It would be extremely desirable for Nostr relays to collect OTS proofs and splice them into the events they return. These proofs should be treated as having similar cryptographic importance as signatures. Do any of the larger relays, say wss://relay.damus.io ( @jb55 ), provide any OTS infrastructure? If not, can they please do so? Here is a quick reference for what an OTS-enabled relay should do:
inkan's avatar inkan
I made an OTS-enabled version of the strfry relay. The relay collects timestamp proofs from clients, verifies these proofs against BTC headers, picks the best verified proof for each reference event, and splices it into the event's JSON when a client retrieves the reference event over REQ. So as a result, when a client requests an event, the BTC timestamp (if available) arrives with it, in the same response. https://gitlab.com/inkan_dev/ots-enabled-strfry #nostrdev #OTS
View quoted note →

Replies (3)

I'm still looking for relays that are open to adding all-important OTS support. No takers so far. Maybe wss://nostr.land ( @semisol ) could be interested? See below for what I think is needed.
inkan's avatar inkan
It would be extremely desirable for Nostr relays to collect OTS proofs and splice them into the events they return. These proofs should be treated as having similar cryptographic importance as signatures. Do any of the larger relays, say wss://relay.damus.io ( @jb55 ), provide any OTS infrastructure? If not, can they please do so? Here is a quick reference for what an OTS-enabled relay should do: View quoted note →
View quoted note →
Nostr.land is planning to support OTS. I’m planning to use the standard OTS spec that is already well-implemented. “Inline” proofs is not something I plan to support as a client can easily request OTS proofs when necessary. Also, I hate AI written “specs” so much as they have so much useless junk (like wtf is the s tag for?) and overcomplicate things.
The s-tag is for the signature of the reference event. Just to be sure, when you refer to the "standard OTS spec," I hope you don't mean the old broken NIP-03 where the item that gets timestamped only includes the event_id and does not include the signature (see below for some discussion of this). Also, I think splicing the OTS proof into the event that gets returned by the relay is essential so that clients get proof simultaneously with the event. I think that's precisely the role that relays need to play because it's a huge pain for a client to have to fetch an OTS event separately when trying to filter out events that don't have a valid OTS proof. I guess I find AI-written specs useful. They are both human-readable and can be fed to an AI to get oriented in the project without skipping too much detail. Also, the proposed modifications to the relay were made by an AI and the AI has a better grasp of the details.