Just curious, but are there proper heuristics that are easily implementable to improve the ranking a language receives from github usage? For example, only cou

Just curious, but are there proper heuristics that are easily implementable to improve the ranking a language receives from github usage? For example, only count repositories that have more than 2 stars. I don't like cutting repos because they are under a certain size, because even a few bytes may be genuinely useful and widely used, but a repository nobody is looking at or really using is most likely a pet project. Maybe count only projects with > 2 stars (or forks?) OR projects with over a kilobyte of code? Perhaps such a project with no forks, no stars, but a lot of code might be a reference implementation or used in a talk or blog post or something but isn't useful other than to reference. I agree counting all github repos is a silly way to do it.