Evidence can narrow, constrain, and improve a public decision without becoming the source of political authorization.
Part VI · Chapter 14
From Evidence to an Authorised Decision
Evidence can narrow, constrain, and improve a public decision without becoming the source of political authorization.
Learning objectives
What this chapter asks you to be able to do.
- Explain why evidence can constrain and inform public decisions without independently authorising them.
- Specify the proposition, jurisdiction, option set, time horizon, reversibility, resource commitment, and rights exposure of a public decision before evaluating it.
- Distinguish empirical disagreement from normative disagreement and identify when a policy choice requires political judgment rather than additional measurement alone.
- Evaluate cost-benefit analysis, expected-value reasoning, precaution, robustness, and adaptive approaches as decision aids rather than universal constitutions of public value.
- Explain why majority rule is powerful within a defined domain while remaining sensitive to membership, agenda, rights, thresholds, and institutional authorization.
- Introduce Condorcet cycles, Arrow's impossibility theorem, and decision-rule sensitivity without treating social choice theory as proof that democratic choice is futile.
- Explain why rights may constrain ordinary trade-offs while still requiring legal source, jurisdiction, interpretation, factual support, and review.
- Distinguish epistemic deference, expert delegation, political responsibility, and lawful human override.
- Identify when a nominal human decision is actually a rubber stamp for upstream control over options, salience, timing, and override costs.
- Produce Decision Dossier XIV: an Authorised Decision and Public Reason Record that preserves the complete transition from recommendation to legitimate human authorization.
Chapter summary
The argument in inspectable form.
A decision proposition should specify the action, scope, jurisdiction, population, time horizon, resource commitment, reversibility, implementation authority, and rights exposure.
Option-set construction is a location of political power. A formally democratic vote can still be upstream-captured if meaningful alternatives were excluded without scrutiny.
Empirical assessment, normative judgment, legal or constitutional constraint, and political authorization are distinct layers even when they interact.
Quantitative scoring can expose trade-offs, but weights and objectives are political or constitutional choices whose provenance must be recorded.
Rights are not automatically reducible to weighted preferences, but rights claims still require legal source, jurisdiction, interpretation, factual support, and review.
Cost-benefit analysis can clarify opportunity costs and welfare effects while remaining limited by valuation, distribution, discounting, uncertainty, and nonmarket values.
Expected-value, precautionary, robust, and adaptive approaches respond differently to uncertainty; none should be treated as a universal public decision rule.
Majority rule has strong procedural attractions in a defined domain but does not determine the electorate, agenda, rights constraints, quorum, or institutional authority.
Condorcet cycles and Arrow's theorem show that collective aggregation can be sensitive to rules and agendas; they do not prove democracy impossible.
Supermajorities, vetoes, and quorums can protect important commitments while also creating obstruction and minority veto power.
Decision-rule sensitivity should be exposed. Changing the rule after the likely winner is known is a form of procedural power.
Public-reason traditions differ, but they illuminate why authorized institutions should normally offer reconstructible reasons for coercive decisions.
Political judgment is bounded discretion under persistent factual, normative, legal, and institutional disagreement.
Epistemic deference does not erase political responsibility. Officials who depart from strong expert advice should preserve and explain the divergence.
Human authority includes the power to reject AI or expert advice; it does not make false factual premises true or legitimize every override.
A nominal human signature is not meaningful authorization when upstream systems control options, salience, timing, or practical override conditions.
Strong evidence, public support, expert consensus, urgency, or AI confidence cannot silently create jurisdiction.
Participation records should inform representative institutions without being collapsed into one public-preference score.
Minority and losing reasons should remain auditable because future evidence, appeals, or outcomes may make them relevant.
Current NousPolis canon strongly separates epistemic assessment, normative mapping, AI recommendation, and human authorization, but a mature system still needs to operationalize the full transition without allowing a `HUMAN_AUTHORIZED_DECISION` label to legitimate upstream capture.
Decision Dossier XIV records proposition, authority, option-set provenance, evidence, participation, rights, values, rule, reasons, dissent, human authorization, divergence, accountability, review, and Chapter 15 reopening conditions.
Key terms
Concepts to carry forward.
- public decision
- An institutional choice that authorizes, directs, permits, prohibits, funds, or otherwise commits public power within a defined jurisdiction and procedure.
- decision proposition
- The precise act or question being decided, stated narrowly enough that later reviewers can determine what was and was not authorized.
- option set
- The alternatives made available for authoritative choice, including status quo, delay, pilot, or non-intervention where materially relevant.
- political judgment
- Reasoned choice under conditions in which evidence, values, rights, uncertainty, and institutional responsibility do not mechanically determine one outcome.
- public reason
- A family of theories holding that the exercise of political power, especially coercive power, should be justifiable through reasons appropriate to citizens as political equals; theories differ about scope and content.
- majority rule
- A decision rule under which the alternative supported by more than half of the relevant votes defeats the other alternative, subject to the institution's membership and procedural rules.
- supermajority
- A decision threshold requiring support greater than a simple majority, often used for specially protected or entrenched decisions but capable of creating minority veto power.
- Condorcet cycle
- A pattern in which collective pairwise majority preferences can cycle, for example A over B, B over C, and C over A.
