[Foundation-l] Analysis of lists statistics: community in decline

Thu Oct 30 11:28:43 UTC 2008

The problem we have is well measuring the community and the activities. I am
looking forward that the UNU-MERIT survey will be a huge step to knowing
"us" better.

Activity on malinglists and Wikimedia Statistics can only give hints, but
not tell us about the quality of mails and edits. This is again trying to
read something from Statistics that cannot be read from them.

A decrease of the total number of edits in a WP language edition can mean,
for example, a decrease of vandalism. Or: WP rules have been discussed
largely, now Wikipedians do less discussions. Or: Wikipedians have learned
to do more within one single edit when writing an article etc. Or: The
Poplar Bluff syndrome (bot generated geographical stubs / pseudo articles)
with its aftermath has settled.

A "New Wikipedian" can be simply a vandal, having made 5 edits in a month
(en.WP will not ban him so fast). Maybe the potential vandals have lost the
fun, or are all blocked, that's why we have less new Wikipedians etc.
It is also natural that people create less articles or edit less because
many articles already exist and are well written.

It may be true that "there is a stagnation or decline", but I cannot see
substantial evidence for ... well, what is it exactly we mean that shows
"stagnation" or "decline"?

Ziko van Dijk

2008/10/30 Milos Rancic <millosh at gmail.com>

> On Thu, Oct 30, 2008 at 10:07 AM, Yaroslav M. Blanter <putevod at mccme.ru>
> wrote:
> >>> They do not use any mailing lists at all as far as I know. At least I
> am
> >>> not aware of existence of any mailing lists for ru.wp
> >>
> >> First, as you (Yaroslav) is active here, I would like to know what
> >> Russian Wikipedians are using for communication. Just wiki? Some other
> >> ways of communication? Wikipedia in Russian is not a small project, as
> >> well as it is growing -- which demands some level of systematic
> >> coordination. I think that the answer on this question may be very
> >> significant at least in understanding some part of lists traffic
> >> decrease.
> >>
> >
> > The main channel of communication is wiki: basically, the village pump
> > which is structured as a number of pages, for instance for instance,
> > general, news, rules, technical issues and such. So far it worked.
> > Important issues are branched out as separate rfc's.
> >
> > Some of the Wikipedians (not me) are full-time present on irc channel.
> >
> > A creation of a mailing list for sysops only was recently proposed but
> was
> > eventually rejected.
> >
> > There is of course personal communication, and there are even some
> > Wikipedians (not me again) who insist that most of the issues should be
> > discussed privately.
> >
> > The Russian chapter is in the creation stage and does not have its own
> > communication channel as far as I know.
> >
> > I am not aware of any other signifgicant communication channels (I
> > recollect there is a collective blog which is half-dead and there is a
> > decision of arbitration committee that this blog is not a part of the
> > community), may be there are some more I do not know of, but it is
> > unlikely. There are information channels of course, for instance, similar
> > to wikizine. but they do not involve any discussions and are not
> > widespread (at least now).
> >
> > Some of the prominent ru.wp editors are subscribed to this list, I assume
> > they will correct me if they have a different perspective.
>
> Hm. At the first sight, I haven't found anything interesting. But, I
> took a look to statistics (stats.wikimedia.org) and I realized that
> one of the possible answers is laying in ru.wp community.
>
> First, I didn't see anything special; just different communication
> channels. But, it may be the part of the answer, too. Using wiki as a
> communication channel is more productive.
>
> Then I went to stats (statistics are from May 2008) and compared the
> data for the next Wikipedias:
> - (English and German are not so relevant because statistics are old.)
> - I took French and Italian as "the ordinary cases" -- as their lists
> show decline.
> - Russian and Japanese are "not so ordinary cases".
> - I took Serbian, too, to compare data with my knowledge.
>
> So, before I started, I had supposed the next:
> - French [1] and Italian [2] should show signs of stagnation and
