Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT
Much of today’s organizational knowledge still exists outside of formal information repositories
and often only in people’s heads. While organizations are eager to capture this knowledge,
existing acquisition methods are not up to the task. Neither traditional artificial intelligence-
based approaches nor more recent, less-structured knowledge management techniques have
overcome the knowledge acquisition challenges. This article investigates knowledge acquisition
bottlenecks and proposes the use of collaborative, conversational knowledge management to
remove them. The article demonstrates the opportunity for more effective knowledge acquisition
through the application of the principles of Bazaar style, open-source development. The article
introduces wikis as software that enables this type of knowledge acquisition. It empirically
analyzes the Wikipedia to produce evidence for the feasibility and effectiveness of the proposed
approach.
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 71
isting knowledge from organizational sources, tions work well for high-volume applications
especially from people? And how can we man- such as credit approval. Even then, the knowl-
age the maintenance so as to assure that the edge creation effort is highly resource-inten-
stored knowledge is accurate and up-to-date? sive (Lee, 2001). When insufficient data vol-
Discovering answers to these questions is im- umes thwart data mining efforts, the acquisi-
portant for organizations as information work tion activity has to elicit knowledge directly
becomes knowledge work, thus requiring from experts as rules and facts or similar formal
knowledge to support non-routine decision representations. This should be done under the
making (Drucker, 1993, 1999). It is similarly im- guidance of knowledge engineers trained in
portant for organizations whose corporate por- knowledge elicitation, formalization, and repre-
tals that were set up years ago increasingly are sentation. Yet a knowledge engineer’s produc-
becoming dated and stale (Newcombe, 2000). tivity is limited to hundreds of rules per year for
Furthermore, it is important for organizations in development and maintenance (Sviokla, 1990;
the business of creating knowledge assets who Turban & Aronson, 2000). This productivity
are faced with increased costs of knowledge level may be acceptable for high value-added
creation, shorter knowledge life cycles, and in- projects but limits the broad applicability of the
creased knowledge obsolescence. approach. Smaller projects have attempted to
Seeking a solution to the problems of or- rely on capturing knowledge without knowl-
ganizational knowledge acquisition, the article edge engineers, relying on end-user develop-
makes the following argument. First, it intro- ment. The latter has not been very successful
duces previous approaches to knowledge ac- (Wagner, 2000, 2003). Wagner found end-user
quisition, identifies four limitations, and of- expert systems often to be poorly structured,
fers evidence for these limitations. The article incomplete, highly coupled, and thus, difficult
then refers to Bazaar style (software) devel- to maintain. Artificial intelligence-based meth-
opment (Raymond, 2001) as a potential direc- ods thus are facing considerable applicability
tion for knowledge asset creation. It then ex- constraints. Consequently, organizational
plains the concept of conversational knowl- knowledge management efforts have sought
edge management and advocates wiki tech- to capture knowledge in less formal ways; for
nology and the “wiki way” (Leuf & instance, by extending document management
Cunningham, 2001) as a possible approach to and groupware systems into knowledge man-
using Bazaar-style methods in conversational agement systems (Davenport & Prusak, 1998;
knowledge management. An empirical analy- Holsapple & Joshi, 2002) in part through better
sis of the viability and effectiveness of the indexing, search engines, and linking.