- aggregation fragility
- A condition in which a materially reasonable change in aggregation rule, threshold, sequencing, or admissible options changes the decision-relevant result.
- cost-benefit analysis
- A decision aid that estimates and compares monetized benefits and costs of alternatives; its interpretation depends on assumptions about valuation, distribution, discounting, uncertainty, and institutional purpose.
- rights constraint
- A legal, constitutional, or normative limit or heightened justification requirement that cannot simply be treated as another ordinary preference without establishing its source and scope.
- expected value
- The probability-weighted average outcome of an uncertain choice under a specified probability and value model.
- deep uncertainty
- A condition in which decision-makers cannot confidently agree on the relevant models, probabilities, or valuations needed for ordinary predictive optimization.
- robust decision
- A choice designed to perform acceptably across a wide range of plausible conditions rather than optimizing for one forecast alone.
- meaningful human authorization
- Human authorization in which the responsible actor has practical access to evidence, alternatives, uncertainty, dissent, and a realistic capacity to modify or reject the recommendation.
- override
- An authorized decision that accepts, modifies, or rejects an advisory recommendation while preserving the original recommendation and the reason for divergence.
- public reason record
- A reconstructible account of the proposition, authority, evidence, values, rights, decision rule, reasons, dissent, authorization, and review conditions attached to a public decision.
Review & discussion
Questions for seminar, revision, or assessment.
- Why can evidence eliminate some policy options without uniquely determining the authorized choice among the options that remain?
- Rewrite the proposition “Asterbridge should adopt the transport reform” so that it can function as an auditable authorization record.
- Give an example in which an option set is legally constrained and an example in which the same language could conceal agenda power.
- Why is a weighted multi-criteria score mathematically precise without being normatively self-justifying?
- When should a right be treated as a hard constraint, a rebuttable presumption, or a consideration in a balancing test? Why can this not be answered without jurisdiction?
- What information must accompany a positive cost-benefit result before an authorized institution can use it responsibly?
- Compare expected-value reasoning, precaution, and robust decision-making in a case with a small probability of severe irreversible harm.
- What does May's theorem establish about majority rule, and what does it not establish about the legitimacy of majority rule in public institutions?
- Construct a three-option Condorcet cycle for Asterbridge and explain how agenda order could change the final winner.
- What is the political lesson of Arrow's impossibility theorem if the lesson is not “democracy is impossible”?
- When can a supermajority requirement protect legitimacy, and when can it become minority domination?
- Why should decision-rule sensitivity be reported even when the institution is not required to change its precommitted rule?
- What does public reason add to a formally valid council vote?
- Give one example of a legitimate value-based override of expert advice and one example of an override resting on an empirical falsehood.
- Why is “human in the loop” insufficient as a constitutional safeguard?
- Which upstream conditions must be present for a human reject button to constitute meaningful override capability?
- Can a public decision be legally valid but democratically weak? Can it be democratically popular but unlawful? Explain both cases.
- How should an elected council treat a representative survey, a mini-public recommendation, and a self-selected consultation when they point in different directions?
- Why should losing reasons remain attached to a decision after authorization?
- Which Chapter 14 field would be most important in detecting method-shopping? Which would be most important in detecting option-set capture?
- What does current NousPolis canon already require about human authorization, and what would be misleading to describe as implemented public-decision infrastructure?
- Explain why a `HUMAN_AUTHORIZED_DECISION` object could still be constitutionally weak if the upstream option set and value weights were controlled by the orchestration layer.
- Design one decision for which a reversible pilot is preferable to immediate permanent adoption even though the expected-value ranking favours immediate adoption.
- At the end of Chapter 14, which questions now belong to implementation and learning rather than authorization?
Further reading
Continue into the literature.
For public justification and the relationship between democratic coercion and reasons, begin with Rawls's “The Idea of Public Reason Revisited” (1997), then compare it with Habermas's *Between Facts and Norms* (1996) and Gutmann and Thompson's *Why Deliberative Democracy?* (2004). Pettit (1997) is useful for connecting public power to contestability rather than treating periodic electoral consent as sufficient. For social choice, Arrow's *Social Choice and Individual Values* (1963) remains foundational; May (1952) provides a compact characterization of simple majority rule, while Schwartzberg (2013) examines the origins and limits of supermajority rules. Readers should treat these works as tools for understanding institutional trade-offs rather than as a substitute for democratic theory. For decision analysis, Boardman et al. (2018) provide a practical introduction to cost-benefit analysis; Adler (2016) is particularly useful on distributional weighting and the assumptions behind welfare aggregation. The European Commission's 2000 communication offers an institutional formulation of precaution that includes proportionality, action-versus-inaction comparison, and review rather than a simple command never to act under uncertainty. Lempert (2019) and the open-access volume edited by Marchau et al. (2019) introduce robust and adaptive approaches to decision-making under deep uncertainty.
Companion, not replacement
The textbook carries the complete argument.
This page reproduces the chapter's study and navigation layer from the current living manuscript. The Asterbridge narrative, historical and methodological argument, figures, Political Science Lens boxes, NousPolis Canon crosswalks, labs, red-team exercises, and N5 Practice boundaries remain in the canonical textbook rather than being republished wholesale here.