Having tired of listening to arguments about what goes on at AFD but having little data, I decided to engage in a little experiment to analyze voting patterns at AFD. I wrote a computer script to parse over 100 days of AFD logs and compile a variety of statistics by looking at the bold faced terms (eg. keep, delete, merge, etc).
Obviously this is imperfect since AFD is suppose to be about discussion and consensus building (somehow despite somehow having 100 noms and 875 votes cast a day), but it is good enough to reveal some interesting patterns. For example, I would not have guessed that on a forum with 7200 participants over 100 days that the top 5 would account for 10% of all votes. Or that for all articles with 2/3 of the votes cast favoring deletion, only 6 would also have 10 or more keep votes.
However, I am not all that surprised that regulars cast delete votes 76% of the time whereas occasional participants (the vast majority of all voters) say delete only 56% of the time.
The full listing of statistics compiled so far can be seen at:
http://en.wikipedia.org/wiki/Wikipedia:AFD_100_days
Things are still somewhat preliminary as I may yet try to further improve the parser and the last three days of my AFD sample have not entirely closed, but I doubt those things with substantially change the patterns.
Anyway, I hope this provides a better grounding for some of the endless discussions about what to do with AFD (and whether or not it has serious problems).
-Dragons flight
PS. I need to get back to real life now, so if there are questions/requests/complaints, it will probably be several hours or a day before I get back to respond to them.
Well, since keepable articles shouldn't end up on AFD to begin with the number of delete votes don't really surprise me. The amount of votes the top 5 provide are stunning, though. I really wonder how they read all those articles they're voting on and still have time to do other stuff.
--Mgm
On 9/19/05, DF dragons_flight@yahoo.com wrote:
Having tired of listening to arguments about what goes on at AFD but having little data, I decided to engage in a little experiment to analyze voting patterns at AFD. I wrote a computer script to parse over 100 days of AFD logs and compile a variety of statistics by looking at the bold faced terms (eg. keep, delete, merge, etc).
Obviously this is imperfect since AFD is suppose to be about discussion and consensus building (somehow despite somehow having 100 noms and 875 votes cast a day), but it is good enough to reveal some interesting patterns. For example, I would not have guessed that on a forum with 7200 participants over 100 days that the top 5 would account for 10% of all votes. Or that for all articles with 2/3 of the votes cast favoring deletion, only 6 would also have 10 or more keep votes.
However, I am not all that surprised that regulars cast delete votes 76% of the time whereas occasional participants (the vast majority of all voters) say delete only 56% of the time.
The full listing of statistics compiled so far can be seen at:
http://en.wikipedia.org/wiki/Wikipedia:AFD_100_days
Things are still somewhat preliminary as I may yet try to further improve the parser and the last three days of my AFD sample have not entirely closed, but I doubt those things with substantially change the patterns.
Anyway, I hope this provides a better grounding for some of the endless discussions about what to do with AFD (and whether or not it has serious problems).
-Dragons flight
PS. I need to get back to real life now, so if there are questions/requests/complaints, it will probably be several hours or a day before I get back to respond to them. _______________________________________________ WikiEN-l mailing list WikiEN-l@Wikipedia.org To unsubscribe from this mailing list, visit: http://mail.wikipedia.org/mailman/listinfo/wikien-l
Interesting stuff. James Burns has basically made a career out of AfD, so I'm not surprised to see he is the top contributor in this sample. I think his contributions and Etacar's and mine are probably distorted somewhat by selective voting. For instance I won't bother voting at all on most nominations because they're going to be deleted anyhow and I agree with the deletion. The few that I do look at and vote delete on are generally cases where I think it look marginal and I investigate. Probably most of the 500+ keeps I made were related to topics of interest to me. Programming languages, schools, roads. Stuff that I think should be in the Library of Babel.
Also I have the worst judgement for nomination of anyone I know. If I nominate an article for deletion, it will probably end up as a FAC within a month.
Anyway, I hope this provides a better grounding for some of the endless discussions about what to do with AFD (and whether or not it has serious problems).
Wow... I didn't realize I had a 77% delete (I'm #107, whee!)... of course that includes my early "deletionist" days so it's not too surprising.
Etacar has a 95.9% delete rate... my word :).
That's the coolest thing ever... thanks DF!!!
Thanks, RN
hmm I suspect the low delete rate amoung no regulars may be infulence by people trying to protect the articles they have created.
On 18/09/05, DF dragons_flight@yahoo.com wrote:
Having tired of listening to arguments about what goes on at AFD but having little data, I decided to engage in a little experiment to analyze voting patterns at AFD. I wrote a computer script to parse over 100 days of AFD logs and compile a variety of statistics by looking at the bold faced terms (eg. keep, delete, merge, etc).
That's a magnificent piece of work, thank you very much!
Any chance the script could analyse the data to see how votes end up being closed? I.e. the % delete votes articles which actually are deleted get?
Dan