approach follows. The article ends with impli- Yet challenges remain. When organiza-
cations and conclusions about the future of tions try to make sense out of large volumes of
conversational knowledge management. documents in their document management sys-
tems, they usually need search engines, text
KNOWLEDGE ACQUISITION mining, and automatic indexing tools, resulting
in an expensive solution with limited success
Approaches to (Bygstad, 2003). Furthermore, this approach is
Knowledge Acquisition well suited only for relatively stable and cen-
Organizations that try to acquire organi- tralized knowledge bases. Users of such knowl-
zational knowledge formally (based on artificial edge bases often encounter information over-
intelligence methods) have relatively few avail- load, irrelevant responses, or no response to
able alternatives. For application areas with large queries. Alternatively, organizations might use
amounts of transaction data, data mining can expert reports and harvest expert knowledge to
induce rules from that data. Data mining solu- capture the methods used by domain experts
(Snyder & Wilson, 1998). Again, this method
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
72 Information Resources Management Journal, 19(1), 70-83, January-March 2006
often is limited to niche applications, requires volunteers. Open source projects engage soft-
considerable effort, and still faces knowledge ware developers, wherever they may reside, and
maintenance difficulties (Malhotra, 2000). Other have them collaboratively develop the knowl-
solutions, such as corporate controlled portals, edge asset (the software). Surprisingly, this
can quickly suffer from outdated knowledge activity takes place with little centralized man-
and lack of maintainability (Newcombe, 2000). agement. Raymond (2001) characterized this
approach to software development as the Ba-
Knowledge Acquisition Bottleneck zaar style in contrast to the traditional cathe-
In summary, we can describe the knowl- dral style of development. Cathedral is a meta-
edge acquisition bottleneck as follows (Wagner, phor for the development of a large monolithic
2000; Waterman, 1986): artifact through a structured and lengthy de-
velopment process. Fundamental to the ca-
• Narrow bandwidth. The channels that exist thedral style approach is that source code is
to convert organizational knowledge from only widely available at release dates with ac-
its source (either experts, documents, or cess restricted to a few developers between
transactions) are relatively narrow. release dates. Bazaar style development, how-
ever, occurs over the Internet in constant pub-
• Acquisition latency. The slow speed of ac-
lic view. Raymond identified principles of this
quisition frequently is accompanied by a de-
development style that challenge the assump-
lay between the time when knowledge (or
tion that large and complex software assets
the underlying data) is created and when
need to be built via an a priori, centralized ap-
the acquired knowledge becomes available
proach. Overall, four themes guide this devel-
to be shared.
opment approach, which can be characterized
• Knowledge inaccuracy. Experts make mis- as follows: (1) design simplicity of the artifact,
takes and so do data mining technologies (2) team work, (3) frequent creation of a visible
(finding spurious relationships). Further- work product, and (4) development as an on-
more, maintenance can introduce inaccura- going conversation. This section introduces a
cies or inconsistencies into previously cor- framework of open source (software) develop-
rect knowledge bases. ment, identifies its benefits, and derives les-
• Maintenance trap. As the knowledge in the sons about the applicability for knowledge as-
knowledge base grows, so does the require- sets other than software.
ment for maintenance. Furthermore, previ-
ous updates that were made with insuffi- Open Source Software Development
cient care and foresight (“hacks”) will accu- Open source software development, as
mulate and render future maintenance in- described, for instance, by Raymond (2001),
creasingly more difficult (Land, 2002). Benkler (2002), and Markus, et al. (2000), relies
on several factors to achieve success (and thus,
Given these challenges, it appears that performance of the knowledge creation effort).
there are few opportunities for breaking the Key success factors (see Figure 1) consist of a
knowledge acquisition bottleneck. The next suitable artifact, a skilled and motivated team
section will propose one possible remedy. of volunteer users and developers, a lean and
transparent development process, and light-
LEARNING FROM weight but effective governance. Added to this
SOFTWARE DEVELOPMENT is an enabling factor; namely, an appropriate
One area that has offered lessons for the technology infrastructure, which, for instance,
successful creation of knowledge assets is soft- permits frequent releases, accommodates vot-
ware development, and specifically open source ing mechanisms to govern the community, or
software development by distributed teams of enables fast and reliable version management,
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 73
Artifact Group
• “Value proposition” • Core team
• Credible core • Skilled individuals
• Modular design • Voluntary participation
• Standards based • User / developer duality
Performance
Process Governance • Fast development
• Frequent, early releases • Lightweight, but with • High quality
• Continuous peer review operational, collective • Free product
• Parallel, distributed choice, and constitutional
development choice rules
• Version management • Free redistribution of the
• Low participation overhead work product and its
• Full disclosure of technology derivatives
functions • Open license
all with little overhead and few transaction costs. be applied: (1) simplicity of design and frequent
With these factors in place, open source soft- redesign (refactoring) to maintain simplicity, (2)
ware development promises faster development teamwork (3) frequent creation of a small work
speed than proprietary approaches (including product available for review and testing, and
higher developer productivity) and a better (4) development as conversation to facilitate
quality product, which is also free. back-up, clarity, and shared understanding.
