Re: screwy statistics

Brion Vibber <[email protected]> Wed, 24 Dec 2003 23:36:22 -0800
Newsgroups gmane.science.linguistics.wikipedia.international
Message-ID <[email protected]>
On Dec 24, 2003, at 21:19, Jay Bowks wrote:
>> Who takes these statistics, how? Have there in fact been deletions?
>
> The stats I found were at:
> http://en2.wikipedia.org/wiki/Wikipedia:Multilingual_statistics

Short story: there's a gap in all stats for August and September. Then 
there's an unusually dense period of statistics recorded for October, 
with the ia figures very high for October 1-20 and then dropping by 
about 1600 between data points at October 20 and 25. The rate of growth 
over October-December is fairly steady if you discount the one-time 
drop.

I note from my mail archives that I gave you (Jay and Ensjo) the admin 
password for ia.wikipedia.org on October 20. Did you guys do a lot of 
clean-up work perhaps, getting rid of junk pages or such?

Note that we don't store access logs that far back.

At the time of conversion, the page count for ia.wikipedia.org was:
1757 (old-style comma count)
2859 (new-style links count; this does not exclude redirects)
3232 (total pages)

This is consistent with the recorded stats for October 25-December 19. 
The total size is too small to account for an accidental miscalculation 
of the October 1-20 stats (such as forgetting the "]]" and accidentally 
searching all pages), so either the high stats were _totally_ wrongly 
collected somehow, or a lot of pages were deleted between October 20 
and 25.

Compare the generated statistics:
http://download.wikimedia.org/wikistats/EN/ChartsWikipediaIA.htm

This shows a jump up in August, then consistent growth from then on. 
These come from the current state of the database (including history), 
so will not be able to count any pages that were since deleted.

> Paolo Castellina complained to
> me that a great number of his
> articles were missing, I took it
> to imply that these were deleted.

If you can point me to a specific list, that'd be great.

Also, note that the user contributions list for a post-conversion 
account doesn't list edits by the user pre-conversion unless the user 
ID numbers have been corrected in the database.

> How does one check for deletions
> in the new software?

The deletion log tracks all deletions, which should be at:
http://ia.wikipedia.org/wiki/Wikipedia:registro_de_eliminationes

Also, to restore a deleted page:
http://ia.wikipedia.org/wiki/Special:Undelete

-- brion vibber (brion @ pobox.com)

_______________________________________________
Intlwiki-l mailing list
[email protected]
http://mail.wikipedia.org/mailman/listinfo/intlwiki-l
PGP.sig (application/pgp-signature, 186 B) - not displayed