2013/1/22 Ravishankar <ravidreams@gmail.com>

Arjuna,

Your approach to have an idea of a community's activity by combining some metrics is interesting. Also, to compare monthly averages instead of end of year performance is also good.

Thanks.
One clarification.
I took the sum of the metrics for the entire year rather than average.

Increase in database size, number of most active contributors (making more than 100 edits a month) and page views can be real indicators of growth and activity. Is there any way to find database size? We don't have that and many other useful stats after May 2010. Check http://stats.wikimedia.org/EN/TablesWikipediaTA.htm for example.

Yes and No, as any metric can be compromised. If database size is a parameter, too many stub articles may adversely affect the quality.

Number of most active contributors is too small a number in most Indian language Wikipedias. There has been debate about the % of people who contribute to Wikipedia in English. While Jimmy is reported to have said that it is small fraction which contributes the bulk, Aron Swartz did an analysis that the bulk of contribution is made by large number of wikipedians based on the bytes added metric. In wikipedia, I think the former is more true, though it could be different if there are people who are contributing stubs rather than reasonable sized articles.

You can get an account on Toolserver and obtain the database size information.

Also, any inference based on page views should normalize it based on the population of native speaking people to have a better perspective of the community's performance compared to its potential / size.

My analysis is more focussed on relative change in each language than absolute numbers. The graphs included the information for all the languages to get a general feel of the languages together.

It is not the population alone which can be a significant factor, as there are several other factors like the love for language and geographic distribution of its speakers. Population numbers could be considered, but then there is lot of variation depending on the sources.

An exhaustive study would be useful, if we can determine a independent measure (may be by annual survey) as an impact of Wikipedia in accomplishing its mission, then the various metrics available like edits, page views and their combinations can be tried using statistical algorithms to arrive at key parameters or parameter combinations.

I think it would be good for WMF to work on such a thing, as otherwise our focus could be lost by looking at simple metrics like active editor/page views.

Cheers
Arjuna