Open source software development has Applications of this kind exist within organiza-
had remarkable successes, creating software tions, and among organizations and people. For
that appears to break long-standing rules of example, companies could conceivably turn
software evolution (Scacchi, 2004). For example, their traditional help desks into open help desks,
open source software size has been shown to where customers would openly share their prob-
grow super-linear (exponential) rather than lin- lems with others, help each other, and free up
ear or inverse-square (Mockus et al., 2002). company experts to tackle only the most diffi-
cult problems. Unfortunately, companies fre-
Bazaar-Style quently do not want to relinquish control of
Knowledge Management their (closed) help desk. Open help desks exist
Can Bazaar-style development be applied on the Web, typically as discussion forums of
successfully to the creation of knowledge as- questions and answers. While they embody
sets other than software? Several leaders of teamwork and conversation, the resulting work
the open source community have hypothesized product often is not simple and well-structured
this, including Torvalds (Hamm, 2004). Yet but lacks organization and is filled with repeti-
Torvalds also acknowledged that not all knowl- tion and inconsistencies.
edge assets are equally suitable, as the cre- Consequently, one necessary condition
ation process may be too personal or too linear. for this research was to find a knowledge asset
Hence, in order to extend the lessons and ben- that was highly amenable to the Bazaar-style
efits of Bazaar-style development, we should development approach and that used a tech-
target applications where the core themes can nology that facilitated this type of development.
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
74 Information Resources Management Journal, 19(1), 70-83, January-March 2006
The selected asset was an online encyclopedia Wiki Structure and Principles
— Wikipedia (wikipedia.org) — that employs A wiki is a set of linked Web pages cre-
wiki technology and the “wiki way” of knowl- ated through the incremental development by
edge asset creation (Leuf & Cunningham, 2001). a group of collaborating users (Leuf &
The article will provide more detail on the Cunningham, 2001) as well as the software used
Wikipedia application, following a briefing on to manage the set of Web pages. Ward
knowledge management with wiki technology Cunningham developed the first wiki in 1995 as
and the wiki way. the PortlandPatternRepository in order to com-
municate specifications for software design
CONVERSATIONAL within a large, heterogeneous community. The
KNOWLEDGE MANAGEMENT term wiki (from the Hawaiian wikiwiki, mean-
ing fast) references the speed with which con-
WITH WIKIS
tent can be created with a wiki. Wiki key char-
Knowledge management with wikis has
acteristics are as follows:
recently drawn media attention (Brown, 2004;
Hof, 2004; Ripley, 2003) as a new, end user de-
veloped approach founded on collaboration • It enables Web documents to be authored
and conversation. Collaborative knowledge collectively;
management means that many people work to- • It uses a simple markup scheme (usually a
gether to create or acquire knowledge instead simplified version of HTML, although
of a few individual experts. In other words, a HTML frequently is permitted);
community (of practice) will jointly create and • Wiki content is not reviewed by any editor
maintain the knowledge. Research elsewhere or coordinating body prior to its publica-
(Cheung et al., 2005) suggests that conversa- tion; and
tional knowledge management is well suited for • New Web pages are created when users navi-
this challenge, whereby conversations (i.e., gate a hyperlink that points nowhere.
