tl;dr version?

My understanding: What software authors need from a language varies so greatly across each of those tiny squares that sampling from it with quantitative research doesn’t tell you very much at all. But I’ve only read the tweet!