> Keybase was sort of trying to do this with social proof, though it's reliance on a centralized provider (keybase itself) made it brittle.

I think something like this could've worked, which is why I was so sad over its demise. I think Keybase (like the MIT PGP keyring, which it is sort of a fancier version of) was predicated on the idea that it's much easier to build a centralized keyring and then decentralize it once it's widely adopted than to build a decentralized keyring from the ground up.