questions and answers) become the source of
relevant knowledge. Underlying these characteristics are spe-
Conversational knowledge management cific principles that have shaped wiki software
has become popular in communities that form as well as its use. They are intended to produce
around discussion boards. Leading solutions a development environment where multiple
such as ezboard or Yahoo groups are now used people easily can create and modify a set of
by millions of communities1. Yet while discus- jointly owned Web pages. Wiki pages are ex-
sion forums have been a simple and practical pected to be open, incrementally developed,
solution to share knowledge through conver- and organic; require little markup; have consis-
sation, they lack several useful knowledge rep- tent edit functions and clear naming, be heavily
resentation and maintenance features. For ex- hyperlinked and easily observable (found). As
ample, discussion forum postings, even within a result, wiki pages are expected to change and
a single thread, often do not build upon each improve incrementally.
other. As a result, the latest post may not be an
incremental improvement of earlier ones. An Wikis in Use
alternative technology, which combines the
most desirable features of other conversational
Creating Wiki Pages
technologies, is the wiki. This section discusses
Creating and editing wiki pages is a simple
wiki technology and its suitability for knowl-
activity. A wiki author will use a Web-enabled
edge management.
formfield to enter a comment he or she wishes to
publish. Authors can use plain text or a simpli-
fied markup language. The system then auto-
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 75
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
76 Information Resources Management Journal, 19(1), 70-83, January-March 2006
debate how much of their content is informa- later, the (English) Wikipedia contains about
tion instead of knowledge, encyclopedias con- 280,000 articles.
tain insights (factual), rules (inferential), prin- Wikipedia, applying wiki principles, ap-
ciples (inferential), and so forth. They also fit pears to enable its developers to use a Bazaar-
the definition of information in context (Dav- style approach. Specifically, writers can make
enport &Prusak, 1998), since they frequently incremental changes and then commit and pub-
link concepts to other concepts (cross-refer- lish them immediately. Also, articles can be writ-
encing). By design, encyclopedias also are ten by numerous writers as joint authors, thus
relatively loosely coupled knowledge assets, building on the work of others or correcting
whose components (articles) can exist inde- mistakes. Furthermore, Wikipedia rules stress
pendently. Encyclopedias frequently are com- an authoring etiquette that incorporates rules
piled from the work of a group of authors who of article design and redesign targeted toward
know little about each other or each other’s simple and clear articles. In other words, it is
work. Encyclopedia articles have common possible for Wikipedia authors to follow the
structural elements, since all articles are defi- main themes of Bazaar-style development.
nitions. They typically also follow some stan- Whether authors do so and whether the out-
dards for articles of a similar type (e.g., all biog- come of their efforts is consistent in its effec-
raphies are structured similarly and different tiveness with Bazaar-style software develop-
from city descriptions). ment needs to be determined empirically.
The majority of digital encyclopedias,
such as Britannica, Encarta, Compton, or Grolier, Research Questions
is closed source. They are compiled by a rela- The research sought to address two ques-
tively small group of commissioned writers and tions through empirical analysis.
editors. The result of their work only becomes
available to the readership once the entire edit 1. Is conversational knowledge management,
process has been completed and the new en- as demonstrated in Wikipedia, consistent
cyclopedia version is released. Yet, because of with Bazaar-style knowledge asset creation?
their loosely structured nature, encyclopedias 2. Is conversational knowledge management,
(and other, similar knowledge assets) also can as illustrated by Wikipedia, able to achieve
be created in Bazaar style, given certain condi- the benefits of Bazaar-style development?
tions. The work product cannot be an off-line
product such as a book or a CD; the technology The research thus needed to determine
in general has to be amenable to Bazaar-style whether “Wikipedians” would follow Bazaar-
knowledge acquisition and representation, and style knowledge acquisition and whether the
the organization creating the encyclopedia has effect would be improved knowledge acquisi-
to formulate procedures and methods that en- tion. Based on the criteria in Figure 1, numer-
able this type of knowledge acquisition. Bazaar- ous questions would have to be addressed.
style knowledge acquisition, therefore, becomes Yet, as Table 2 illustrates, compliance with the
a possibility when the asset is created following majority of criteria was confirmed from
the wiki way. Hence, Wikipedia, the online ency- Wikipedia information (Wikipedia Web site and
clopedia developed as a wiki, was used as the Wikimedia Meta-Wiki), leaving four core ques-
knowledge asset to be analyzed for this research. tions to be answered.
