And of course basing the github measure on lines of code gives quite an unfair advantage to verbose languages like Java.
Perhaps they could look at compression ratios when (say) gzipping a decent sample of each language, and use these to weight the LOC metric?
I mean, yeah, most of the comments indicate that we agree that this kind of graph or "false statistic" is flawed.
But can we find any value in it? How do we interpret this graph, despite its basic problems? What good stuff can we do with it?
Perhaps the useful information is more in the correlations between different related metrics, than in drawing up ranked lists (surprise surprise, Java > Haskell!). This graph helps visualise the correlation between these two, which seems significant but far from perfect. Outliers like SQL can then be identified, which point to problems with the metrics (e.g. with SQL, presumably github fails to spot lines of SQL embedded in other langauges).
Also a lot of repos contain multiple languages (web languages more often than not do)
Github is a reliable source but I am not sure whether it is representative. At least it appears to be very popular with web folk. Other sources are missing such as bitbucket, especially since it allows for free of charge private repositories. Like this, the only thing this says is that github use correlates with the lines of JavaScript. Not sure whether this is a good measure for popularity. Much like a kid in kindergarten, who shows up every day and talks to everyone. Yes, his presence correlates with the presence of others, but what if he is just a jerk and nobody really likes him?