Question
Preserve the lived problem, competing frames, standing, and the institutional question before one administrative definition hardens into the answer.
The framework
One architecture at three levels of detail.
The Reasoning Polity uses five stages to explain the idea, six recurring dossier functions to classify the work, and fifteen records to preserve the sequence. These are different views of one architecture—not competing lifecycles.
The Public-Reasoning Chain
Question → Evidence → Challenge → Authorization → Learning.
A reasoning polity is an institutional process in which claims can be inspected, challenged, assigned to responsible actors, authorized by legitimate human institutions, and reopened when reality changes the record.
Evidence
Research, measure, compare, interpret, and synthesize with provenance, uncertainty, alternative explanations, and explicit limits.
Challenge
Expose assumptions, dissent, rights claims, trade-offs, methodological weaknesses, and category errors before they become invisible.
Authorization
Carry the best-supported options to the legitimate human institution that has the authority—and responsibility—to decide.
Learning
Compare the authorized baseline with implementation and outcomes, then reopen inquiry when evidence or institutional conditions justify it.
Learning → QuestionMonitoring can justify inquiry or reopening. It does not automatically prove cause or authorize a new policy.
Three views of the same process
5 stages explain. 6 functions classify. 15 records preserve chronology.
The levels deliberately do different work. The function labels classify what kind of work is being performed; the record numbers preserve the chronological sequence.
Public-facing stages
Question, Evidence, Challenge, Authorization, and Learning give readers the simplest constitutional explanation of the reasoning chain.
Dossier functions
Define, Constitute, Investigate, Synthesize, Decide, and Learn classify the work. A function may recur later.
Dossier records
The records preserve the accumulating reasoning history. Readers can follow what changed without regrouping the sequence by function.
Six dossier functions
Functional categories, not six consecutive phases.
Preserve the public question and competing frames before administrative refinement becomes an answer.
Make standing, authority, jurisdiction, distribution, power, legitimacy, and the AI boundary explicit.
Design research and build qualitative, quantitative, causal, and computational evidence with visible limits.
Bring heterogeneous evidence and participation into an inspectable record without manufacturing consensus or authorization.
Record the binding human-authorized decision, its authority basis, reasons, dissent, advice, and overrides.
Compare implementation and outcomes with the authorized baseline, preserve adaptations, and define reopening triggers.
5 ↔ 6 ↔ 15 crosswalk
The relationship is many-to-many conceptually.
Every dossier record has one primary function, but a public-reasoning stage can depend on several functions. Challenge in particular cross-cuts the record and can reopen earlier work.
Public stage
Question
Related dossier functions
Define · Constitute
Define preserves the public question; Constitute constrains framing through standing, jurisdiction, authority, and power.
Public stage
Evidence
Related dossier functions
Investigate · Synthesize
Investigation creates and tests evidence; synthesis brings heterogeneous findings together without manufacturing consensus.
Public stage
Challenge
Related dossier functions
Constitute · Investigate · Synthesize
Challenge cross-cuts the record and can reopen assumptions, methods, standing, authority, evidence, or interpretation.
Public stage
Authorization
Related dossier functions
Constitute · Decide
Constitutional and institutional constraints determine who may act; Decide makes the human authorization point explicit.
Public stage
Learning
Related dossier functions
Learn · Define · Constitute · Investigate
Implementation and monitoring create reopening triggers that may send the polity back to framing, authority, or investigation.
Safeguards
The system must remain challengeable.
Transparency is necessary but insufficient. A reasoning architecture is defensible only when people can understand where claims came from, challenge how they were produced, and identify who remains responsible.
Human authority
AI systems advise, research, compare, and critique. Legitimate institutions retain the power—and responsibility—to decide.
Epistemic pluralism
Different models, disciplines, stakeholder perspectives, and value claims reduce dependence on a single reasoning path.
Auditability & provenance
Important claims, evidence, transformations, disagreements, and decisions remain traceable enough for later scrutiny.
Contestability & reopening
Minority reports, adversarial review, appeal, monitoring, and explicit reopening triggers prevent “the system said so” from becoming an endpoint.
“AI analyses. Humans authorise.”Evidence before authorization; jurisdiction and legitimate authority constrain the process throughout.
See the framework in use
Follow the case or inspect the reasoning record.
Asterbridge shows the political problem unfolding across the book. The Decision Dossier shows how the reasoning can be preserved, challenged, and revisited.