The stats are interesting, but still dangerous to draw direct conclusions from. (What the graphs suggest may be accurate, you just can't assume it from the art

The stats are interesting, but still dangerous to draw direct conclusions from. (What the graphs suggest may be accurate, you just can't assume it from the article alone.) For example, suppose a language's standard library is already powerful, or its 3rd party libraries are very reliable. If one language has one Really Good library to do X, and another language has 7 mediocre libraries because no one does X well, which language is more useful? Do you count the 7 projects required in the 2nd language as evidence of higher popularity, when they're all just picking up the slack? Yet another example is projects that use multiple languages. For instance, I can combine scripting languages and lower-level languages easily. This may technically be evidence of the "popularity" of the lower-level language, but my hybrid could be using the scripting language much more.