One kind of optimization involves reducing the number of instructions in the algorithm, and the limit of this process appears to lead to a twenty-two byte (non-parasitic) replicator. This replicator is almost a quarter the size of its original eighty byte ancestor, and replicates almost six times as fast.