[Foundation-l] Social networking (was: Analysis of lists statistics: community in decline)

John Vandenberg jayvdb at gmail.com
Tue Nov 4 13:46:32 UTC 2008


On Tue, Nov 4, 2008 at 11:13 PM, Milos Rancic <millosh at gmail.com> wrote:
> On Tue, Nov 4, 2008 at 12:13 PM, John Vandenberg <jayvdb at gmail.com> wrote:
>> I don't see any reason for alarm in the data that we do have.
>
> According to statistics which you gave (btw, thanks for pointing to
> them, I didn't know where to find them):
>
>> http://stats.wikimedia.org/wikispecial/EN/ChartsWikipediaCOMMONS.htm
>
> Commons is in a constant and significant decrease since May 2007.

Most people have "found" Commons by now.  "Binaries" continues to climb.

>> http://stats.wikimedia.org/wikispecial/EN/ChartsWikipediaINCUBATOR.htm
>
> In not so strong decrease since January 2008 (but we don't have data
> after May 2008)

Edits per month continues to climb.

Incubator fluctuates as projects migrate, and groups of people will
arrive and leave together; as a result we would need to understand how
this affect those stats in order to make good deductions from them.

>> http://stats.wikimedia.org/wikispecial/EN/ChartsWikipediaSOURCES.htm
>
> Old Wikisource is not so big project and it is not possible to make
> precise conclusions.

We'll come back and look at this one in a year! ;-)

>> http://stats.wikimedia.org/wiktionary/EN/ChartsWikipediaZZ.htm
>
> All Wiktionaries together stay well, this is true.
>
>> http://stats.wikimedia.org/wikisource/EN/ChartsWikipediaZZ.htm
>> http://wikisource.org/wiki/Wikisource:ProofreadPage_Statistics
>
> It seems that all Wikisources together had begun decrease at the
> beginning of 2008. However, according to the second link, it seem that
> they stays well.

The only stat going down is "new wikilibrarians".  The number of
Contributors continues to climb.  The RC feed is increasingly becoming
impossible to monitor; I'm not imagining things!.

> (BTW, I would like to see a short explanation of the
> significance of ProofreadPage extension and pages which used them.)

A "page" in those stats indicates a page that has an accompanying
image of the *original printed page* , which means that
1. anyone can transcribe the text (even without understanding the language)
2. the rest of the world can know with 100% certainty that our edition
is perfect, and has  accurate bibliographic and provenance
information.

> BTW, again, number articles *will* raise except there are big
> problems. One new page per month means that there is one article more
> and somewhat bigger database. I explained in one of the previous
> emails [1] why some data are more relevant than others. (If you have
> objections to this approach, please let me know what are the errors of
> the method.)

Your focus on stats on "users" leads to bad results.  All languages
have a finite number of people that understand them, and the graph of
new contributors is indicative of the gradual growth of the wiki into
that population.  When a wiki is small, the population doesnt know
about it.  When the wiki is large, the majority of the population
knows about it, and most will have already decided whether they wish
to participate or not.  So, I dont put much weight on stats of new
contributors.  Also, most newcomers dont get the "wiki" bug.  They
deposit one or two pages, and then go away happy.

The number of active contributors is more important, but is still
indicative of the stage the wiki is at, in relation to public
awareness.

I understand that you were using stats about users to learn something
about the health of the "community", and can see some value in it,
however I much prefer to look at the content related stats : the
growth of the wiki.  The content.  And all indicators there are
looking OK on the projects.

I fail to see what is the problem when all of the indicators show the
content namespaces are growing, even if it is a linear growth.  We
know that contributors often leave, but new people are filling their
places, or the old people are being more productive.

The more difficult aspect to measure is the quality.  For example, the
German Wikisource stats look like they are having a hard time... their
stats fluctuate a lot.

http://stats.wikimedia.org/wikisource/EN/ChartsWikipediaDE.htm

The reality is that they have been actively turning away contributors,
because they have decided that they will not accept any text that isnt
accompanied with page scans.  Most people are not so dedicated that
they will go to such lengths.  I think it is a bad decision, but the
result is that they have very good quality throughout their wiki, and
the project members are more proud of their work, because they are
working in a very orderly environment.

Quality attracts a different class of new contributor -- a rarer
breed, but more likely to make highly valuable edits.  But quality is
_hard_, and enforcing quality results in less new contributors.

> And, again, I would be really happy to see that I am wrong. I didn't
> spend significant time in analysis just because I like to spread
> defeatism; but to point to the problem.

The most important problem is that the statistics are stale.  If you
want to make big decisions, you need good data, and analyse it from
many angles.

--
John Vandenberg



More information about the foundation-l mailing list