Halfaker et al.’s paper presents interesting research on quality control in Wikipedia. Since Wikipedia is often the first place one looks for information, this is an important topic to many of us. The content is prepared (and corrected) by millions of volunteers who work with little formal organization. Many of us who use Wikipedia wonder about the quality of the entries, including how they are corrected and improved.
When a Wikipedia article is written and stored by a registered editor, edits can be made and stored by both the original writer and other registered editors. If any edit is considered incorrect, a special kind of edit is made by any registered editor to revert the article to its original reading. Halfaker et al. use a random sample of approximately 1.4 million revisions attributed to registered editors, to examine two categories of factors expected to predict the revisions that will be reverted: measures of quality and factors unrelated to quality. They carefully explain their reasoning for the design and procedures of their research.
Certain factors strongly predict whether an edit will be reverted: “edits that remove established words,” edits from “editors with a history of high quality contributions,” edits from “editors who have been reverted recently,” and “edits that remove the words of active editors.” One factor--“stepping on toes”-- appears to be an example of ownership behavior, even though Wikipedia discourages this. The paper also explores two other factors: editor experience and “editors who cite policy.”
The results presented in this paper are useful for understanding how quality in Wikipedia is maintained.