[WikiEN-l] Some more unscientific findings

Geoffrey Burling llywrch at agora.rdrop.com
Fri Dec 9 18:51:30 UTC 2005


I've been spending some of my time over the last few days looking for 
articles on Wikipedia that could be added to [[Wikipedia:Good articles]], 
& noticed an interesting pair of statistics.

The first statistic comes from the results of a number of Random page 
tests that have been performed & recorded over the past. From the list 
at [[Category:Random Page Tests]], & adding to this my own survey of 
20 pages found at random & Alphax's survey of 10 he posted to the 
list, I came up with a sample of 230 pages, 16 of which the participants 
rated as Good or better, which works out to a proportion of a little 
more than 5%. This sample is large enough I feel (based on my own 
admittedly feeble knowledge of statistics) to suggest that about 5% of 
all Wikipedia articles are of "good" quality or better.

Now while some people will use this as proof that Wikipedia badly needs 
improvement and eoter rant that we need to work on our quality, or 
dismiss this finding as belaboring the obvious, a little math will show 
that if the 5% figure is correct then there ought to be at least 
42,000 articles that are Good in quality -- or better. That's not a 
bad statistic, especialy when one considers some proprietary encyclopedias 
don't have that many articles *in total*.

The second statistic comes from my search for Good articles. Now I 
admit that I like to find the easiest solutions to problems, & the 
hardest way to find Good articles would be to click on the "Random 
article" link, & hope that at least every 20th article is worth listing. 
A far easier way would be to find the articles that have gone thru 
the Featured Articles Candidate process, & examine the rejects to determine 
whether any would be a "Good article." And so last night I started to 
add up all of the articles listed in [[Category:Old requests for peer review]] 
& found this category contained only 1328 articles. In other words, for 
every *40* Good articles in Wikipedia, only *1* article gets any 
special attention to be singled out for consideration and review!

And it's a plausible assumption that a good chunk of the articles 
submitted to Peer review should never have been considered a "Good 
article" to begin with -- which makes the number of Good articles found 
& considered for this process even smaller.

Let's stop making assumptions, and do some more math: more articles 
are listed & debated on AfD than on Peer review. It's hard not to 
conclude from that simple calculation that Wikipedians are more motivated 
to remove an article than to improve it. Small wonder, then, that so 
many people complain about the quality of the articles on Wikipedia: 
that's  what they look for -- the bad articles instead of the good.

I know this would be unenforceable, but wouldn't it be a good policy 
that before an editor nominates an article to AfD, they have to also 
nominate an article to Peer review? And that before an editor adds to 
the debate at AfD, they have to contribute to the debate at either 
Peer review or Featured articles candidates? Maybe that might make some 
people complain about those fora like some people complain about AfD now, 
but it just might help uncover more of those 42,000 unrecognized Good 
articles.

Geoff




More information about the WikiEN-l mailing list