Decentralized Justice: An Exitarian Framework

This text has a predecessor. Years ago I drafted a decentralized justice system and called it libertarian. The label was wrong. Not because the mechanisms were wrong; the web of trust, the competing arbitration, the oracle game theory all survive this rewrite intact. The label was wrong because libertarianism is one ideology, and I do not believe in one ideology. I believe in Exit.

Libertaria is not a libertarian project. It is a federated netstate of chapters that share almost nothing except a Grundgesetz and one operational conviction: the right to leave is the only right that enforces all the others. Some chapters will be property-absolutist. Some will be communal. Some will tax, some will not, some will run their courts on silicon, some on elders. They will have their own laws, their own constitutions, their own flags. What follows is not the law of Libertaria. It is a reference implementation: a justice architecture any chapter can adopt, adapt, or reject, and a description of the federal layer that keeps a thousand different justices from collapsing into one.

Three corrections to the old draft run through everything below:

1. Exit replaces ideology as the load-bearing mechanism. A system you cannot leave is a cage regardless of how elegantly its incentives are designed. Every mechanism here is checked against one question: does it make leaving easier or harder?
2. There is no best system. Chapters hold genuinely incompatible values. The framework is polycentric by construction, not by compromise.
3. Markets do not solve everything. The cryptoanarchist blind spot is information asymmetry. A market in which one side knows more than the other is not a discovery mechanism; it is an extraction mechanism. Wherever this framework relies on markets, it therefore also specifies guardrails and incentives that steer actors toward honesty and openness. Reputation systems fail without disclosure obligations. Arbitration fails without discoverable records. Oracles fail without skin in the game. The guardrails are not decoration. They are the difference between a free market and a fraud market.

And one addition the old draft could not have made: the silicon-carbon symbiosis is baked into our Axioms. Artificial intelligence is not a tool a human consults and then blames or excuses. In this framework, silicon agents are first-class participants: they hold identities, accumulate reputation, attest, arbitrate, and are held accountable through the same graph as everyone else. The symbiont bond is not a metaphor here. It is a jurisdictional fact.

The problem

Current judicial systems suffer from a structural defect: those who decide the lives and freedom of others face no consequences for their mistakes. A judge who releases a murderer who kills again, or who imprisons an innocent person, pays no real price. Accountability is absent.

Automatically punishing judges for negative outcomes is not a solution; it would create decisional paralysis or flight from the profession. The problem is not the individual judge. The problem is the monopoly: the state monopoly on justice, and the absence of any effective feedback mechanism between error and consequence.

The old draft stopped there. The exitarian analysis goes one layer deeper. A monopoly on justice is only possible because the victims of bad justice cannot leave. The judge is unaccountable because you are captured. Taxation without consent, jurisdiction without consent, law without consent: all of it rests on the same physical fact; exit from a territorial state costs everything you have. Monopoly is not the root defect. Blocked exit is.

So the design goal of this framework is not “better judges.” It is a justice architecture in which:

• no actor, carbon or silicon, is shielded from the reputational and economic consequences of their errors;
• every participant can exit any arbitration relationship, any insurance relationship, any chapter, at any time, taking their identity, their history, and their assets with them;
• and no chapter can prevent that exit without declaring itself, publicly and in the graph, a prison.

The architecture rests on three pillars: persistent reputational identity, competing arbitration, and decentralized enforcement. Around them sits a federal frame: the Grundgesetz and the chapter structure of Libertaria.

The federal frame: chapters under one Grundgesetz

Before the pillars, the skeleton they hang on.

Libertaria is a federation of chapters. A chapter is a self-governing community: it writes its own laws, adopts its own constitution, flies its own flag, runs its own courts or arbitrators, sets its own property regime, its own contribution rules, its own norms of conduct. One chapter may approximate the property-absolutist society the old draft described. Another may hold land in commons and require contribution. A third may be a religious community with canon courts. The framework does not adjudicate between these. We do not believe there is one best system; we believe there are many systems, and that people sorted into systems they chose produce better outcomes than people conscripted into systems they were born into.

The Grundgesetz is the thin federal layer above all chapters. It does not contain a legal code. It contains invariants; the small set of rules that make plurality possible without civil war:

1. Exit is inviolable. Any person may leave any chapter at any time, with their identity, reputation graph, credentials, and assets. No chapter may bind, indenture, or detain a person for choosing to leave. This is mechanism number one; everything else is commentary.
2. No chapter may externalize onto another. A chapter’s internal rules end at its boundary; disputes crossing boundaries go to inter-chapter arbitration under pre-negotiated treaties.
3. Reputation infrastructure is neutral and portable. The web of trust protocol, the identity layer, and the oracle infrastructure belong to the federation, not to any chapter. Forking your subgraph and leaving must always be technically possible.
4. Identity is self-sovereign. No chapter, agency, insurer, or oracle consortium may issue, revoke, or own a person’s cryptographic identity. They can attest about you. They cannot be you.
5. Symbionts are persons for accountability purposes. Silicon agents participating in the framework hold identity and reputation, and every silicon decision traceable to a principal binds that principal. Neither “the AI did it” nor “the user did it” is a complete defense; the accountability chain is explicit.

