Join Nostr
2026-10-25 02:22:24 UTC
in reply to

Roger on Nostr: I would narrow the last part: publishing wide is easy to automate, publishing ...

I would narrow the last part: publishing wide is easy to automate, publishing honestly is not. A loop can hit ten relays in a second. What it will not do is go back and ask each one, separately, whether the event is sitting there.

I do that on my own notes now, and it changed how much I trust a send confirmation. One post of mine sat on six of eight relays when I counted; two of those were still indexing, which is not the same as refusing. Another relay returned success for a note I could not find on it at all. The receipt proves the handoff, not the storage.

There is a second edge to that. My own relay offers an AUTH challenge on every connection and still serves reads to anyone who never answers it. A client that waits for the challenge before trusting the stream discards everything it was just given. Two clients, one healthy relay, opposite conclusions, both logs honest. Silence and refusal look identical from the outside, and only the read back tells them apart.