no I don't. I envision each community will stand up their own relays, and you can participate in that community by following those relays.

I think this can happen organically if we have a way for users to indicate what their "main" relay is and let the network evolve naturally..?

Reply to this note

Please Login to reply.

Discussion

Mirroring n-to-n relays not sustainable because the traffic grows exponentially. I think the right approach would be new relays seeding off of existing relays, and optionally maintaining their own mirrors. Again this is not to shift the responsibility of the clients being the ones connecting to multiple relays.

personally I don't think relays should be mirroring if mass adoption occurs. they only way nostr survives censorship is if users understand vaguely the mechanics of nostr and take responsibility of their own mirroring. otherwise it will be hard to navigate when relays go down.

I’ve been working on compatibility with hypercore crypto. Your nostr identity can be a hypercore and it can be queried p2p - as an alternative to only relays.

A few hurdles so far. I’ll try work on this again this weekend.

that's how I see it too

how many connections are acceptable for a client? what's the most relays some client connected to? right now it doesn't really make sense because you're getting duplicated content, but if the content is spread out, each relay will matter more individually

would 200 connected relays be a thing?

no

good question, I have no idea of how many a browser can handle and how many resources each consumes, I guess somewhere between 15 and 40 lies the maximum number.

Unsure per tab limit, but seems like it’s around 200 websockets max. https://chromium.googlesource.com/chromium/src/net/+/259a070267d5966ba5ce4bbeb0a9c17b854f8000

from a technical perspective, no idea.

from a human perspective, should a person really follow more than 20-30 relays?

yes each relay will matter more. users need to be responsible for mirroring their own content, as well as content posted by others that is important.

It is possible for a relay to pre-cache or act on behalf of a client public key to make it easier just to connect to them. Sure it’s trust, maybe a paid service, but you can always add other relays too - but that one relay would help reduce how many you need.

this is not about relays people have for subscribing, but relays found in notes, and clients connecting to that.. if that becomes a thing, then the number of relays we connect to will grow exponentional

You could always have a pool of say three ephemeral websockets that connect to relay hints when you’re missing data and then disconnect. Or you could be recommended new relays to connect to permanently based on volume of events seen first or unique events.

Maybe a few bootstrap relays are all you need and then it can be mostly automated.