Friday, April 23, 2010

"Results from the past offer no guarantee for the future ..."

This is part of a mandatory quote when products of the financial sector are advertised in the Netherlands. In life insurance, it is calculated what average result can be expected. As an individual you can improve your odds with a proper diet, exercise and staying out of any firing lines.

When you plan a strategy for a multi-faceted organisation like the Wikimedia Foundation, numbers give a baseline, something to measure against. However as we decide to focus on what has the biggest impact, what is known to work best, we ossify. We expect all our smaller Wikipedias to grow in a similar way, some do and some don't.

We do know that we can influence results but in order to do this we first need to be convinced what is to be done. Numbers will not help us because they at best show the impact after the fact. A good example is the effect of the LocalisationUpdate extension. It has had a measurable impact on the quantity of MediaWiki localisations at and, there are indications that this had an impact on the traffic on some Wikipedias but sadly there are no hard numbers that establish a relation between traffic and localisation.

The aim of the Wikimedia Foundation is to increase the reach of our projects. This means that we want to have more traffic. When you look at the official WMF documents, you find this list that is incomplete. The reason given is that the source for these numbers is not that great outside of the first world. It is however not the only source, Alexa is more informative. So with numbers from elsewhere, we have more of a clue about our impact in a given country and consequently we are better able to "measure" the evolution of our impact.

When numbers are available, we can look for anomalies. Why does the Indonesian Wikipedia outperform all the Indian Wikipedias combined. According to Alexa the impact of Wikipedia is higher in India.. and, it is not that they do not know their English in Indonesia. Problematic is the support for the scripts of the Indian languages. This showed itself when we tried to support  the mobile Wikipedia interface and it showed in the problems realising the 500 article Malayalam Wikipedia CD. The Indonesians have fewer editors but many more readers ...

There may be other factors as well but, is it not simpler to fix the known issues first? We know that many of our languages are badly supported, the fact that Wikipedia supports more languages then the CLDR makes that sufficiently obvious. What does it take to support our languages? What does it take to overcome the bias that has us only support the Wikipedias that are already big? What does it take to make Wikipedia a platform that will truly bring the sum of all knowledge to everyone?

This does not completely answer the question, what does it take to invalidate a strategy. More is needed to do that.
Post a Comment