Wikipedia is one of several knowledge products Thus, the research questions focused
developed over the last few years with wiki tech- on the incremental nature of the knowledge
nology and the wiki way of development. Other acquisition effort, the multi-person effort, and
applications include Wikitravel and Wikibooks. the effect on the growth and quality of the
Development of Wikipedia began in 2001. As work product, as described in the following
of May 2004, less than three and a half years subsections.
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 77
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
78 Information Resources Management Journal, 19(1), 70-83, January-March 2006
to extend it and possibly to correct mistakes. written and the number of edits it took to create
This would reflect one of the key themes of open the article version. This ratio was calculated for
source, also called Linus’ [Torvalds] Law; articles in their early stages (20% of develop-
namely, that “given enough eyeballs, all bugs ment effort) and at their present state. Decreas-
become shallow.” Hence, the research sought to ing ratios would indicate more effort being spent
determine whether enough eyeballs were scruti- over time on article refinement. To exclude in-
nizing each article, at least more than two. Hence, significant edits, the research only considered
the analysis focused on whether article publica- non-minor changes (counted separately in
tion and maintenance was a multi-person effort. Wikipedia). The expectation was that, over time,
more effort would be devoted to increased ar-
H2: Knowledge acquisition and maintenance ticle quality. It is a stated Wikipedia goal to
in individual Wikipedia articles is a multi- increase quality as articles mature (see, for in-
person effort. stance, the reply to objections concerning
Wikipedia, which discusses quality and growth
Effectiveness issues, at http://en.wikipedia.org/wiki/
The research sought to determine Wikipedia:Replies_to_common_objections).
whether encyclopedia development adopting The corresponding hypothesis concern-
the wiki way would be effective. In this explor- ing quality improvement was as follows:
atory study, effectiveness was measured
through two variables; namely, (1) growth of H4: Edit effort targeted at quality improve-
the knowledge asset and (2) quality improve- ments for individual Wikipedia articles
ment efforts. Growth of the knowledge asset will increase over time, demonstrated by
was determined, based on the increase in the reduced article growth per edit.
number of articles in the Wikipedia over time.
In line with other open source successes Data Source
(Mockus et al., 2002), the expectation was that Wikipedia is an open encyclopedia in
growth would be linear or better (super-linear). many ways. In addition to articles being freely
accessible, so is the history of their creation,
H3: Wikipedia growth in terms of number of including dates, content of each version, and
articles will be linear or super-linear. author information. Hence, it was possible to
trace changes, change frequencies, and author
Unable to assess the overall quality of contributions. To address the first two ques-
the Wikipedia objectively vis-à-vis other ency- tions, 80 articles were randomly selected with
clopedias, the research focused on process the one qualification that 40 of them had to be
quality and specifically quality improvement created originally in 2001 and 2002. More re-
efforts. These efforts were operationalized by cent articles were ignored because of their short
the ratio of edit efforts vs. the growth of history. To determine knowledge asset growth,
Wikipedia articles. In other words, the research Wikipedia summary statistics were accessed,
tested whether editing efforts were devoted to which logged the number of articles written
increasing the size of articles or to refining ex- each month from the start of Wikipedia.
isting articles. The assumption was that refine-
ments (without significant increase in size) Results
would improve overall quality, for instance,
through an increase in presentation quality, Incremental Development
content quality, or the inclusion of more view- (Release Early and Often)
points (diversity). This analysis focused on two samples of
Hence, we computed a words-per-edit 40 articles from 2001 and 2002. For both of these
ratio based on the number of words (per article) samples, the edit efforts for the first 20% of
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 79
each article’s existence (up to the measurement Given these results, what is the likelihood
point in March 2004) were compared against that articles overall were predominantly single-
the entire development effort. The results do authored? Virtually none. A t-test showed very
not support the notion of a short effort burst significant differences between the actual au-
but one of incremental development, as shown thor numbers and the possibility of single au-
in Table 3. thorship. This is a strong result, yet the reader
For articles started in 2001, the first 20% is reminded that the sample articles were old
of an average article’s existence accounted for articles. More than half of the Wikipedia ar-
about 34% of the article’s size (793 words out ticles were less than 12 months old (as of June
of 2,319) and less than 6% of its edits (17 out of 2004) and will have been edited by fewer people.
