I find that readers ARE the disproportionate cost. Without client auth you will eventually be taken down by spammy aggregators unless you police bandwidth consumption.

Reply to this note

Please Login to reply.

Discussion

Is there a NIP already covering client auth?

I hope everybody will require AUTH soon, Snort has supported AUTH for a long time but it seems like no other clients want to implement it.

Another good way to track this outgoing cost would be to track the number of events served, if clients are bad at querying or are inefficient then it's going to cost the user more so it will also force client devs to create more efficient queries

Yes - client auth seems to be the beginning of actually being able to charge for use as far as I can tell.

My problem is the current NIP-42 forces you to open the websocket connection first which makes absolutely no sense for a private relay.

I would like to see auth in a header instead.

I definitely want to add AUTH to coracle, just haven't been able to get to it. It's my opinion that private relays are the future of nostr, so AUTH support will eventually be a sine qua non.