When Judgment Hardens 1.2.8-5

The Moral-Hazard Dimension

The staged-judgment lifecycle — hypothesis, scoped finding, decision recommendation, binding rule — was introduced as a cross-cutting application pattern. Epistemic integrity (1.2.1) established the verification gates. The restraint principle (1.2.3) established reversibility and bounded scope. Answerability (1.2.7) mapped the four institutional transitions and the design requirements at each gate. This sidebar addresses the moral-hazard dimension: premature closure as a specific hazard — the moment when "good enough" coherence acquires coercive force, especially in incentive-driven and AI-amplified environments where fluent synthesis is cheap and authoritative presentation is easy.

Premature closure is a moral-hazard failure. The agent who hardens a judgment prematurely — converting a hypothesis into institutional force without the verification the lifecycle demands — retains the benefits of decisive action (speed, authority, the appearance of competence) while shifting the costs of inadequate verification to the parties who are constrained by the premature conclusion (who bear wrong policy, unjust enforcement, or misallocated resources). The consequence-agent separation that defines moral hazard is present in its full structural form.

The Premature-Closure Hazard

The staged-judgment lifecycle assumes that each transition is governed by a gate — a structural checkpoint that prevents the judgment from hardening until it has earned the level of force it will carry. Premature closure is the failure mode in which the gate is bypassed, weakened, or captured.

Three structural conditions produce premature closure.

Coherence as Authority

The first condition is the confusion of coherent presentation with verified truth. A judgment that is presented fluently, in authoritative language, with the structure and confidence of an established conclusion, creates social and institutional pressure to treat it as a finding even when its evidential basis warrants only hypothesis status. The coherence is persuasive. The persuasion short-circuits the verification gate.

This condition is dramatically intensified under AI conditions. Large language models produce coherent, authoritative text by design — their output has the structure and confidence of expert synthesis regardless of its evidential grounding. An AI-generated analysis that is internally consistent and fluently presented will pass the coherence test for institutional adoption even when its claims have not been independently verified, its assumptions have not been stated, and its scope has not been defined. The hazard is that the institution treats the presentation quality as a proxy for epistemic quality — a Goodhart effect applied to the judgment lifecycle itself.

The counter-gaming response: verification gates must evaluate evidence, not presentation. The institutional design requirement from (1.2.7-5) applies — each transition requires proportional evidence, and evidence is assessed by its grounding, not by the fluency of its delivery. AI-generated analysis must be subject to the same verification requirements as human-generated analysis, with the additional requirement that the sources and reasoning be independently confirmable.

Incentive Pressure to Conclude

The second condition is the structural incentive to resolve uncertainty quickly. Institutional actors who operate under time pressure, performance metrics that reward decisiveness, or political environments that penalize ambiguity face a structural incentive to harden judgments prematurely. The incentive is not to find the right answer — it is to find an answer that can be presented as right, quickly enough to satisfy the pressure.

This is a moral-hazard failure in the direct sense: the agent who concludes prematurely retains the benefit (appearing decisive, satisfying the timeline, meeting the metric) while shifting the cost (implementing a premature conclusion that may be wrong) to the parties affected by the conclusion. The consequence-agent separation is structural: the decider benefits from the decision's speed; the affected parties bear the consequences of its accuracy.

The counter-gaming response: the hazard scan's second diagnostic question — Which metrics have become targets? — must be applied to decisiveness metrics specifically. If an institution measures and rewards the speed of decision-making, the metric creates a structural incentive for premature closure. The Goodhart map's gaming-pathway analysis applies: can the metric be satisfied by producing fast conclusions rather than good ones? If so, the metric is a premature-closure engine.

Sunk-Cost Entrenchment

The third condition is the institutional reluctance to reverse a judgment that has already acquired force. Once a hypothesis has been treated as a finding — resources allocated, decisions made, commitments undertaken — the institution faces a structural incentive to defend the premature conclusion rather than revise it. The sunk costs create political capital that is forfeit if the conclusion is reversed. The agents who championed the conclusion face professional consequences if it is repudiated. And the institutional narrative has already been constructed around the conclusion's validity.

This is the reverse of the softening requirement from sidebar (1.2.7-5). The institutional design specifies that binding rules must include revision mechanisms. Sunk-cost entrenchment is the hazard that prevents those mechanisms from functioning — not because they do not exist but because the incentive structure makes their use costly to the agents who should invoke them.

The counter-gaming response: compensation clawbacks and disclosure rules apply to institutional judgment as well as to financial positions. If the agents who championed a premature conclusion face no consequence for the premature closure — and face significant consequence for admitting the error — the incentive structure rewards entrenchment. The design correction: institutional review processes must reward honest correction at least as strongly as they reward initial decisiveness.

The Governing Principle

The staged-judgment lifecycle, viewed through the moral-hazard lens, produces a specific extension of the unit's central claim: premature closure is a moral-hazard failure in which the benefits of decisive action are retained by the agents who act while the costs of inadequate verification are borne by the parties constrained by the premature conclusion.

The hazard scan's five diagnostic questions apply directly:

• Where can agents privatize gains and socialize losses? — wherever the agents who harden judgments retain the reputational benefits of decisiveness while the affected parties bear the consequences of error.

• Which metrics have become targets? — wherever institutional metrics reward speed of conclusion rather than quality of verification.

• Where is risk being shifted to parties who cannot see it? — wherever a premature conclusion constrains parties who were not consulted in the judgment process and cannot evaluate the evidence on which it was based.

• Where has oversight been captured? — wherever the review mechanism for the judgment is staffed by the same parties who championed the conclusion.

• Where is risk accumulating in the tails? — wherever a premature conclusion, if wrong, would produce consequences disproportionate to the cost of additional verification.

The principle connects systemic moral-hazard awareness to the staged-judgment thread's governing commitment from sidebar (1.2.7-5): no claim may acquire coercive force without a defined appeals/exit pathway. Premature closure is the specific hazard that violates this commitment — and the hazard scan, applied to the judgment lifecycle, is the detection mechanism.