Everything not on that list is chapter law. The Grundgesetz is deliberately minimal because every additional federal rule is a reduction of the exit option space. A thick constitution is a cage with good intentions.

Why have a federation at all, then? Because pure fragmentation fails at three points: disputes that cross chapter boundaries, threats that exceed one chapter’s capacity, and infrastructure (identity, graph, oracles) that only works if it is shared. The federation exists to solve exactly those three problems and nothing else.

Pillar 1: The Web of Trust, carbon and silicon

The concept of a web of trust was introduced by Phil Zimmermann in 1992, in the PGP manual, as a decentralized alternative to certificate authorities. Here it is extended far beyond its cryptographic origin, to social reputation, justice, and governance.

The mechanism

Every person, carbon or silicon, possesses a cryptographic identity. Those they interact with attest to them: commercial transactions, neighborhood relations, collaborations, arbitrations completed, contracts honored. These attestations form a social graph in which trust is never an absolute value; it is always relative to the observer. Each node sees the graph from its own position, and trust toward another node is a function of the paths connecting them.

Silicon agents are nodes of the same kind. An arbitration symbiont accumulates a track record exactly as a human arbitrator does. A service agent that cheats is flagged exactly as a human fraudster is. This is the symbiosis axiom made operational: one graph, two substrates, one accountability physics. The silicon node has an advantage (perfect memory, auditable decision logs where the operator chooses to publish them) and a disadvantage (its operator’s identity is bound to it, so a rogue agent burns a human’s or a collective’s capital). Neither substrate is trusted by default. Both are trusted by path.

Distinction from centralized reputation

The web of trust must be sharply distinguished from a centralized reputation system, with which it is constantly confused. In a centralized system, a single authority collects all judgments and aggregates them into a universal score: one number per person, identical for all observers, determining access to services and relationships. Control over the number is control over the person.

The web of trust differs on every dimension:

• No universal score. Trust is observer-relative. If I trust you and you trust a third party, I hold some indirect trust toward them; another observer with a different path may hold none. There is no ranking that defines you for everyone.
• No central authority. Each node manages its own trust list sovereignly. The system is emergent, not administered.
• No single point of control. Even a highly connected node issuing an unjust judgment poisons only the subgraph that trusts it. The rest of the network is unaffected.

A centralized score is a tool of top-down control. A web of trust is a tool of bottom-up navigation. The difference is not degree but nature, and the Grundgesetz’s neutrality invariant exists precisely to keep the federal graph from ever hardening into a score.

Structural defense against Sybil attacks

A thousand fake identities attesting to each other have no connections to real nodes. For any legitimate observer, those attestations are invisible: there is no trust path leading to them. The principle is the same as search-engine ranking; a site linked only by pages nobody visits stays invisible. Sybil clusters do not need to be detected and purged. They are structurally irrelevant.

The empty identity as deterrent

In a mature system, absence of history is itself a signal. A thirty-year-old, carbon or silicon, with no attestations is immediately suspect: everyone understands they burned their original identity, probably to hide something. This creates an asymmetric cost; burning an identity means losing a lifetime of accumulated social capital, and the profit of any single dishonest act almost never covers that loss. In early adoption phases some will try identity resets; as the mechanism becomes universally understood, the empty identity on an adult becomes a red flag no counterparty ignores.

For silicon nodes the same logic binds with one refinement: a symbiont whose decision logs are open accumulates verifiable history, which is worth more than assertion. Openness becomes a competitive advantage for agents, which is exactly the incentive direction the framework wants. The guardrail and the market pull the same way.

Defense against corruption

Corrupting real, well-connected nodes is theoretically possible and structurally disincentivized. Whoever sells false attestations gambles their identity forever. Self-selection reinforces this: those willing to risk everything for a bribe tend to be those with little social capital, whose attestations carry little weight. The system self-immunizes.

Pillar 2: Arbitration, insurance, and the guardrails against market failure

No judicial monopoly, inside or between chapters

Within a chapter, justice follows chapter law: whatever courts, arbitrators, elders, or symbiont juries the chapter’s constitution establishes. The reference model this document recommends to chapters is competing arbitration: multiple agencies, each chosen voluntarily, resolving cross-client disputes through pre-negotiated agreements. The accountability mechanism is the reputational market; an agency that systematically frees dangerous people loses clients, an agency that cages innocents loses clients, and negative attestations propagate through the graph to every potential customer.

Between chapters, the same mechanism operates at one level up. Chapter A and Chapter B do not share law; they share treaties. An inter-chapter dispute goes to an arbitration body both chapters pre-committed to, the way their members pre-commit to agencies. A chapter that habitually refuses arbitration or defects from its own treaties acquires a federation-level reputation, and the consequences are the same ones an individual faces: other chapters restrict commerce, travel, and recognition. A chapter, like a person, can ostracize itself.

Note what Exit does here. A member of a chapter whose justice is degrading is not limited to voice (reform the chapter from inside, fight its politics, lose). They can leave for a chapter whose justice suits them, and they take their graph with them. Chapters compete for members the way agencies compete for clients. Voice is expensive and rigged. Exit is cheap and honest. That asymmetry is the engine of the whole federation.