> decline (according to the data from the lists).
> - Serbian [3] should show signs of stagnation and decline (according
> to my knowledge: it is in the similar situation as other "ordinary
> cases").
> - Russian [4] should show signs of stagnation: it started to raise,
> but it should be around the peak. So, if not decline, stagnation is
> expected.
> - Japanese [5] should show signs of low raising (according to the data
> from the list).
>
> I was just partially right:
> - I was right for French, Italian and Serbian Wikipedias.
> - Russian doesn't show signs of stagnation, but signs of linear raise
> (not exponential, like it was during the first years of Wikipedia, of
> course).
> - Japanese is in stronger decline than French, Italian and Serbian.
>
> Before the particular analysis, just to explain what which behavior
> means to us in the case of new users:
> - Linear growth: (a) in the sense of project growth: exponential
> growth (n^m); (b) in the sense of long term sustainability: linear
> growth.
> - Stagnation: (a) in the sense of project growth: linear growth (n*m);
> (b) in the sense of long term sustainability: stagnation (which is
> just fine at the position where Wikimedia is globally now).
> - Decreasing: (a) in the sense of project growth: logarithmic growth
> (n*1/m); (b) in the sense of long term sustainability: decline.
>
> Here is the explanation of importance of charts:
> - Contributors: This is not the best chart to look in. Number of
> contributors can't fall and it is reasonable to expect some raise
> every month -- if it is not about really small projects nor very big
> problems at some project. Also, changes are not so visible (analysis
> by just looking into it assumes measuring of curve angle).
> - New Wikipedians: This is a very good and visible indicator. Linear
> raising of the number assumes exponential growth; stagnation in
> numbers assumes linear growth (the best possible development for us in
> this situation), while decreasing number of Wikipedians means problems
> for us.
> - Active and very active Wikipedians: They are connected to the new
> Wikipedians. If the number of NW (per month) raises, the number of AW
> raises, if number of NW stagnates, number of AW stagnates; if number
> of NW decreases, number of AW will decrease. The connection is simple:
> some Wikipedians are leaving; if there is a number of others to
> replace them, the number will stagnate; if there are more new
> Wikipedians, the number will decrease; if there are more than enough
> newcomers, number of active Wikipedians raises.
> - Edits per month. This indicator has one more important value than
> just signaling the number of active and very active contributors. A
> lot of house keeping tasks may be done automatically, so this number
> has to be in more or less constant correlation with the number of
> articles.
> - Other charts are relatively straight-forward. One more created
> article means growing of database size and growing of number of
> articles; and similar.
>
> So, conclusion related to analyzed Wikipedias (and according to the
> data up to May 2008) is:
> - There are problems with French, Italian and Serbian Wikipedias. If
> such trends continues, we would loose sustainability there.
> - Japanese Wikipedia has serious problems; even data from the list
> (from January to May) shows different situation. Again, it should be
> good to hear, this time because the opposite reason, what is going on
> there.
> - Russian Wikipedia (up to May 2008, of course) is going fine (linear
> growth at the project level, stagnation at the long term
> sustainability). Why is it so -- it should be analyzed.
>
> The only reasons which I may detect is a very strong ArbCom. If ArbCom
> is able to say "that something is not a part of the community", then,
> AFAIK, it is stronger than (still strong) en.wp ArbCom.
>
> [1] - http://stats.wikimedia.org/EN/ChartsWikipediaFR.htm
> [2] - http://stats.wikimedia.org/EN/ChartsWikipediaIT.htm
> [3] - http://stats.wikimedia.org/EN/ChartsWikipediaSR.htm
> [4] - http://stats.wikimedia.org/EN/ChartsWikipediaRU.htm
> [5] - http://stats.wikimedia.org/EN/ChartsWikipediaJA.htm
>
> _______________________________________________
> foundation-l mailing list
> foundation-l at lists.wikimedia.org
> Unsubscribe: https://lists.wikimedia.org/mailman/listinfo/foundation-l
>

-- 
Ziko van Dijk
NL-Silvolde