create “the largest” collections..
The development of collections on the Web has grown and matured in recent years. Early Web
collections employed a strategy of collecting as many pages as possible (more pages increase
chances of a “hit” at the site). Soon, this strategy was replaced by a strategy of collecting as

Draft 8                                     8/9/2012 6:11 PM                             Page 11 of 15
Rein & McCue                                                        Intranets: Experiences at Xerox
                                            *** DRAFT ***

many links as possible since links are smaller and easier to store and, performance considerations
aside, functionally equivalent to “having” the document. But having long lists of links is of
limited use: the link names are insufficiently descriptive and organizing them is difficult. The
solution to that problem is a strategy of developing the best search tools possible. This is where
the action is today as exemplified by the many powerful search tools now on the market (e.g.,
Yahoo!, Lycos and AltaVista).5 Seeing how far this still leaves us from the capabilities of paper
collections offered by libraries, we predict that the next focus will be on the development of Web
tools for collection development and management, and when these tools exist people will then
seek to develop the most extensive collections they can.

In an attempt to understand why some intranet applications are less successful than others, we
have looked at a sampling of applications from Xerox. For the successful applications (e.g., the
company phone book), we saw there was a match between the affordances of the electronic
medium and the applications’ needs. For these applications, the intranet effectively solved the
distribution problem for current factual information. For the less successful applications (e.g., the
project repository), the intranet solved the information distribution problem and simultaneously
intensified the document collection management problem.
We cannot over-emphasize the importance of understanding the work processes of an application
in the successful application of intranet technologies. The successful applications had relatively
simple, well understood processes. In contrast, the less successful applications were complex:
they had data and work processes that were not well understood. Sorgaard’s study of the
successful integration of the Web into work processes [7], confirms this point: all of the
applications he discusses have the same attributes as our successful applications, and he
discusses no applications as complex as the project repository.

Roles that Web-Based Tools Need to Support
If we step back from the specific work processes of our applications for a moment and consider
the general work processes that are involved, we can identify three roles that it seems Web
technologies need to support: writer, reader, and collection manager.
             The writer of a document is responsible for acquiring or generating the information
              content of the document, organizing that information, and choosing layout and format
              for effectively communicating that information to the intended audience.
             The reader of a document is first a seeker of information. This search process
              identifies and locates the relevant documents. The reader then uses the documents for
              some purpose (research, review, education, enjoyment, etc.)

    5 Yahoo! is a trademark of Yahoo! Corporation, Lycos is a trademark of Lycos Incorporated, AltaVist
      is a trademark of Digital Equipment Corporation.

Draft 8                                       8/9/2012 6:11 PM                                  Page 12 of 15
Rein & McCue                                                   Intranets: Experiences at Xerox
                                         *** DRAFT ***

             The collection manager is a service provider, matching authors to readers by
              managing the artifacts of their communication: the documents. These documents
              must be stored in some accessible place and for some applications it may be desirable
              that readers be notified of their availability.
In using the Web, a typical user plays all these roles at one time or another, sometimes in rapid
succession. For example, when actually creating document content one is a writer; when looking
at existing Web documents, perhaps to see what else is out there or to find a related item, one is a
reader; and when one places the finished document out on the Web, one is playing the collection
manager role. In contrast, the work of the Web master is primarily in the collection manager role.
Current Web technology does a decent job of supporting writers and readers for certain aspects
of their tasks. Specifically, for writers, many desktop publishing systems provide HTML
translators and editors. For readers, there are many powerful search engines. Support for the
collection manager role, however, is virtually non-existent. The few collection manager
functions, such as versioning and dead link detection, that are provided in some commercial
products are positioned as authoring tools. The association of collection management tasks with
the wrong role suggests that the collection manager role is not well understood, perhaps even

Insights from Library Science
The library science perspective on collection development may provide some insights about the
collection manager role. As Buckland [1] explains “Collecting material does not create material.
It affects where copies are located.” Commercial products’ positioning of collection management
functions as authoring tools is confusing collection with creation. We suggest that it would be
beneficial if we started referring correctly to these functions as collection management functions.
Buckland [1] continues, “Library collection development is a matter of ‘file organization,’
concerned with where copies of documents are to be located and for how long.” This is
interesting—collection management for librarians deals with the activity of organizing copies of
documents. For the Web, we are not so much concerned with copies of documents, because the
ubiquity of Web access and its world-wide reach mean that the master document can be
wherever the reader is. Therefore, we suggest that it is useful to view collection management as
dealing with the organization of documents (delete “copies of”), including where they are located
and for how long.

