20:31:38 <kayabanerve[m]> Just to jot the thought down, a curve cycle and IVC scheme would halves currently benchmarked FCMP times per input *and* make additional inputs only logarithmically expensive.
20:33:17 <kayabanerve[m]> IVC folding also wouldn't leak input paths to other inputs, like aggregate input proofs would without a MPC protocol.