Blockchains:
Verification cost: O(N)
Communication cost: O(N^2)

ZK:
Verification cost: O(log N)
Communication cost: O(log N)

engineer / founder 🇦🇺 I love zk proofs, vm's, mimetics, navigating latent space. Live by gradient ascent

Things have a strong path to getting better:

- hardware accelerated provers are being built as we speak
- new proving systems come out every day
- tooling is always being improved
- recursion is being implemented by every team atm

Recursion btw is what enables logarithmic communication cost. This is from the Goliath whitepaper- https://i.imgur.com/vRrdoUk.png

I'm at the cusp of the ZK readings, because I'm an obsessive manic.

Tech drawbacks atm:
- proofs too large (20kB)
- proofs too long to generate (5s-20mins)
- no generally implemented recursion
- dev tooling all still early

We are on the cusp of a revolution.

BDFL of Quilibrium, Eng @ Farcaster, ex-Coinbase, always opinionated, never hydrated

/quilibrium

This is excellent and you're 100% right - I was simplifying the finer details to get the message across.

This isn't quite right:

zkSNARKs:
ver: O(1)*
com: O(1)*
tx proof: ~200B

zkSTARKs:
ver: O((log n)^k)
com: O((log n)^k)
tx proof: ~45kB

bulletproofs:
ver: O(N)
com: O(log n)
tx size: ~1500kB

It comes down to balancing complexity growth, SNARKs being the worst