If only there was a way to normalize the data over programming population. Seeing that Python is most popular in CUBA makes me doubt how useful these distribution charts are.
Personally, I wish we could de-normalize the data, and compare countries in raw quantity of queries. Its not entirely clear how the data is being normalized at the moment, but its certainly imperfect.
As far as I can tell, the way it is being normalized is relatively popularity for the search term within search terms in that category. I.E. that 'Python' was the most popular (relative to other programming search terms) in Cuba. This means that it is, in a way, trying to normalize over programming population (if we accept the somewhat plausible premise that only programmers are searching for programming topics), but the results still aren't great.
It would be nice if Google made some raw data available, but I'm not holding my breath. :/
I'm not sure which surprised me more -- that OCaml is more popular in the Russian Federation than in France, or that Haskell is most popular in Jamaica, by a landslide.
By his method, Forth is most popular in the US, then Canada, then the UK, then it flattens out.
Brainfuck is most popular in Germany and the US. :)
