Things like this can often be assessed on a macro level. When you start to get the number of users sites like Facebook have and sites like Reddit claim they have, you end up with content that's reflective of a broad sample of society. You do have that on Facebook, you do not have that on Reddit.

I suspect Reddit is intentionally overcounting by doing things like counting multiple devices as different users, multiple accounts as different users, making minimal efforts to remove bots, counting dynamic IPs as distinct users, and so on. You could even count API callers as users, but that is stretching the limits of plausible deniability. The thing is - their content isn't reflective of the popular town bar, it's representative of an insular clubhouse with some small rooms in the backyard for 'normies.'