Jimmy Wales wrote:
Guy Chapman aka JzG wrote:
It worked, though.
It is not clear to me that it did. I would love for us to have some serious analysis of that.
I don't mean to be contentious on this issue, but whenever it comes up it kind of bugs me that no serious analysis was apparently even _planned_. One can't really call something an "experiment" if one doesn't intend to analyze at the results.
It seems to me that there should be some relatively straightforward facts and figures that could be dug out of the edit history database on this. Perhaps a graph of the average age of deleted articles would do for a start; a drop in speedy deletion compared to other forms of deletion should result in an increase in the average age of articles upon deletion.
My sense is that the number of articles created by unknown people is about the same, but that they now sign up for an account first. This is not helpful, because whereas before we had the rough indicator of "ip number equals newbie" (imperfect), we now have less of an indicator.
In this case we'd see a significant increase in the number of new accounts that make just a few edits after creation and then are abandoned.
There is also the question of whether the net production of new articles by ip numbers was sufficiently worthwhile to mean that preventing those cost more than it was worth.
A bit harder to judge numerically since it involves "worth" judgements, but how about looking at the rate of creation of articles that aren't deleted within a week of their creation? That gives articles time to go through the full AfD gauntlet, which (perhaps debatably :) acts as worth-based filter. All ideas are just off the top of my head, of course.
Fortunately thanks to Wikipedia's retention of full edit and deletion histories one can still do detailed analysis after the fact. But IMO this should have already been prepared for before the switch was thrown.