I don't know the details of their methodology either. The delta is certainly high, maybe suspiciously so. Most likely it's a combination of multiple factors: 1

I don't know the details of their methodology either. The delta is certainly high, maybe suspiciously so. Most likely it's a combination of multiple factors: 1. Actual increased Linux usage 2. Reduction in Windows usage (replaced by mobiles / tablets) 3. Systematic measurement errors, for example a change in which websites use their ad network. For a different data set, here's a plot of a few years of StatCounter global desktop usage data (just Linux and ChromeOS), http://jsnell.iki.fi/tmp/statcounter.png http://jsnell.iki.fi/tmp/statcounter.png . There are definitely 5 month periods there that could be used to show similar absurd growth ratios, but the overall growth is a lot slower that that (even if clearly trending upwards). Even so, on the whole the numbers look pretty consistent with my first link.