Oddbean new post about login | logout It's a two-step process. You can do one without the other. The more important part is the resolving, the less important part is the short name. Think of it this way. It's like building a file system. Then adding sym links. The file system is the more complex part, the sym link is the easier part, after we have working file system (there is a solution here but I'm stickin to a single topic here). It's not just DHT. DHT is for the censorship reistance. It's a home server, which is fully powerful and scalable remote storage with micro apps. The DNS part just does the resolving. In fact names are easy to fix, the question is what if two keys want the same name, what tie breaker do you use. That's a good problem to solve, and it's the next part of the two-part process. #pubky isnt confusing the first step with the second, which I think is a good approach.
yes, I know. but that's just what DNS and the naming authorities solve, isn't it? solving names in a decentralized way is hard but important for usability. Namecoin(👴) and ENS seem to be still the best solutions for that part?
The naming part is actually easy once everything else is there. You just start off by using your npub as name. The resolver is the powerful thing. Then there are so many paths. Nomen was going to be the great solution based in bitcoin, but fiatjaf and semisol wanted to kill it, and so it died. What you need simply is a fair naming system that is a level playing field and with a wide eco system to ensure a small clique doesnt grab the good domains. namecoin got rekt and it didnt allow transfers, which was a glaring mistake. ENS is very scammy and they raised the prices massively. the fact that there is a "they" is also a problem. The only solution to this is naming on bitcoin which is pretty a pretty much solved at this point, you would use an OP_RETURN UTXO runes or a soft-fork of runes with an OP_RETURN flag. But let's see how it evolves. Have a look at resource oriented computing on wiki for a more in-depth explanation. But, the npub resolver alone is huge because it gives you decetralized domains and also SSH keys for free, which is a massive upgrade. https://stacker.news/items/256435
forgot to say, also you could tie it into nip-05 too if you wanted
which would be again relying on DNS :)
Yes, bridge to the older system. Which isnt going away.
one last point is that it works out-of-the-box with nostr (and could/should with alby too) click on settings -> paste seed, and it will generate you a pubky address from your nostr privkey its actually an ed25519 key z32 encoded, which allows you to use, ssh, git, matrix, signal, gpg, hypercor, Tor, wireguard and much more, out of the box -- something we probably should have done on day 1
I disagree with this, I think phone numbers, and even browser address bars autocomplete prove that subjective local naming of keys are enough for the vast majority of use cases. I don't need to type your pubky from memory every single time, in fact I don't think I remember the exact handle of every twitter user I follow or the domain of every blog I like to visit. Even when contacts and bookmarks are not enough, duckduckgo takes me the rest of the way. human readable names are not worthless but they are not nearly as valuable as people make them sound, and when they are absolutely necessary, ICANN is fine. Onion addresses prove my point too.
Onion addresses also prove that with the increased number of randomised characters in the domains the end user experience is extremely low and subpar when compared to short names. Long randomised domains create a perfect environment for various fishing scams.