Jimmy Wales wrote:
My research (conducted in December) showed that
half the edits by logged
in users belong to just 2.5% of logged in users. It would be extremely
interesting to run tests to compare "edit dispersion" for new articles,
old articles, heavily edited articles, highly watched articles, heavily
trafficked articles, etc.
Those sound like plausible numbers, but I think your methodology is
somewhat overstating the cohesiveness: The "drive-by editors" who are
the least associated with an identifiable Wikipedia community are the
users who haven't even bothered to go so far as to create a user
account, which you explicitly exclude from your count.
Users who aren't logged in make only around 18% of all edits. It
doesn't change the results materially.