Interesting topic! Here is a useful analogy regarding the distribution of sizes. There
has been study of how big cities are within countries or worldwide, and there are
recurring patterns of the scale of the largest to the second largest, and the
second-largest to the third, and so forth.
Without getting into this too deeply you might at least check if the size relations among
Wikipedias are like those of cities, that is, if they have a similar-looking distribution.
If they do, the underlying forces and dynamics for city sizes might also apply to
wikipediae or other sites.
The math is described by Zipf’s law and/or Gibrat’s distribution.
https://en.wikipedia.org/wiki/Zipf%27s_law
<https://en.wikipedia.org/wiki/Zipf's_law>, and
https://en.wikipedia.org/wiki/Gibrat%27s_law
<https://en.wikipedia.org/wiki/Gibrat's_law>. The work by Xavier Gabaix, cited
there, was my introduction to it.
Like the choice of what city to move to, the relevant Wikipedias for a user will usually
need to be “close” — geographically for a city, or to the languages the user knows for a
Wikipedia. There are other factors driving a user’s choice, if we think of the user as
choosing. If the user wishes to study an obscure academic subject, they may have to use a
large wikipedia, and that drives them to also participate there. If the user is focused
on a geographically local subject, that drives the choice. A larger wikipedia is more
useful than a small one, therefore the distribution of wikipedia sizes would be more
unequal than the distribution of personal languages.
It sounds like, based on Poland and Korea, you can show that Internet availability is not
driving all the difference. Good to know. — peter meyer
On Jul 24, 2018, at 11:30 AM, James Salsman
<jsalsman(a)gmail.com> wrote:
Why do you think different language
Wikipedia's have different
sizes, outside of the popularity of a given language?
Piotr, if you model organic editing production with a Poisson
distribution, which is reasonable for a first approximation, 3x+
disparities are just natural for the same population sizes:
https://en.wikipedia.org/wiki/Poisson_distribution
I'm not sure the images in that article capture the wide platykurtosis
of large Poisson distributions.
Best regards,
Jim
_______________________________________________
Wiki-research-l mailing list
Wiki-research-l(a)lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/wiki-research-l