On 15 June 2010 14:54, Andrew Gray andrew.gray@dunelm.org.uk wrote:
On 15 June 2010 19:52, Gregory Maxwell gmaxwell@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