288). For the 2002 articles, the first 20% ac- An additional sample of 40 randomly selected
counted for about 36% of article size and 15% articles started in 2003, though, still corrobo-
of article edits. Overall, this was considerably rated the results (average of 48 authors, t =
less than expected according to the Pareto rule. 7.164, p = 0.0000). In other words, as time
All results are highly significant (p < 0.0001). progresses, Wikipedia articles are scrutinized
Hence, size grew relatively incrementally with a by “many eyeballs”.
somewhat larger upfront effort (about 35% of
size produced in 20% of the time). Wikipedia Wikipedia Growth
edits were even more incremental with a dis- Data points concerning the growth of
proportionately small number during the early Wikipedia illustrate dramatic growth. Although
existence of an article (15% or less of the edits Wikipedia has existed since 2001, more than
in 20% of the time). half of its approximately 280,000 articles (En-
glish articles as of May 2004) were written since
Multi-Person Effort June 2003. (See http://www.wikipedia.org/
Each of the 80 articles in the two samples wikistats/EN/TablesWikipediaEN.htm).
also was evaluated according to the number of To explore the growth pattern further, the
authors. None of the articles in the sample was analysis targeted the numbers of new articles
co-authored by fewer than 18 people, and the created each month. Three different time series
maximum number of authors for any article was were compared: number of articles, log of num-
285. On average, more than 96 authors worked ber of articles, and square root of number of
on an article (Table 4). articles. For each series, the fit was computed
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
80 Information Resources Management Journal, 19(1), 70-83, January-March 2006
to determine which one best predicted releases in an environment that enables con-
Wikipedia growth. As Table 5 illustrates, versational knowledge acquisition. In the case
Wikipedia growth is best explained by a qua- of Wikipedia, this was possible for several rea-
dratic function (R2 = 0.988, highest). In other sons. First, Wikipedia was able to draw a large
words, Wikipedia article growth is most likely and quadratically growing developer group (ap-
quadratic and, thus, super-linear, which is an proximately 7,000 as of May 2004).
aggressive growth pattern. Quadratic growth Second, Wikipedia pages are highly
also best explained the increase in the number decoupled from each other so that new authors
of Wikipedians and in the number of edits can write with little concern for the current con-
(changes) made to Wikipedia. tent of other pages. When an author breaks a
hyperlink or negatively affects content, it be-
Quality Improvement comes quickly apparent, and other Wikipedians
The second effectiveness measure, the will fix the problem. Third, when authors make a
allocation of effort to quality improvement, sug- contribution, whether writing a new page or
gested a shift toward more quality as Wikipedia changing an existing article, the result is imme-
articles aged. Table 6 illustrates that during the diately visible to the entire community, thus
first 20% of an article’s existence, each edit re- enabling quick releases with minimal latency
sulted in about 60 additional words vs. 11 or and multi-user quality assurance. Therefore, the
fewer words for the remaining 80% of the transaction cost of making a contribution is low,
article’s life (up to the measurement date in much lower than in any peer-reviewed or closed
March 2004). source authoring environment (Ciffolilli, 2003).
The differences in the means of these ra- Fourth, there is no individual ownership of
tios were highly significant, confirming that Wikipedia pages, which are developed by vol-
later effort is an investment in article quality unteers; thus, everyone works to improve
rather than article length. everyone’s contributions. Quality is everyone’s
responsibility. Fifth, Wikipedia has strong edit-
Discussion ing guidelines that are motivated by the
Results of the exploratory study confirm refactoring rules of software development and
what has been expected. Hypotheses H1, H2, principles of objectivity. This ensures that ar-
H3, and H4 were all confirmed. Knowledge ac- ticles, which might have suffered in readability
quisition efforts apparently can successfully from the disjointed work of multiple contribu-
adopt Bazaar-style development with multi-user tors and commentator, ultimately become very
involvement, incremental changes, and quick readable again.
