[Wikimedia-l] Editor retention (was Re: "Big data" benefits and limitations (relevance: WMF editor engagement, fundraising, and HR practices))

[Federico Leva (Nemo)]

Andreas Kolbe, 10/01/2013 17:24:
> On Wed, Jan 9, 2013 at 6:41 AM, Federico Leva (Nemo) wrote:
>
>> The main pattern, ie a turning point in 2007, is the same in all projects,
>> and almost in all language versions of them: [...]
>
> Actually, Nemo, I don't think that is right at all. If you look at the
> German, Spanish or French Wikipedia, for example, the German and Spanish
> are totally stable, with no decline at all discernible around 2007, while
> editor numbers for the French Wikipedia are actually growing:
>
> http://stats.wikimedia.org/EN/TablesWikipediaFR.htm
> http://stats.wikimedia.org/EN/TablesWikipediaDE.htm
> http://stats.wikimedia.org/EN/TablesWikipediaES.htm

I said "a turning point", i.e. a singularity; mainly, from positive to 
non-positive derivative, whether negative or not. Of course, it's easier 
to see in a graph than in a table.
I don't see French growing: except an outlier in November 2012 for 
active editors, which is not reflected in the very active editors count, 
in the last few months it's at the same level as in January-March 2008, 
4800-5000 active editors.
It's the same in Italian, growth till January-March 2008 and then 
oscillation/stagnation: http://stats.wikimedia.org/EN/TablesWikipediaIT.htm
Anecdotally in WMIT, we've been repeating "it.wiki has 500 very active 
editors" for a while, and we've stopped updating this figure a long time 
ago. :-)

Of course I'm only playing the stats dilettante here.

> Summaries and charts for all projects are available here:
>
> http://stats.wikimedia.org/EN/Sitemap.htm
>
> These three projects are of a similar age to the English Wikipedia, and
> they are definitely not following the same editor retention pattern at all.

You're using the "editor retention" term quite incorrectly by the way: 
those tables show only total active editors, old or new, not how many of 
the new editors are still active, nor how many "really new" editor we had.

Richard Symonds, 10/01/2013 17:37:
 > Which column are you looking at to give you the growth numbers on those
 > projects?

I think I've replied already. :-)
"New editors" is not reliable because one edit is enough, number of 
edits or (new) articles have too much bot noise, database size/words is 
often useful but even more often not available for WikiStats performance 
limitations.

Nemo