Programs often break when interesting evidence is mistaken for permission: when CMC is treated as a technical appendix, when translational signal is over-read, when clinical expansion outruns closure, when vendor choices are made without consequence mapping, or when capital and claims advance faster than the underlying proof state.