Testing of a proposed Ethereum privacy pool put the creation of a single withdrawal proof at 28.6 milliseconds; a separate batch of 128 withdrawals took 87.6 seconds to assemble.
The single-withdrawal test used a depth-32 Merkle tree, which organizes deposits so a withdrawal can establish that its deposit is part of the pool. An earlier design needed 0.81 seconds to create one proof in the comparison. The faster result measures proof creation for one withdrawal, rather than the separate task of assembling proofs from many withdrawals.
In the batch test, assembly took 1.7 seconds for four withdrawals and 87.6 seconds for 128. The completed proof occupied 343.9 KiB at both reported batch sizes, while verification took 46 to 64 milliseconds across the tests. The unchanged proof size did not mean the batches took the same time to assemble.
The design uses Flock, a proof system intended for post-quantum security. It runs 34 SHA-256 compression steps through repeated uses of a small circuit instead of building one circuit containing the full sequence. SHA-256 is also used for the deposit tree and for nullifiers, which prevent the same deposit from being withdrawn twice. These steps explain how the design creates an individual proof; batch assembly remains a separate operation.
The proposal also outlines a recovery route if quantum attacks make the cryptography of older privacy pools vulnerable. Under that approach, wallets would derive deposit secrets from a seed. An owner could later use those secrets to establish a claim through a post-quantum recovery verifier, provided the pool supports it.
The pool would need to retire its old verifier before that cryptography is compromised. Deposits created using the older random-secret method could become inaccessible once the old verifier is retired. A pool that cannot be modified has no way to add the recovery verifier; an upgradeable one would require a governance change.
