> What I find striking here is that there are both fewer questions and fewer answers.

Assuming the ratio of answers per question remains the same (or drops because the answer quality raises) this is the sort of thing I would suspect.

It wouldn't even be theoretically surprising if the number of Qs & As on MathOverflow trend down purely because all the easy questions get answered. Wouldn't necessarily expect that, but because we can't really anticipate whether questions should trend up/down/sideways I don't think analysing MathOverflow is really all that interesting. Like the writer observes, kids these days are probably all over on Discord.