Proof of sequential work

If the majority determines distribution of rewards, then there exists some incentive to create Sybil identities. Therefore the assumption that only 1/4 of nodes are bad may not be realistic.