Testing What Holds: A Practical Audit of Contact, Burden, and Revision Abstract

This paper presents a practical audit for testing whether an argument, relationship pattern, institutional process, AI workflow, policy, or personal rule actually performs the function it claims to perform. The method begins from a simple distinction: visible coherence is not sufficient evidence of adequacy. A structure should be examined through its stated function, repeated effects, contact with relevant evidence, burden distribution, correction pathways, scale, and behavior under challenge. The audit contains ten steps: name the local structure; state its claimed function; compare claim with repeated effect; identify relevant contact; trace carried and exported burden; locate real correction authority; test for capture by a proxy or self-protective objective; check scale adequacy; observe response under pressure; and issue a provisional verdict with explicit falsifiers and a revision path. The method is positioned near organizational learning, sensemaking, systems thinking, institutional analysis, and human-centered design. It rejects two temptations: a single universal score and dramatic inference from thin evidence. Instead, it uses an Answerable Holding Profile that records strengths and failures across dimensions without pretending they share common units. Four worked cases - AI deployment, relationship conflict, institutional metrics, and a personal rule - show how the audit changes when domain-specific evidence changes. The contribution is methodological: a portable way to keep interpretation close to observable consequence and revisional power.