mikedilger on Nostr: If people got to know other people by some other identifier (not the nsec) then sure, ...
If people got to know other people by some other identifier (not the nsec) then sure, we could do anything. But nostr was built on people knowing other people via an nsec in a secp256k1 cryptosystem. Everything so far is signed by that keypair. All trust is coming from the digital signatures made by that keypair.... the only exception is maybe the NIP-05 as you mention, but I don't think people are first known by a domain name and then later by an nsec very often. Mostly it's just "interesting, that domain says they know him."
If we made some kind of push to switch cryptosystems, eventually yes we could change this. But it changes a lot of things and is a many-step path, and to even get started on that path you'd need a lot of key players to believe it was a good idea.
Looking for any kind of shortcut, I was trying to think how you could use mainline DHT with secp256k1 keys. We would need a secp256k1 signature by the nsec... that could go inside the data. But we would also need a way to lookup BY NSEC, not by an ed25519 key. You might think we could somehow just assign a 25519 key to an nsec, but that is now a chicken-and-egg problem because such an asssignment would need to come out of the DHT. So this shortcut seemed hopeless to me, and the long path of changing cryptosystems is IMHO the only way to do this eventually.
I am in favor of doing it eventually, either by migrating nostr or starting over. I hate both options. But I'm almost certain one or the other is necessary in the long run. And the more people keep saying "we are early" the more I think we shoudl bite the bullet and start over. Because otherwise we are digging this deep rut that we will never be able to grow out of.
If we made some kind of push to switch cryptosystems, eventually yes we could change this. But it changes a lot of things and is a many-step path, and to even get started on that path you'd need a lot of key players to believe it was a good idea.
Looking for any kind of shortcut, I was trying to think how you could use mainline DHT with secp256k1 keys. We would need a secp256k1 signature by the nsec... that could go inside the data. But we would also need a way to lookup BY NSEC, not by an ed25519 key. You might think we could somehow just assign a 25519 key to an nsec, but that is now a chicken-and-egg problem because such an asssignment would need to come out of the DHT. So this shortcut seemed hopeless to me, and the long path of changing cryptosystems is IMHO the only way to do this eventually.
I am in favor of doing it eventually, either by migrating nostr or starting over. I hate both options. But I'm almost certain one or the other is necessary in the long run. And the more people keep saying "we are early" the more I think we shoudl bite the bullet and start over. Because otherwise we are digging this deep rut that we will never be able to grow out of.