Oddbean

I’d prefer a DHT for mapping npubs & nip-05 identifiers to relays where you can find their content. But purplepag.es is a good stopgap. That said is the crawler working @PABLOF7z? We’re building stuff and using purple pages to find a users up to date relay list and it’s not working well. Some data is there but others we’d expect is missing.

You can't bootstrap a DHT for the same reason you can't bootstrap Bitcoin, just accept the inconvenience and use Mainline and ed25519 keys, Pkarr is ready for prime time in JS and Rust and bringing TLDs to individuals pkarr.org

I seriously would have done whatever it took if it was at all possible to support secp keys, but it is not possible, just... accept it like you accept nip05 and move on, it is the most minimal layer that prilovides robust battle tested credible exit, more and more projects are considering using it from Fedimint, to Iroh (a Hyperswarm alternative that works in all environments including Rust) to even DIDs (regardless of your opinion of DIDs). Let's get this over with. I am not even selling you something new, Pkarr is literally DNS packets over Mainline DHT, not a single thing more.

I’m not saying run our own distinct DHT. Just we should push signed kind 10002 (nip-65) events to some DHT that’ll accept them and let us look up by npub as the key. @Nuh 🔻 are you interested in building it and writing a nip? We’d like to support the effort and adopt it for @nos.social if you did.

What I am trying to say is, this is not technically feasible, you can pretend that it is, by using some fake DHT like IPFS, but might as well use Purple pages at this point. If you want to use a real one, the only alternative is Mainline, and what you are suggesting is not doable, Happy to jump on a call to explain it all if you care because text makes me sound like a fanatic, but it is just the technical reality.

Well purplepages is a stop gap and creates a single point of failure. Not using anything means folks need to concentrate on a few relays which although not a single point of failure it is a major point of centralization.

What can I say, I built the thing that I told you I would build, and I am now telling you it is the only option, but it requires treating ed25519 keys as the root of identity, like some clients may choose to treat nip05 as such, except it is sovereign unlike nip05. Now either use it or not, I am trying to make peace with the probability that no one in Nostr will use it because the different key type is too much of an ask. And I know for sure Fiatjaf will fight against it becoming too successful for the same reason he fought against nip26 (adds too much complexity or whatever, even though it doesn't)

Nostr addresses aren't the final solution for the problem of human readable names since ICANN ultimately controls nostrplebs.com or whatever regular domain name. BitDNS (ie. Namecoin) names are a perfect fit for Nostr. They're lightweight, ridiculously censorship-resistant, easily managed and can be maintained with Bitcoin Lightning payments. We have integrated BitDNS for our Nostr client, Nymble, and the app will be available in the next couple of days. If you follow, I'll send you an update to try once it's released.