Load-bearing claim

A claim that actually supports the budget request, as opposed to a decorative one that adds nothing. The discipline is to make only the smallest set of load-bearing claims, because every claim made is a claim that must be defended, and a non-load-bearing claim is pure attackable surface with no upside. Test whether a claim is load-bearing by asking whether removing it would actually change the ask, the risk picture, or the alternative comparison; if the budget survives its removal untouched, it was decorative. Watch, too, for claims that quietly depend on each other, where losing one weakens a second you thought was independent, and for the fact that which claims are load-bearing can shift with the audience: a claim that carries no weight with a CFO may be exactly what a values-minded director needs to hear.

A statement in the chapter that a reviewer is specifically trained to test first, because the compliance conclusion depends on it (the lawful basis, the special-category screen, the provenance of data feeding the system's highest-risk output). Load-bearing claims need the strongest evidentiary support in the chapter; background context does not need the same density of citation.

A specific, checkable assertion (a statistic, figure, citation, name, or fact) that a deliverable's main conclusion depends on, such that if it were false the conclusion would collapse; the priority target for verification.

The specific assumption in a design that, if wrong, breaks the robot's core case; experienced reviewers deliberately search for this claim and the thinnest evidence supporting it before asking their hardest question.

A statement a decision actually rests on; the claims that receive full verification because the decision is wrong if they are wrong.