Editor retention

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

Editor retention

Gwern Branwen
http://online.wsj.com/article/SB10001424053111904875404576532431335938862.html
September 2011

> Adding more editors “is one of our top priorities for the year,” says Howie Fung…aims to increase the number of editors across all languages of Wikipedia to 95,000 from 81,450 by June of next year.

From http://stats.wikimedia.org/EN/TablesWikipediaZZ.htm using the >5
edits a month metric used in WMF docs:

1. July 2012: 76,400
2. June 2012: 74,402
3. May 2012: 76,956
4. April 2012: 75,141
5. March 2012; 76,274

The high water mark, incidentally, seems to have been March 2007 with
90,618 editors >5 edits that month. So we have been shrinking ~2.8k
editors a year  ((91 - 77) / (2012 - 2007)). In retrospect, my 75%
prediction that this priority would not be achieved
(http://predictionbook.com/predictions/3241) was ludicrously
optimistic, given that the 95k editor mark has *never* been reached.

--
gwern
http://www.gwern.net/In%20Defense%20Of%20Inclusionism

_______________________________________________
WikiEN-l mailing list
[hidden email]
To unsubscribe from this mailing list, visit:
https://lists.wikimedia.org/mailman/listinfo/wikien-l
Reply | Threaded
Open this post in threaded view
|

Re: Editor retention

Andrew Gray-3
On 4 September 2012 21:01, Gwern Branwen <[hidden email]> wrote:

> http://online.wsj.com/article/SB10001424053111904875404576532431335938862.html
> September 2011
>
>> Adding more editors “is one of our top priorities for the year,” says Howie Fung…aims to increase the number of editors across all languages of Wikipedia to 95,000 from 81,450 by June of next year.
>
> From http://stats.wikimedia.org/EN/TablesWikipediaZZ.htm using the >5
> edits a month metric used in WMF docs:
>
> 1. July 2012: 76,400
> 2. June 2012: 74,402
> 3. May 2012: 76,956
> 4. April 2012: 75,141
> 5. March 2012; 76,274
>
> The high water mark, incidentally, seems to have been March 2007 with
> 90,618 editors >5 edits that month. So we have been shrinking ~2.8k
> editors a year  ((91 - 77) / (2012 - 2007)). In retrospect, my 75%
> prediction that this priority would not be achieved
> (http://predictionbook.com/predictions/3241) was ludicrously
> optimistic, given that the 95k editor mark has *never* been reached.


I don't disagree with the overall results - editor numbers are still
in decline - but I think it's worth including the caveat that the
numbers reported on the wikistats site have recently been adjusted
downwards by around 5% -
http://blog.wikimedia.org/2012/08/31/improving-the-accuracy-of-the-active-editors-metric/

The result is that Howie's quote above is doubly unlikely - it's based
on an inflated estimate of how many editors we had then. Our figures
for Aug 2011 are now 76,126 rather than the 81,450 quoted; adjusting
his target accordingly, this would make it around 89,500. Still a long
way to go, though, whichever you use!

I've quickly graphed the recent numbers (just focusing on the period
after the initial exponential growth, so starting in 2007):

https://commons.wikimedia.org/wiki/File:Wikipedia_editor_numbers_2007-2012_recalculated.svg

It seems that the rate of decline may be slowing, at least. Which I
guess is something.

The really interesting part comes from separating out the data for the
English Wikipedia, which is 30% down from its 2007 peak, down from
50,000 editors to 35,000 - the graph is smoother, but otherwise looks
very similar to the first one.
https://commons.wikimedia.org/wiki/File:Wikipedia_editor_numbers_2007-2012_recalculated_-_English_only.svg

In fact, the decline on en.wp is so steep that it actually represents
almost all the overall decline in the projects; if you look at the
totals for every project except en.wp, the graph jumps around a bit
but is actually quite stable. Since the start of 2008, the three-month
moving average has always been 40,000 editors +- 5%.

https://commons.wikimedia.org/wiki/File:Wikipedia_editor_numbers_2007-2012_recalculated_-_Non-English_Wikipedias_only.svg

One last interesting point: the 2010 drop was mostly a non-en.wp
event; the drop on en.wp was proportionally much less. I have no idea
as to the likely cause of this.

