Sample size is a bit small. With its "most popular" metric being only a small sample of top Alexa ranked sites. Loosing a single install, or failed detection can lead to massive swings of "market share". Plus comparing forum software installs to WordPress installs is since they are both "CMSs" is a bit asinine. Also note none of the other forum software like vBulletin or phpBB have enough datapoints to on that site display similar historical data. Here is a better apples-apples comparison, note I couldn't find IPB to add to the chart.
The web is growing far faster than XenForo's user-base is (or isn't, it is hard to tell from these stats) and it contains a lot more than just forums.
