Urbit: The good, the bad, and the insane

Definitely the most concerning part of Urbit, for me at least, is the obscurantism of Hoon. Much like the author says, it's totally incomprehensible to anyone who hasn't read through every bit of the (lacking imo) documentation. Even a domain-specific language with a familiar syntax would be better, since at least a first-timer could grok basic logic/structure at first glance. This only compounds the security concerns, since reviewing any Hoon code requires being a domain expert - I doubt any security/crypto researchers will want to take a few weeks to learn Hoon just to audit something as obscure as Urbit. Hoon even has a kind of Lisp-y feel to it and Lisp lends itself well to the purely functional goals of Hoon/Nock, so I don't see why they couldn't have just made Hoon a Lisp.

On the whole however, I'm hesitantly optimistic about the project. Most other works aiming to address Internet centralization just seem like band-aids (federated social networks still have problems with moderation/deplatforming, self-hosting just means running it in someone else's data center, etc.), and Urbit seems to be one of the few things willing to deal with the fundamental problems from the ground up.

Also, I'm kind of glad the author didn't mention Urbit's original author despite most other posts on the topic feeling the need to mention it. He's long since left the project but it seems as though it'll never be rid of his influence. At this point I just feel bad for the maintainers/users that will constantly be associated with the guy from now on.

On the other hand... is walltime the only criterion worth considering?

The major design concersns raised by the above comment and the OP are inseparable from the original author, because he chose an extremely idiosyncratic language and system design that eschews all known standards and practices.

If people want to make the idea work, they should port it to a reputable language.

It is indeed quite similar to mastodon in the sense that other instances can ban you. The difference is that in mastodon your identity is usually located on an instance, and if that instance bans you, you lose your identity. In Urbit every identity has its own instance so your personal contact list will still be able to reach you.

Of course requiring everybody to run its own instance adds quite some friction for user onboarding, but that is an issue with the urbit network, not with the identity layer. You could imagine having your identity and linking it to a hosted instance, and later switch instance if the needs arise.

This is what I like about the idea of having your identity as a NFT, but clearly the way urbit currently uses it does not exploit the full potential of the idea.

As for losing your keys, if such a system would become widespread, most people would chose to have their keys hosted on a third party service, and only people who really care about maintaining full ownership would keep the keys to themselves.

Re-reading it now, I'm still a little confused.