--
- Andrew Gray
  [hidden email]

_______________________________________________
WikiEN-l mailing list
[hidden email]
To unsubscribe from this mailing list, visit:
https://lists.wikimedia.org/mailman/listinfo/wikien-l
Reply | Threaded
Open this post in threaded view
|

Re: Editor retention

Gwern Branwen
On Tue, Sep 4, 2012 at 6:22 PM, Andrew Gray <[hidden email]> wrote:

> I don't disagree with the overall results - editor numbers are still
> in decline - but I think it's worth including the caveat that the
> numbers reported on the wikistats site have recently been adjusted
> downwards by around 5% -
> http://blog.wikimedia.org/2012/08/31/improving-the-accuracy-of-the-active-editors-metric/
>
> The result is that Howie's quote above is doubly unlikely - it's based
> on an inflated estimate of how many editors we had then. Our figures
> for Aug 2011 are now 76,126 rather than the 81,450 quoted; adjusting
> his target accordingly, this would make it around 89,500. Still a long
> way to go, though, whichever you use!

Whups.

>  One last interesting point: the 2010 drop was mostly a non-en.wp event; the drop on en.wp was proportionally much less. I have no idea as to the likely cause of this.

Perhaps the damage has already been done on En? I would've suggested
that maybe the WMF retention initiatives might have not failed
entirely, except I don't remember any of them being finished in 2010,
much less being able to affect the overall wiki so much.

--
gwern
http://www.gwern.net

_______________________________________________
WikiEN-l mailing list
[hidden email]
To unsubscribe from this mailing list, visit:
https://lists.wikimedia.org/mailman/listinfo/wikien-l
Reply | Threaded
Open this post in threaded view
|

Re: Editor retention

WereSpielChequers-2
One of my experiences from the death anomaly project is that different
language communities vary quite sharply as to their tolerance of unsourced
edits.

I suspect that by 2010 the English Wikipedia had already gone through the
transition on this, and that those editors we were going to lose by having
their unsourced edits reverted had largely gone. We still lose a large
proportion of newbies, and I think we should find better ways to
communicate with those who make unsourced edits. But I can understand that
this would lead to EN wiki being further on the path to a smaller community
that edits to a more rigorous standard.

What would be interesting would be to chart our different communities and
their retention rates by the point that they went through various
transitions. This could test both the theory that reversion of goodfaith
but unsourced edits is causing the change, and it would test and I hope
disprove the theory that flagged revisions deters new editors.

The other big change that has happened on EN and is EN specific is the
growth of other projects that may have syphoned away a large proportion of
editors who started in EN. Has anyone used SUL to chart patterns of editors
shifting their project focus? My suspicion is that many formerly active EN
Wikipedians are still active elsewhere in Wikimedia, it would be
interesting to know which were our recruiter and destination projects and
how that has changed over time.

WSC,

On 5 September 2012 01:57, Gwern Branwen <[hidden email]> wrote:

> On Tue, Sep 4, 2012 at 6:22 PM, Andrew Gray <[hidden email]>
> wrote:
> > I don't disagree with the overall results - editor numbers are still
> > in decline - but I think it's worth including the caveat that the
> > numbers reported on the wikistats site have recently been adjusted
> > downwards by around 5% -
> >
> http://blog.wikimedia.org/2012/08/31/improving-the-accuracy-of-the-active-editors-metric/
> >
> > The result is that Howie's quote above is doubly unlikely - it's based
> > on an inflated estimate of how many editors we had then. Our figures
> > for Aug 2011 are now 76,126 rather than the 81,450 quoted; adjusting
> > his target accordingly, this would make it around 89,500. Still a long
> > way to go, though, whichever you use!
>
> Whups.
>
> >  One last interesting point: the 2010 drop was mostly a non-en.wp event;
> the drop on en.wp was proportionally much less. I have no idea as to the
> likely cause of this.
>
> Perhaps the damage has already been done on En? I would've suggested
> that maybe the WMF retention initiatives might have not failed
> entirely, except I don't remember any of them being finished in 2010,
> much less being able to affect the overall wiki so much.
>
> --
> gwern
> http://www.gwern.net
>
_______________________________________________
WikiEN-l mailing list
[hidden email]
To unsubscribe from this mailing list, visit:
https://lists.wikimedia.org/mailman/listinfo/wikien-l