FAQ entry about zkSNARKS

[heeckhau]

context: tlsnotary/tlsn#1062 (comment)

[themighty1]

Thanks for starting this convo.
Some thought:

Let's reserve the use of "TLSNotary" for just the MPC-TLS protocol.

It's easier to conceptualize that the zk part of TLS comes after MPC-TLS if and is something that is provided by the tlsn crate, not part of the TLSNotary protocol.

Also, let's not mention the attestation here, since for someone who is simply reading an FAQ section, it is not obvious why we would make a caveat about attestation in the zk section.

This is the polished version of this section:

Can I prove statements about TLS data in zero-knowledge with TLSNotary? Does it use zkSNARKs?

The TLSNotary protocol itself is an MPC-TLS protocol; the zero-knowledge functionality lives in the higher-level tlsn crate built on top of it.

The tlsn crate currently supports proving hash commitments to transcript data in zero-knowledge, and we plan to extend it to support proving more complex statements about the transcript data in the future; we use the QuickSilver zk proving system.

QuickSilver is an interactive zero-knowledge proving system (in contrast to the non-interactive zkSNARKs), with very attractive performance characteristics in terms of proof generation time and memory usage. Its interactive nature makes it exceptionally well suited as the proving system inside an already interactive MPC-TLS protocol.

Once you have a hash commitment produced and proven with tlsn, it is straightforward for the prover and verifier to use a general-purpose zk proving system (such as Noir) to open that commitment in zk and prove arbitrary statements about the committed transcript.

---

Feel free to mix and match the and make the final decision on the final wording.