Lab: free tools · size first · stock scanner · crypto scanner · journal

Governance

Recursive Proofs (ZK)

A recursive proof is a zero-knowledge proof whose statement includes checking another proof.

Stack that up and many computations can be compressed into one small proof, which is how systems like Mina keep the chain tiny and how ZK rollups combine many batches. Recursion makes proving more complex, but it is one of the main techniques behind scaling with zero-knowledge proofs.

Why it took a while

The 2019 Halo paper by Bowe, Grigg and Hopwood describes recursion as "proofs that attest to the correctness of other instances of themselves." Before Halo, practical recursion needed a trusted setup and special pairs of elliptic curves built to work with each other. The Electric Coin Company blog calls a trusted setup a systemic risk that has to be repeated for major upgrades. Halo was the first practical recursive proof composition without one. It ran on a cycle of ordinary 255-bit curves, and proof size and verification time don't grow with the depth of recursion. ECC looked at it for Zcash.

Mina, the plain example

Mina is the easiest place to see it. Nodes don't store the whole history. They keep one proof of constant size, built with recursive zk-SNARKs, that stands in for the chain. Mina's blog quoted 22 kB for that proof; a later technical reference measured it at about 11 kB. Mina's Berkeley upgrade, completed in June 2024, enabled recursive proof composition for zkApps. Because the proof stays that small, a phone or a browser can check the Mina chain without downloading years of blocks.

The catch

Somebody still has to run the hardware that produces every proof. As a trader, I take a project saying it "uses recursion" as a statement about verification costs, not about whether anybody uses the chain. I don't know which proving systems will win out, and I'd be wary of anyone who claims they do. For the rollup side, start with how ZK rollups work and proof aggregation.