just stumbled onto the ETH r&d project TLSNotary.org -+ such a slick use of tls session keys to do MPC 🦀

🌐📶🧑‍💻

https://mac.hot.page

Collusion problem to generate arbitrary proofs

Working on building zkTLS infra/products. A ruthless empiricist and meme connoisseur.

/opacity

and

/eulerlagrange

I saw someone post about running TLSNotary on EigenLayer AVS. This does not work in the general case.

EL requires defined rules to slash stake for trust to propagate to the AVS. If you verify a proof by committee, the user can change the truth in b/w generating a proof from the committee.

https://warpcast.com/eulerlagrange.eth/0xa1d57804

I explain here why it’s problematic

Do you have any ideas on how to reasonably reduce collusion for mpc TLS schemes ?

Oracles:

1. Don’t use private user data
2. The truth value can’t be changed

So you can use a committee that all make the request and check for agreement. In practice I think a lot of them have strong trust assumptions.

What do oracles do to try to solve this problem that you think doesn't work?

 How could it be improved?