Open Questions
A traditional librarian (or collection manager) performs many actions on a document. He/she
catalogs it, gives it a call number (stable identifier), stores it in some accessible place, creates
bibliographic entries for it. He/she also serves readers by notifying them of new bibliographies
and holdings and by giving advice on use of search tools and the search process. Reference
librarians seek information for reader. It would be a useful exercise to consider how the
affordances of the Web might support these document and service activities. It may be the case

Draft 8                                    8/9/2012 6:11 PM                             Page 13 of 15
Rein & McCue                                                Intranets: Experiences at Xerox
                                       *** DRAFT ***

that we cannot come up with technology solutions that take advantage of the affordances of the
Web. In these cases, well developed social protocols can partially compensate for lack of
functionality in tools.
We also have to ask new questions (librarians may not have thought about them). For example,
one open question that we have touched upon concerns community maintenance versus central
maintenance. We need to learn what maintenance method works when. Insights may be gained
from reflecting on the non-electronic world. For example, in our group we have observed how
the community maintenance model works in two different conference room settings: one room,
Kearns, is a show case facility; the other room, 114/100, is a simple, basic facility. Community
maintenance works especially well for Kearns and is much less successful for 114/100. Both
conference rooms are booked solid for months in advance; both are used by many different
groups (i.e., the rooms do not belong to any one group); food may be served in both.

Suggestions for Future Design Efforts
In this paper, we have implicitly suggested a way of working as we pursue the design and
development of future Web-based tools and environments. Explicitly, we should look to other
relevant disciplines for guidance and ideas. We should ask new questions—the affordances of
the Web may allow new ways of working. We should look at work processes in the paper-based
world—understanding the differences between the affordances of paper and the Web may lead to
profoundly new ways of working.

We gratefully acknowledge the constructive comments and feedback of our colleagues on earlier
drafts of this paper: Herve Gallaire, Rick Peebles, Judith Slein, Abi Sellen, and Kenton O’Hara.
We thank Michael Bieber, Tomas Isakowitz, and Fabio Vitali for organizing this special issue of
the Communications and for their patience and understanding in our needing more time for the
first draft.

1. Buckland, M. Redesigning Library Services: A Manifesto. American Library Association,
   Chicago, Illinois,1992.
2. Gibson, J. J. The Ecological Approach to Visual Perception. Houghton Mifflin, New York,
   New York, 1979.
3. Levy, D. Fixed or Fluid? Document Stability and New Media. In Proceedings of the
   European Conference on Hypertext Technology ’94, Edinburgh, Scotland, ACM, 24-31.
4. Luff, P., Heath, C., and Greatbatch, D. Tasks-in-interaction: Paper and Screen Based
   Documentation in Collaborative Activity. In Proceedings of CSCW ’92, 163-170.
5. O’Hara, K. and Sellen A. A Comparison of Reading on Paper and On-line. To appear in
   Proceedings of CHI ’97, Atlanta, GA, 1997.

Draft 8                                 8/9/2012 6:11 PM                            Page 14 of 15
Rein & McCue                                             Intranets: Experiences at Xerox
                                     *** DRAFT ***

6. Sellen, A. and Harper, R. Paper as an Analytic Resource for the Design of New
   Technologies. To appear in Proceedings of CHI ’97, Atlanta, GA, 1997.
7. Sorgaard, P. Work Behind the Service: Web Publishing and Changes in Document
   Production. In proceedings of the 1996 Information and Process Integration Conference
   (IPIC ’96), 387-396.
8. Unni, A. and Bhamidipati, R. Documents and Business Process, Understanding the Links. In
   proceedings of the 1996 Information and Process Integration Conference (IPIC ’96), 11-26.
9. Yin, R. Case Study Research: Design and Methods. Second Edition. Sage Publications,
   Thousand Oaks, California,1994.

Draft 8                               8/9/2012 6:11 PM                         Page 15 of 15

To top