Join Nostr
2025-09-30 04:33:02 UTC
in reply to

mccrmx on Nostr: Yes that's a potential problem for sure. I was thinking about refining the DHT ...

Yes that's a potential problem for sure.

I was thinking about refining the DHT approach by:
- Checking the NIP-11 for policies etc.
- Sending one event and checking it can be read back in reasonable time.

That would whittle the original list down to higher quality relays.

It's also not either-or. You can throw some DHT relays into the mix to get extra decentralization and also connect to well-known ones.

If you're connecting several you'd hope some of them are higher quality. The algorithm is robust to lists that vary, so it doesn't matter too much if there are some bad quality relays in the list.

When I did a rough census previously a majority of relays were running standard software. I wonder how many people are modifying the default settings?
https://github.com/chr15m/nostr-relay-research/blob/main/README.md?plain=1#L20-L27

I guess what I'm thinking is there are probably a lot of great relays, at least as good as e.g. relay.damus.io and nos.lol but people continue to default to the latter. That seems like a waste of good public resources.

What would be cool is if there was some NIP-11 flag for "This relay is happy to be part of the global grid and accept your events" (though maybe this is implied just by running an open relay). Also an uptime report. Or some other metrics clients and users could use to evaluate relays.

I will continue to test this approach some more in my own apps. Thanks for your feedback!