On 15 June 2010 14:54, Andrew Gray <andrew.gray(a)dunelm.org.uk> wrote:
On 15 June 2010 19:52, Gregory Maxwell
<gmaxwell(a)gmail.com> wrote:
Though I wouldn't recommend trying it _first_
nor would I recommend
trying it while the press is talking about. Perhaps it would be an
intolerable train wreak only because the press is spreading the name
of that article around. It would be unfortunate if we reached
incorrect conclusions on the effectiveness of pending vs protected on
high traffic articles simply due to some temporary attention skew.
Mmm. If we've got a queue - an idea which I have to say I quite like,
even if I was initially a bit confused by it - then why not schedule
in some articles that we expect it not to work very well on? It could
be it has unexpectedly less terrible effects.
Well, part of the objective here is to see whether we get enough
encyclopedia-worthy edits to determine if it is worthwhile removing
protection. Myself, I'd generally be happy if we saw a 1:10 useful edit to
vandalism ratio on most articles, but most articles aren't going to get that
many edits anyway. There are some high-viewership articles in the early
going, so we'll see pretty quickly how much of a difference the pending
changes level makes. However, that same ratio isn't particularly workable if
we're talking about an article that starts getting 50 or more edits a day,
especially when the article involved is a {{good}} or {{featured}} article;
remember that even 5 vandalism hits a day is almost invariably sufficient to
semi-protect an article, not just because of the visible vandalism, but also
because it is a huge waste of volunteer time, and it also impedes the
continued improvement and maintenance of articles.
Unfortunately, we don't have a way of keeping track of the number of pending
changes that are (a) rejected as vandalism/BLP problem, (b) accepted
directly into the article or (c) some other variation, such as putting the
proposed edit onto the article talk page for discussion. I am hoping that
we might be able to track how many pending edits are made by anonymous/newly
registered editors versus autoconfirmed editors, though, and what percentage
of edits by autoconfirmed editors winds up being held because of an earlier
pending revision.
We really do need some hard numbers here, so that the community can make
informed decisions about the results of this trial.
Risker/Anne
I