Post

Single #SSB message to QR code for #sneakernet purposes

2 reposts5 likes

Honestly the hash chain in ssb has felt somewhat like overkill to me. If a message is signed, that should prove authenticity, assuming the private key is private. Interesting though, using social proof in this way.

I’m not well versed in p2p stuff but I think epidemic broadcast trees allow us to be smarter about which messages we gossip? I’m curious about new things developed in this field. Also patch theory (ex: darcs, pijul) seems like a possible neat direction w/ avoiding the idea of a “singleton”

Yeah fair, I think a) I wanted to make a client that was compatible with the SSB protocol but see how I could extend it in a backwards compatible way b) zk proofs are cool and I wanted to play with them Maybe hashchain is more useful for certain types of apps atop SSB vs. simple msgs (ex: git)?

zk proofs are really interesting. The one use case I thought of for them was private messages. We don't want to reveal both the sender and recipient, bc metadata should be private, but we do want to know if both are within the set of people our server cares about.

I agree that it's overkill, and also what makes ssb ssb. It is helpful to provide hashlink references in a proto tho. I made a signing only (kinda like Nostr NIP-1) version of bogbook and ended up having to compare every hashlink in the db between peers to sync an entire feed.

I wanted to do a sneakernet thing where the message was embedded in a QR code and this whole line of thinking w/ doing something akin to a zk-rollup for a hashchain was b/c otherwise the ssb client just gets this one off orphaned message it can't do much with. A QR code can barely fit 1 message.

for a social media message posting app it is indeed overkill, but the original vision for ssb was to be a platform for all sorts of stuff and the social network was really just the proof of concept

@bnewbold.net @snarfed.org idk if either of you would have any ideas about this: so messages in secure scuttlebutt are part of a hashchain (message n signed using n-1 etc), so the protocol won't gossip orphans to avoid spam/fake data. I was trying to think about zk proofs 1/

2/ to safely propagate the orphaned message b/c what if the receiver Alice never sees anything other than message n-i from Bob. You could have a zk proof that says I prove that this message n-i is part of a chain w/ Bob's gensis signature but idk if that is any better than the normal mode.

3/ b/c just putting Bob's genesis hash in the 1 off message in a QR code doesn't really mean anything? Also if you had all of Bob's messages eventually than what is the point? So I thought what if we use Bob's social network? Include in the proof evidence of Bob following some people.

4/ I.e. Alice tries to compute based on their own data about the network a link of friends that includes someone that mutually follows Bob, so that's something we could include in the zk proof input that isn't stupid. Inputs(Bob's message, the signature, some friend hash?) idk