What does it take for a legal proposition to be safe to rely on?

Verification, as the profession now practices it, asks whether a case is good law. Attorneys need the answer to a narrower and harder question: whether this proposition, drawn from this case, is usable here. A holding can be real, accurately quoted and clean in every citator, and still be dead as a proposition, displaced in this forum, announced under a different procedural standard, adverse to the client, or already offered to a court in this jurisdiction and rejected. None of those failures announces itself. Opposing counsel may miss them. The court may miss them. The client never sees them.

This program builds the vocabulary and the protocols for closing that gap. The Proposition-Usability Model, set out in Good Law for What?, replaces one question with four: is the proposition still valid, does it govern in this forum and posture, whom does it help, and has the argument already been rejected. The Case Treatment Taxonomy classifies treatment at the level of the holding, by scope, severity and mechanism, because one opinion can overrule one holding, distinguish another and follow a third. The Posture Mismatch Taxonomy names the failure no citator flags: right law, wrong stage. The Completeness Protocol turns to what a brief leaves out, because motions are lost on the complement as well as on the contents.

Why this matters: citation checking has become the profession's default answer to AI risk, and it is a legitimate control. But it inspects citation-sentence pairs. It cannot see the wrong standard applied correctly, the controlling adverse case that was never pulled, or the argument this court has already refused. A verification layer is a smoke alarm, not a building code. The work here is to specify what the building code should say, in terms a librarian, a judge or a malpractice carrier can apply without trusting any vendor, the Institute's own sponsor included.