Wednesday, August 10, 2011

For the love of #statistics

Planning activities based on statistics is like driving a Formula One car based on what can be seen in the rear mirrors. The problem is that it does not show what is ahead and, it does not show everything what is behind.

The Wikimedia Foundation is increasingly data driven and, consequently it is vital to assess the quality of the data that policies and developments are based on. The data is based on the results of traffic to computers and mobile phones from Wikimedia servers. At the same time the data of Wikipedia, maybe all the Wikipedias, is available to organisations like Facebook. As Wikipedia articles become better available in such a way, it does affect not only the reach of Wikipedia but also the ratio between readers and editors.

Browsers are not the only way to read data from Wikipedia; there are apps for tablets and mobiles that make use of an API to retrieve articles from the WMF servers. These numbers are not included in the statistics either. As apps for tablets and phones are increasingly popular, it is not unlikely that they are yet another factor that confuse a full understanding of the status of the wiki.

When language support is to be provided for all our users, it is important to know to what extend all platforms are used. We know that mobile use is growing rapidly and the numbers of tablets sold is also quite staggering. What should be given priority for support and for language support.. What numbers exist to support whatever choice properly.
Thanks,
       GerardM

1 comment:

the wub said...

The answer to bad data isn't to bemoan a data-driven model. The answer is to *get better data*