Answers are grounded in this paper and cite their sources.
This conceptual paper introduces Subthreshold Accumulation Evading Reopening, a mechanism through which a series of small changes can leave a settled conclusion in place even after the total change would warrant review. Each departure falls below the threshold when measured against the immediately preceding state. Measured from the original baseline, the same sequence can produce a change large enough to reopen the question. The paper assesses a reopening rule by asking which quantity its threshold applies to: the size of each increment, the accumulated change from a specified baseline, the rate of change, or another quantity justified by the purpose of the review. The comparison also depends on what the threshold protects. A gradual change and a sudden change of the same size can have different effects. The mechanism applies when the total would warrant the same review in either case and the way the rule evaluates it accounts for the difference. The paper places the mechanism alongside the Weber–Fechner account of perceptual thresholds and examines two documented cases. In Ratzlaf v. United States, the Supreme Court recounted purchases of cashier’s checks at several banks in amounts individually below the cash-reporting threshold. When two proposed purchases at one bank prompted a warning that a report would be required, the buyers canceled them. The leap-second rule illustrates how a cumulative comparison works. It uses the accumulated difference between Coordinated Universal Time and Earth’s rotational time to keep the two within 0.9 seconds, even though each day’s change is small. Part of the Recursive Inquiry and Encounter series within On Human Modeling, the paper has a prospective application in AI performance monitoring. Suppose a system scores 94 percent on a fixed evaluation set at approval and 93, 92, and 91 percent over its next three releases. A review threshold of two percentage points is never reached in a comparison with the immediately preceding release. Comparing the latest release with the approved version reveals a three-point decline.
No takes yet. Share an insight, caveat, or question.
Alexandra Cary (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: