There are better and there are really really bad ways to do ID checks. In a world that is increasingly overwhelmed by bots I don't see how we can avoid proof-of-humanity and proof-of-adulthood checks in a lot of contexts.
So we should probably get ahead of this debate and push for good ways to do part-of-identity-checks. Because I don't see any good way to avoid them.
We could potentially do ID checks that only show exactly what the receiver needs to know and nothing else.
> We could potentially do ID checks that only show exactly what the receiver needs to know and nothing else.
A stronger statement: we know how to build zero-knowledge proofs over government-issued identification, cf. https://zkpassport.id/
The services that use these proofs then need to implement that only one device can be logged in with a given identity at a time, plus some basic rate limiting on logins, and the problem is solved.