and they will steadily outnumber the public ones because it is so simple to run relays
ok, it took me some time to build a good one but i now use it and automatically now anyone i follow can post on it, and fetch from it, satisfying the in/outbox requirements
i have been thinking about how to open it up a bit to allow strangers to post on it, i will figure out a reasonable minimal "knock knock" scenario that means that it's open enough to allow new people in
👀
What about attestations from other, trusted relays? Like AUTH, but once removed?
i haven't got any specific thoughts about how to most simply open up access without strong trust guarantees but i bet you'll come up with some other ideas
for now, the best i've got is second level - the follows follows - follows mutes... but to actually make all this work there needs to be an active process scanning the network for relay lists and asking for mute and follow lists
it's gonna take a bit of thinking i think, for now i have the basis... i can open up relays to people just by adding their key to my owners list and whoever they follow is automatically allowed, and who they mute is not (obviously mute overrides)
gives me some ideas about multi-level and scores ranking distance from the owners and using that to scale the timestamps so that trusted stuff is deleted after less trusted stuff, this is one idea starting point anyway
I know nostr:npub10npj3gydmv40m70ehemmal6vsdyfl7tewgvz043g54p0x23y0s8qzztl5h was trying to get a Minimum Viable Npub Standard, going, where relays can use associate-trust to keep out spam.
ah, yep, ive been working with the brainstorm.ninja team, learning about nostr:nprofile1qqsw2feday2t6vqh2hzrnwywd9v6g0yayejgx8cf83g7n3ue594pqtcpz4mhxue69uhk2er9dchxummnw3ezumrpdejqzythwden5te0dehhxarjxgcjucm0d5q32amnwvaz7tmjv4kxz7fwd4hhxarj9ec82csvvl4vs's equations for scores. both to use them in relays but also in general as a nostr power user..
wot is just a buzzword, and most current implementations are just really basic combining follows lists.. brainstorm is more calculation heavy and can produce more useful results such as the upcoming grapevine score. (which ive experimentally calculated here)
