Groth16 still lives: exploring the trade-offs of modern ZKProof systems


In October 2019, on the heels of SNARK-tember, Georgios Konstantopoulos gave a well-received talk exploring the tradeoff space of ZKP Systems, titled “Groth16 is not dead”.

Almost exactly 3 years after, we will try to provide an updated view of the ecosystem trade-offs, from the perspective of a budding L1 chain that aims at being ZK-proof friendly.

We will summarize overall directions in the space, offering perspectives grown from contributing to the Winterfell Stark prover, and wrangling and benchmarking Arkworks & BLST. We will insist on:

  1. the relevance of comparing trusted setups to transparent approaches,
  2. the search for prover and verifier performance, along with their curves,
  3. the fragmentation of arithmetization approaches post-PLONK, and its impact on languages,
  4. the various approaches to recursion schemes, with questions on their expressivity.

Among the others, we will cross a few past ZKProof standardization proposals along the way, and we expect the audience to hear more questions than answers. We hope listeners will come back from this talk with a refreshed view of the fast-moving pieces in the space, along with growing empathy for the systems engineers trying to keep a somewhat systematized & informed perspective. We’ll also present interesting preliminary Groth16 proof verification benchmarks using our fastcrypto Rust crate.

The associated repo can be found at Mysten Labs.