Courts and Adjudication
In a state whose corrective acts are as often algorithmic as human, the courtroom is where the machine is made to answer. This chapter describes the bench of a value state: how judges are made independent, how a contested reading is tried, and how the law reaches harms that have a million small victims and no single culprit.
What a court is for when the defendant may be the state's own arithmetic
A notice arrives at a logistics firm on an ordinary morning: a rebalancer has booked a provisional anti-value charge against a class of its shipments, citing a measured ecological effect, a magnitude, an affected cohort, and the commit-hash of the pricing rule applied. The owner reads it twice, decides the attribution is wrong, and presses a single word: contest. From that moment the charge is no longer an act of the state. It is a claim before a court, and the state must now prove it like anyone else.
This is the first thing to understand about adjudication in Axiacracy: the interpretive authority of the state's agents is rebuttable, never infallible. A rebalancer's assessment is authoritative by default, and it earns that default not by rank but by transparency: its logic is public and versioned, its calibration record is on the ledger, its conflicts are disclosed. But an anti-value declaration is only a claim in a lifecycle that runs from provisional to attested to settled, and any affected party can arrest that lifecycle and demand review. The court is the review authority for every contested reading, and it can uphold the charge, adjust it, or reverse it entirely. Above the court sit further courts, and above them constitutional review; there is no final, unaccountable authority anywhere in the structure, only escalation bounded by the Charter.
The dockets of such a judiciary are fourfold. Ordinary disputes remain, because contracts are still broken and rights-violations are still tried under criminal law; a value state does not abolish the quarrels of ordinary life. To these are added three dockets no earlier judiciary carried: review of the rebalancer's corrective acts, contestation of the aggregates themselves, since a cohort reading or an attribution can be wrong and a citizen must be able to say so, and Charter enforcement, the power to void the acts of every other organ, including the legislature and including the algorithms. The courts, in short, are where the sensing state is itself sensed.
The independence of the bench
None of this is worth anything if the judge can be starved or dismissed. Smith stated the requirement with a bluntness that has never been improved upon: a judiciary whose pay depends on the executive's good will cannot be independent, and funding it from a volatile source is improper for an institution which ought to last forever. Axiacracy writes both halves into the hierarchy of norms. Justice holds a stable, constitutionally fixed endowment that no budget vote, no ministry, and no emergency can reduce, and its judges are irremovable. The bench answers for its judgments, never for its livelihood.
Selection follows the same principle that staffs every position in the apparatus: the office goes to whoever delivers the most attested adjudicative quality, and a judicial post may in principle be held by a human or an AI agent, each carrying a public calibration record that any litigant can inspect. But the doctrine reserves the apex. Constitutional adjudication, the final word on what the Charter means, is constitutionally reserved to human judges, because at the summit of the system legitimacy and accountability must rest with beings who can be answerable in the way the Charter's authors were answerable. Below the apex, accountability runs not through removal, which would be a lever in the executive's hand, but through daylight: every ruling is appealable, every judge's record is public, and escalation carries any error upward until it meets the constitutional level.
Two further safeguards close the circle. No axis-ministry may adjudicate a capture claim against itself; the register in which the state is accused is never kept by the accused. And because Smith also warned that positive law is always a more or less imperfect attempt at natural jurisprudence, warped by the interest of government or of the particular orders who tyrannize it, the enacted floors and coercive rules are themselves checked against an independently maintained benchmark of natural justice. Even the justice tier is audited for capture, including capture by its own ministries and its own machines.
Procedure: the countable-effect standard
What may a court of the value state actually try? Only a manifested act. Hobbes observed that of intentions which never appear by any outward act there is no place for human accusation, and that a crime is only such a thing as may be made to appear before a judge. Axiacracy keeps his filter intact under far more tempting conditions: the sensor may read much, but nothing crosses into the coercible set except a perceptible, manifested act, and sensed-but-unmanifested trouble goes at most to care and observation, never to the docket. The state's knowledge is not the court's jurisdiction.
Once an act is before the court, the evidence is counted rather than intuited. The doctrine's oldest line holds here with full force: individual value is subjective, but effects are objectively measurable, and cardinality lives in effects. A charge must specify magnitude per axis, the affected set as read from the flow record, and the actor's intent, and the accused may contest every element. Liability reaches only what was foreseeable at the time of action; no rule may make a person answer for remote consequences beyond any ability to foresee, and no charge may exceed the effects that can actually be counted.
The norms against which those effects are scored are frozen in time. Every governing norm, each weighting vote's result, each version of floor and ceiling and band, each externality pricing function, is recorded on the log with a validity interval, and the court asks one question of the archive: what norm-set was binding at the moment of the act? Evidence updates retroactively, because we legitimately learn what actually happened; norms never do, because judging yesterday's act by today's weights is ex-post-facto law, what Hobbes called not punishment but hostility. The deterrent charge is likewise capped at the penalty foreknown when the act was done. One apparent exception proves the rule: concealed harm remains liable however late it surfaces, because hiding harm violated the contemporaneous norms too; the discovery is late, not the law.
