This sounds like a great first project for some of the stats groups
getting started (Erik Garrison, call your office...)
My personal guess, based on some hasty stats months ago, is that the
rate of good-new-article creation dropped, the rate of bad-new-article
creation dropped by a bit less, and the rate of worrisome-vandalism
(not the obvious kind) didn't drop at all. I would like to see a
serious stats breakdown, from before, during, and after the change to
know better what's going on.
Interesting statistics: rate of reverts, rate of speedy-deleted
articles, rate of AfD entries, rate of new user creation, rate of new
users created who never edit, rate of creation of articles that never
get more than one major edit, rate of good additions to AFC, rate of
good additions to AFC that never get turned into an article...
SJ
On 6/21/06, Bryan Derksen <bryan.derksen(a)shaw.ca> wrote:
MacGyverMagic/Mgm wrote:
As I said before, the experiment did cut down on
a lot of crap, because some
newbies simply don't want to register and instead post to AFC where we can
nip an article's creation at the core.
But what are the numbers? There's all this raw data sitting there in the
database from which it should be relatively straightforward to determine
actual statistical measures of what effect the change in policy had, but
since the analysis hasn't been done we're reduced to relying on
subjective impressions and assertions pulled seemingly out of nowhere.
How can I _verify_ that the experiment did or didn't cut down on a lot
of crap? I don't personally do any of the various sorts of "patrolling"
this change was likely to have an impact on (recent changes, new
articles, etc) so I don't even have subjective impressions of my own to
go on.