Hi @rlombreglia
- I heard Spartan[1] that Microsoft published and claimed it doesn’t need a trusted setup and RedShift[2] is also zkSNARK without a trusted setup, so I said most of them require it. For the latter question, I can’t provide an answer yet.
- As I heard from a participant of a trusted setup ceremony in a ZKP study group in TEM, they usually have to run a script for hours to guarantee the randomness and then pass the result to the next participant. That has to be done by a lot of participants to convince users (e.g. if a global system was set up by a group of only 10 contributors, it’s more likely to be considered insecure). But I’d say that is a practice consideration instead of an academic research conclusion.
- I think these can first be split into 2 groups: non-universal (traditional) and universal. For platforms in the same group, the major differences will be proof size, verification time, proving time, and setup procedure, and other differences will be post-quantum, assumption (security), and so on. As I know so far, Groth16 might be the most important platform in former groups as it has better results than others on the major differences. The latter groups have Sonic, PlonK, RedShift, Halo…etc. I’ve not read into them so I can’t describe the exact differences here, but the major differences mentioned above may be part of those.
- My answer to this question is totally in my opinion. I’d say that the previous work (Groth, Jens, et al., Citation 3 in summary) brought the interest to researchers but had practice issues and Sonic solved them and brought the possibility of adoption. The reason Sonic wasn’t used is the evolution is too rapid.