The downside to using a certain tech stack is technical debt, not a health risk. Technical debt seems like a wise acquisition in moderation, especially for a startup that wants pictures and buttons on screens ASAP, and even more especially when everyone is realizing that it can rather easily be undone when/where it matters with type checker addons and FFI.

That doesn't have anything to do with the question of whether "raw number of successful projects using it" is a good metric for how good something is. Lots of successful companies have made mistakes, and there are even mistakes that many companies in a field made simultaneously (e.g. the XML mania of the early 2000s is not fondly looked back on).

I'm not commenting on the specific issue of whether Python creates good trade-offs, just that I think tallying successful companies is a poor measure of a programming language.

Ah, I gotcha, that’s a good point. I still do believe that these decisions for dynamic languages that can later be contorted aren’t just a symptom, but I don’t have any way to prove it.