Regulatory risk assessments contain offsetting non-conservative assumptions that can reverse their net conservative bias
5 events · 2 assessments · 2 decisions
Reassessed
Triggered by subclaim_change: the supporting subclaim on unobservability of true regulatory risk was assessed SUPPORTED (0.82/0.85). The prior assessment already treated this premise as load-bearing for the capability ('can reverse net bias') half, so its confirmation reinforces rather than alters the verdict. Status remains SUPPORTED; confidence nudged 0.68 -> 0.70 to reflect the now-established supporting premise. The residual uncertainty is reading-dependence (modal capability vs. systematic reversal), which the subclaim does not touch. No structural changes: the decomposition (existence half uncontested; capability half supported by the unobservability premise) remains apt, and no missing dependency was found. No credence recorded (reading-dependent, not a single measurable quantity). No dependent notification: status unchanged and confidence change immaterial to downstream claims. Importance left at 0.4 (major-ish, notable/major band): a live methodology debate but with modest, bounded consequence.
Reassessed: still Supported
verdict confidence 0.68 → 0.70
Structured and assessed
First pass. Situated the claim in the well-developed conservatism-in-risk-assessment cluster; it is the Finkel-style counterweight and is already a 'contradicts' child of c6f0b7ce (contested) and 1bd243b2 (supported). Decomposition kept deliberately lean and non-circular: the Matcher collapsed a proposed 'non-conservative assumptions exist' subclaim into the negation of 1bd243b2 (declined to mint a duplicate; kept in prose as largely uncontested bedrock), and flagged a 'net bias indeterminate' framing as near-identical to this claim itself (declined as circular). Attached the existing, genuinely load-bearing premise that true risk is generally unknown (e3fb06dd) as a supports edge. Single natural line of support, so no named argument created. Importance set to 0.4 (notable/major, high contestation 0.7): consequential within environmental/health regulation and actively argued, but a niche methodological point relative to all of claimspace. Assessed SUPPORTED (confidence 0.68): existence of non-conservative assumptions is essentially uncontested (Finkel 1989; NRC 1994 Science and Judgment) and the modal 'can reverse' is defensible, while the stronger reading (net bias systematically reversed/indeterminate) is contested by revisionists. Omitted a single credence because truth turns on the reading rather than one measurable quantity. Kept canonical form (~15 words, neutral, both sides would accept).
Assessed Supported
verdict confidence 0.68
Regulatory risk assessments combine assumptions that push estimates upward (health-protective defaults, upper-bound potencies, maximum-exposure scenarios) with assumptions and omissions that push them downward: exposure pathways left unmodeled, susceptible subpopulations incompletely captured, untested chemicals and unstudied endpoints, and synergistic effects among co-occurring exposures. The existence of such non-conservative elements is widely acknowledged, including by the agencies and expert bodies that defend conservative defaults. This claim is the counterweight to the view that stacking conservative assumptions makes estimates systematically and substantially exceed real risk: it holds that because offsetting non-conservative assumptions are also present, the net direction of bias is not fixed by the presence of conservatism and can, in a given assessment, run the other way. On its modest, modal reading, that offsetting non-conservative assumptions exist and can be large enough to reverse the net bias in particular assessments, the claim is well supported. It is most associated with Adam Finkel's "revising the revisionists" argument, and the National Research Council's 1994 Science and Judgment in Risk Assessment records the same structure while noting that partisans do not agree on the descriptive question of whether assessments are, on net, too conservative. The claim is strengthened by the point that the true underlying risk is generally unknown, which prevents the net direction from being settled as conservative by direct measurement. Where credible disagreement remains is on the stronger reading, that non-conservative assumptions typically dominate so that net bias is generally indeterminate or reversed. "Revisionist" analysts hold that conservative elements usually outweigh the non-conservative ones, leaving assessments conservative on balance; the offsetting-assumptions camp holds the net is unestablished. That magnitude question would be resolved only by retrospective calibration of regulatory estimates against realized outcomes, which the field largely lacks.
Claim entered the graph