[WikiEN-l] Looking for thoughts on statistics

[Keegan Paul]

On Sun, Mar 28, 2010 at 12:57 AM, Gregory Maxwell <gmaxwell at gmail.com>wrote:

> On Sun, Mar 28, 2010 at 12:49 AM, Keegan Paul <kgnpaul at gmail.com> wrote:
> > http://stats.wikimedia.org/EN/ChartsWikipediaEN.htm
> >
> > <http://stats.wikimedia.org/EN/ChartsWikipediaEN.htm>It's obvious of the
> > peak in January of 2007.
> >
> > What I'm interested in is thoughts of why New Contributors has
> statistically
> > declined sharply, but the list of active contributors has much less of a
> > slope and even less so for very active contributors.
> >
> > What happened in the first six months of 2007?  Did we change template
> > systems?  Did we reword some policies relating to new users?
>
> Careful not to mistake a decline in the derivative of a function to be
> a decline in the function. The number of new contributors _must_
> decline at some point, unless you hold a hypothesis that Wikipedia
> will eventually be driving the growth of human population. ;)
>
> The step function in December 2005 is clearly due to
>
> http://en.wikipedia.org/wiki/Wikipedia_biography_controversy#Wikimedia_Foundation_reaction
>
> The only thing I recall happening around June 2007 was the
> introduction of a real captcha.  Might be relevant if a non-trivial
> amount of the new accounts were spambot sleepers!
>
> I don't recall how those stats are generated. If they are produced
> from the public data then there will be odd distortions due to
> deletions hiding accounts..
>
> I think there were also changes to the upload procedure around that
> time (the interface language abuse for an upload wizard) which started
> directing users to commons to upload... and uploading is a primary
> reason to create an account.   This seems to be at least weakly
> supported by the stats on commons:
> http://stats.wikimedia.org/EN/ChartsWikipediaCOMMONS.htm
>
> I'd guess that like most things its probably a mixture of weak effects.
>
> _______________________________________________
> WikiEN-l mailing list
> WikiEN-l at lists.wikimedia.org
> To unsubscribe from this mailing list, visit:
> https://lists.wikimedia.org/mailman/listinfo/wikien-l
>

Oh sure, figures lie and liars figure.  Like stereotypes, from a broad lens,
they make sense.

I think captcha probably had a good deal to do with it.  Good point there to
mention.  The systemization of procedures is a good point as well, whether
it be uploading or bot-assisted and the functionality of automated tools
like huggle and twinkle.

With these thoughts in mind, the good thing is that the standard userbase
numbers are consistent.

Thanks Greg, other thoughts?

~Keegan
-- 
http://en.wikipedia.org/wiki/User:Keegan