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 81
As a result, in three and a half years of Wikipedia articles, for instance, remains a
existence, Wikipedia has challenged the other- source of arguments. Therefore, future research
wise largest but closed authorship Encyclope- will need to investigate the quality of the re-
dia Britannica (Britannica Online) for leader- sulting knowledge based on content. In addi-
ship in content (Britannica has self-reportedly tion, knowledge creation with wikis relies on a
about 100,000 entries, although with a larger strong and positive social contract among its
word count per article). Other wiki-supported contributors and on subject matters that are
knowledge assets, such as Wikitravel, for in- not controversial. These conditions are not al-
stance, may be able to achieve similar leader- ways present. Wikipedia does have guidelines
ship roles in their knowledge domain. The open, in place to handle disorderly participants and
multi-user model also appears to scale well by to maintain a neutral point of view (NPOV) in
interesting an increasingly larger user popula- articles. But Wikipedia clearly relies on the so-
tion to contribute their efforts, thus keeping cial capital within its community. Studies of less
the article latency at about 10 days for old ar- strong communities will have to be part of the
ticles (initially created in 2001) and less than future research in order to determine knowl-
two days for newer articles (i.e., 2003). How- edge losses due to lack of social capital. Fur-
ever, since wiki technology is relatively new thermore, Bazaar-style knowledge management
and contrary to many organizations’ cultures, relies on volunteers who are genuinely inter-
we should not expect this approach to become ested in the cause. This may not be a paradigm
predominant soon. In fact, the successes are for organizations where knowledge assets are
few at present. However, one should expect an not free. Future research will need to explore
increasing number of wiki software products to the applicability of open source knowledge
emerge in the future and an increasing number management when the intellectual property is
of communities to replace their inferior conver- at least partially proprietary. Finally, the dis-
sational technologies with wikis. cussed approach to knowledge management
appears to work, partly because it can engage
CONCLUSION increasing numbers of participants to deal with
The challenge of capturing and maintain- a growing task domain. One has to wonder
ing exponentially growing volumes of knowl- about the limits of growth of this scenario. Con-
edge requires new ways of knowledge acquisi- sidering both the positive findings and the chal-
tion; namely, on approaches that rely on the lenging questions, it appears that Bazaar-style
contributions of many rather than the expertise knowledge acquisition using wikis will be a
of a few. Wiki technology and the wiki way of promising application for the practice of knowl-
collaboration show a feasible model for knowl- edge management as well as a rich source of
edge acquisition and maintenance. Wikipedia interesting research questions.
offers an illustration of the effectiveness of this
approach. The research demonstrates that us- ACKNOWLEDGMENTS
ers of a wiki-based knowledge asset (i.e., The research described in this article was
Wikipedia) apply Bazaar-style methods and supported in part by CityU grant
techniques in their conversational knowledge SMA#9030992. The author completed a sub-
asset creation. The research also suggests that stantial portion of the research leading to this
knowledge acquisition through collaboration article during two sabbaticals at the School of
and conversation can lead to super-linear Information Science, Claremont Graduate Uni-
knowledge asset growth and continuous qual- versity. The research assistance of Karen
ity improvement. Cheung is thankfully acknowledged.
Not surprisingly, there are several cave-
ats. For instance, knowledge quality cannot be
measured or managed easily. The quality of
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
82 Information Resources Management Journal, 19(1), 70-83, January-March 2006
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.
Information Resources Management Journal, 19(1), 70-83, January-March 2006 83
Copyright © 2006, Idea Group Inc. Copying or distributing in print or electronic forms without written permission of Idea Group Inc.
is prohibited.