Wikipedia: Quality through Coordination
We know that online collective intelligence can achieve incredible things — for example, the ESP Game, Wikipedia, Mechanical Turk. But many of these systems assume that individuals make independent assumptions and that these judgments can be aggregated. Neither of these are true in Wikipedia, where “truth” is generated through an intense combination of coordination and discussion.
So, what happens to article quality when collaboration occurs on Wikipedia? Aniket Kittur and Robert Kraut from CMU set out to explore this in “Harnessing the Wisdom of Crowds in Wikipedia: Quality through Coordination.” They tried to predict public markers of article quality (e.g., “featured article”) using metrics such as number of editors, amount of communication in the talk pages, and whether edits were consolidated under a small number of users or truly crowdsourced.
What did they learn? Surprisingly, adding more editors led to no increase in article quality. More communication and a higher concentration of work both improved quality — but there were interaction effects. When work is highly concentrated, more editors increase quality (but not so when work isn’t concentrated); communication stops improving quality as the number of editors goes up.
Conference report from CSCW 2008.