Model-layer alignment is a game of whack-a-mole: each fix requires a new training run or post-training intervention that is expensive, slow, and unpredictable in its transfer across the problem space. Coordination-layer alignment operates on software-engineering timescales — adding a checker, tightening a permission, raising an audit rate, or deleting an edge can be done quickly and iteratively in production. This feedback-loop asymmetry means coordination-layer alignment compounds while model-layer alignment bets.