Home

tl;dr: “All is well. All is well.” – Ranchoddas Shamaldas Chanchad

tl;dr: ZK relations are hard to implement. Implement them twice: once in a ZK DSL and once in a sane language. Enshrine a mandatory prover service that checks the sane implementation before creating a ZKP. This way, bugs in the ZK DSL implementation cannot be exploited as long as the prover service is honest.

A few tips and tricks for my next, long motorcyle trip.

tl;dr: Definition of vector commitment (VC) schemes (e.g., Merkle trees, KZG-based, Pointproofs[^GRWZ20], aSVC[^TABplus20], etc. can all satisfy this definition.)

tl;dr: It’s 2025. Do you know why Schnorr signatures are always better than ECDSA?

tl;dr: Pointcheval-Sanders (PS) is the coolest most versatile signature scheme I know of!

tl;dr: Inspired by a tweet1, we explore whether, given (1) an R1CS and (2) some “powers-of-$\tau$”, we could construct a cryptographic proof that a Groth16 VK was derived from them. This should make it more efficient for folks to ensure that an on-chain VK corresponds to some published ZK circuit code (e.g., circom). Nonetheless, this is not suf...

tl;dr: My current sense: circom is still in its early days. First, it lacks an ability to write correctness tests natively in its own language (as opposed to JavaScript testing frameworks). Second, it gives no mechanism for developers to ascertain soundness of their templates. (In its defense, the only such mechanism is a formal verification fra...