contributed articles

Doi: 10.1145/1378704.1378720

Why Wikipedia’s remarkable growth is sustainable.

By DiomiDis sPineLLis anD PanaGiotis LouRiDas
the
collaborative
organization
of Knowledge

WIKIPEDIA ( WWW.WIKIPEDIA.ORG) is a freely available online encyclopedia anyone can edit, contributing changes, as well as articles. ¹⁰ With more than a million entries, hundreds of thousands of contributors, and tens of millions of fully recorded article revisions, Wikipedia’s freely available database has also made it possible to study how human knowledge is recorded and organized through an open collaborative process. Although citation analysis⁶ can establish how new research builds on existing publications, the fully recorded evolutionary development of Wikipedia’s structure has allowed us to examine how existing articles foster development of new entries and links. Motivation for our longitudinal study of Wikipedia evolution followed from our observation that even though Wikipedia’s scope is increasing, its coverage is apparently not deteriorating. To study the process of Wikipedia growth we downloaded the February 2006

snapshot of all recorded changes and examined how entries are created and linked. Inspecting the timestamps on individual entry definitions and references, we found that links to nonexistent articles often precede creation of new articles. Also, tracking the evolution of article links allowed us to empirically validate Barabási’s hypothesis on the formation of scale-free graphs through incremental growth and preferential attachment. ¹ Our findings paint a picture of sustainable growth, suggesting that Wikipedia’s development process delivers coverage of more and more subjects.

The phenomenal growth of Wikipedia is attributable to a mixture of technologies and a process of open participation. The key technology behind Wikipedia is that of a Wiki—online lightweight Web-based collaboration. ⁴ Wikipedia content appears online as static HTML pages, though each such page includes an edit button anyone can use to modify its content; editing most articles requires no prior authorization or arrangement. The system maintains the complete edit history of each page and supports a “watchlist” mechanism that alerts registered users when a page they are interested in changes.

The page history and watchlist facilities promote low-overhead collaboration and identification of and response to instances of article vandalism. We found that 4% of article revisions were tagged in their descriptive comment as “reverts”—the typical response to vandalism. They occurred an average of 13 hours after their preceding change. Looking for articles with at least one revert comment, we found that 11% of Wikipedia’s articles had been vandalized at least once. (The entry for George W. Bush had the most revisions and reverts: of its 28,000 revisions one-third were reverts and, conceivably, another third vandalism.) Articles prone to vandalism can be administratively locked against revisions, a step rarely taken; in our study only 0.13% of the articles ( 2,441 entries) were locked.

S TUDYING COOPERATION AND CONFLICT BE TWEEN AUTHORS WI TH HISTORY FLOW VISUALIZATIONS. FERNANDA B. VIÉGAS, MARTIN WAT TENBERG, AND KUSHAL DAVE. PROCEEDINGS OF SIGCHI 2004. H TTP:// ALUMNI.MEDIA.MIT.EDU/~FVIEGAS/PAPERS/HIS TORY_FLOW.PDF