Optimizing for the average case

you’ll be hearing more from me

Brink will guide you home And ignite your bones And I will try to fix [your margins] Get early access to Brink and save up to 90% on transaction cost. 👉 .

Stable Stories EP 11 is live! 🎙️ This week, @riabhutoria & @spjoleh of @deltadotnetwork join @AnnikaSays to dive into: 💠Rethinking L1s 💠Why “centralization is good” (sometimes) 💠Stablecoin sandwichs & cross-border infra 💠Web3 sovereignty + pragmatism Check it out! 👇

So it's clear that snail needs a pcs and then the entire construction is a bit overkill But why is no one doing this? Seems like a really simple and efficient analogue to the univariate quotient zero check

Update: Likely not a room-temperature superconductor - The concrete issue that has been discovered is that Protocol 2 does not bind the prover to the circuit. This might be fixable with a cheap subprotocol, but... - I'm starting to come around to @ronrothblum's consistent feedback that Protocol 2 would collapse the polynomial hierarchy. I've previously thought that the specific algebraic structure of the layered circuit basically guards us against the information-theoretic bounds, but after getting a bit more familiarized with the literature I think I'm in the wrong here - Because of this, I'm tempted to claim that there must be some inherent issue with the "aggregate all layers" approach. - The core construction of SNAIL (Protocol 1) seems to remain intact for now, and should be of interest independently

Introducing SNAIL Verifiable computation at 1.2-1.3x the speed of native execution*, more than 10,000x faster than state-of-the-art SNARKs * for e.g. typical zkVM computations