those conclusions are drawn without controling for a language usage. Since c++ is widely adopted so there will be more instances of a comment where "WTF?" is used.
Why don't use a percentage at least? Even if that was the case, the problem remains... a wtf-y language may be the most avoided and/or not present in github
Well, they DID account for it, but they did it all wrong. They counted WTFs *per repository*