Finally, the reasoning itself is disciplined. A coercive act is valid only if an independent tribunal, knowing nothing but the public rules and the facts of the case, and not the sensor's reading, could reach the same result. Where the aggregate valuation is the true ground of the decision, the act is void. The citizen is judged by rules that could be known in advance, never by a reading that cannot be inspected; lawfulness means predictable, reviewable outcomes. And the court measures the state's own response with the same instrument it applies to the offense: a corrective act is scored by the same anti-value function, and any excess over the defensible level is charged back to the state's own ledger. Why proportionality takes exactly this form, restoration and deterrence as two distinct tracks, is the subject of the next chapter; here it is enough that the court holds the scale.
The citizen's path
Standing belongs to any affected party, and the path is short. An appeal opens a case; the court assembles the contemporaneous norm-set, the evidence including any new facts, and a reviewer independent of the agent that made the charge; the ruling upholds, adjusts, or reverses. Rework between court and agent is bounded, and a case that will not settle escalates to a higher court and ultimately to the constitutional level. The entire proceeding lives on the verifiable log, public and reproducible, so that any citizen can replay the case and confirm that the published rules produced the published result.
Above the procedure stands a guarantee that defines the whole system's temperament: no algorithmic act is ever final against a human being. The sensing engine may inform rules, but it may never be given what Hayek warned against, authority to give any decision the force of law, unquestioningly accepted by the courts. An algorithmic output is never self-executing coercion; it is input to a general rule that a distinct office administers under review, and the ladder of appeal from any such act ends, at its top, before a human bench. The citizen who contests the machine is therefore never contesting a verdict, only a claim, and the right to carry that claim to human judgment is not a courtesy of the system but a condition of its validity.
The court's deference runs, pointedly, in the citizen's favor. Any correction that reaches into the private sphere must be rule-defined, judicially reviewable, and fully compensated, permissible only where the public gain clearly and substantially exceeds the loss, with the benefit of the doubt given to the individual. And there is one thing the tribunal itself may never do: it judges rule-conformity and never re-weighs the Æ-vector, because value-questions that cannot be decided by the application of rules do not belong to judges, and value-arbitration in judicial robes merely launders discretion. The court guards the frame; it does not hold the pen that draws it.
The new docket: systemic harm
The distinctive caseload of a value state is the harm with a million small victims and one diffuse cause. A spoiled product is easy law: one flow, one culprit, a named victim. But a youth cohort's cognitive axis declining faster than its counterfactual baseline, attributable to a whole class of attention-capturing flows, fits no traditional writ. The event is a pattern over many flows, the victim is a cohort, the causation is statistical, and the culprits are many. Twentieth-century law met this shape late and reluctantly, in the tobacco, opioid, and climate litigation; Axiacracy makes it a standing docket.
The trigger is a measured shift: a cohort's axis crossing a threshold beyond its baseline is itself the event that enters adjudication. The causal standard is population-attributable: the claimant organ must establish that the accused class of flows produced an excess of harm beyond the counterfactual, and the accused have full right to contest the confounders, the baseline, and their shares. Responsibility is apportioned by contribution, each actor's proportional exposure, with a multiplier for concealment, and each may argue that its design was not the harmful one or its share is smaller. Restitution is routed class-wide, to the damaged axis of the harmed cohort rather than to named individuals, funding the remediation of what was actually damaged, and the case is brought on the cohort's behalf by the relevant ministry, by the Equity guardian, or by citizen initiative, so that no individual victim need litigate at all.
Where the harm may become irreversible before it can be proven, the court's toolkit bends without breaking: early, soft, precautionary measures on lower confidence, escrow held against the suspected harm and sized to the detection lag, and escalation to genuine coercion only as confidence rises. It is a bounded precautionary principle, never full force on weak evidence. The doctrine is honest about the difficulty: systemic causation is genuinely hard and contestable, and what is claimed is not clean measurability but a disciplined, contestable, precaution-bounded procedure where before there was nothing at all.
Where the hierarchy of norms becomes real
A hierarchy of norms is words on a page until some institution can act on its ordering, and the courts are that institution. A Charter-violating enactment is void ab initio and grounds no obligation, whoever enacted it and however large the majority behind it. A coercion outside the legitimacy wall, without authority, without antecedent published rule, retroactive, or beyond the foreknown penalty, is not defective punishment but hostility, and void. An offense keyed to the sensor's reading rather than to a published rule, some elastic charge of conduct against systemic balance, is void for vagueness, the modern descendant of the old catch-all of public mischief. And because no framer can enumerate every future vector of infringement, the Charter's listed rights are a floor and not a ceiling: the tribunal may void a novel intrusion no text foresaw.
Hobbes could never call a sovereign's law unjust, because the sovereign was its author and he that is bound to himself only is not bound. Axiacracy dissolves the circle by separating the binder from the bound: the constituent Charter above, the constituted code below, and a bench with the endowment, the tenure, and the duty to strike the second whenever it touches the first. That bench is where the promise of Chapter 1, rights above any majority, stops being a preamble and becomes a working fact of daily governance, binding strongest exactly when it is most tempting to break it.
In the doctrine
The doctrinal ground is Rights, Crime and Justice and The Charter and the Hierarchy of Norms; the deep sources on procedure and the coercion gate are Hobbes and Hayek.