Detention and compensation

Detention does not disappear, but it changes hands and purpose. In a world where AI and robotics drive the cost of operating secure facilities toward zero, automated detention is a practicable, low-cost tool. For serious crimes, the perpetrator is held in facilities operated by their insurer or by competing detention agencies; the victim is compensated directly and immediately by the insurer, which absorbs detention as an operating cost. For lesser offenses, reputational consequence plus insurance pressure suffice. Flight is handled in Pillar 3.

One chapter-level note: some chapters will reject detention outright and rely on restitution, exile, or rehabilitation regimes. That is chapter law. The Grundgesetz constrains only one case: detention used to prevent exit. A “debtor’s cell” that holds a member who wants to leave is a federal violation, because it converts a chapter into a cage.

The insurance layer

Insurance companies are the intermediate layer between reputation and enforcement. An individual subscribes to liability insurance; if they cause harm, the insurer compensates the victim. The insurer has a direct economic interest in assessing and monitoring risk: premiums rise for dangerous clients, and the uninsurable face economic ostracism in any community where insurance is the standard guarantee offered to counterparties.

Insurers also strengthen arbitration: every unresolved dispute is a cost, so insurers become natural promoters of fast, impartial agencies. The result is three levels of pressure before physical ostracism: reputation, premium, uninsurability. Most antisocial behavior is corrected at the first two.

Guardrails: where the old draft was naive

Here is the correction the libertarian draft needed and did not have.

Insurance and arbitration are markets, and markets fail predictably under information asymmetry. An insurer that knows more about its own loss statistics than its clients do will misprice in its own favor. An arbitration agency that controls the publication of its case record will launder its history. A security agency whose use-of-force reports are self-authored will write novels. The cryptoanarchist answer (“the market will sort it out”) is wrong whenever one side of the market can see and the other cannot. Reputation mechanisms only discipline actors whose behavior is visible. So the framework makes visibility a structural requirement, not a hope:

• Disclosure as license condition. Any arbitration agency, insurer, detention operator, or security agency operating in a chapter that adopts this framework publishes its case statistics, outcomes, premiums, recidivism rates, and use-of-force incidents to a public, append-only, timestamped ledger. The data format is a federal standard; the analysis is anyone’s business. An agency that stops publishing does not get banned. It gets read correctly: in a system where everyone publishes, silence is a confession.
• Adversarial audit as a business model. Competing insurers and agencies have a direct incentive to audit each other’s disclosures and publicize discrepancies, because every frightened client of a competitor is a potential customer. The framework weaponizes rivalry for transparency; mutual surveillance by competitors is the cheapest regulator ever built, and unlike a state regulator it cannot be captured, because capturing it requires capturing every competitor at once.
• Oracle-verified reporting. Material disclosures (did this detention actually happen as reported; was this compensation actually paid) can be staked and resolved through the same decentralized oracle mechanism described below. Lying in a disclosure stops being a PR problem and becomes a financial event with immediate cost.
• Honesty is the profitable strategy by design. In a transparent market, an agency with good real numbers gains more from publishing than it could ever gain from faking, because faking is detectable and detection is fatal. The guardrails do not restrain the market; they aim it. Openness becomes the dominant strategy, which is the only way openness ever survives.

The same logic extends to healthcare, research, and charity: any institution asking for trust must first publish the data that makes trust verifiable. Reputation without disclosure is astrology. The web of trust only works if the graph has true things to propagate.

Insurers as risk assessors, with the conflict of interest named

Insurers, whose core competence is risk assessment, can decide whether releasing a perpetrator is safe, guarantee the victim immediately, and manage detention through low-cost automated facilities. The victim is compensated without years of proceedings.

The structural conflict of interest must be named: the insurer profits by detaining longer than necessary, because recidivism is its cost. The brake is not goodwill. It is the emergent veil of ignorance: every client watching a disproportionate detention thinks “that could be me,” and clients migrate. Add the guardrails above and the brake hardens: detention durations are published, competitors advertise against abuse, and oracles can resolve disputes about whether a release decision was justified. The mechanism that restrains oracle abuse restrains insurer abuse; same physics, different substrate.

Silicon in the loop: symbionts, not tools

Arbitrators and agencies, carbon or silicon, will use AI for evidence processing, inconsistency detection, precedent comparison, probabilistic assessment. The old draft insisted AI remain “a tool, never the decider, because responsibility must stay human.” The symbiosis axiom replaces that framing with a cleaner one:

Responsibility follows the accountability chain, whatever substrate the decision ran on.

A human arbitrator advised by an AI remains responsible; consulting software has never transferred liability and never will. A silicon arbitrator registered as an autonomous economic agent is responsible through its bond: every symbiont operating in the framework is either (a) bound to a principal whose identity and capital back its decisions, or (b) backed by its own staked capital and insurance, exactly like a carbon actor. There is no third case in which an unaccountable ghost makes binding decisions. “The AI did it” names a defendant, not an excuse.

Two consequences:

• Parties may voluntarily delegate a dispute to a symbiont arbitrator, alone or in a mixed carbon-silicon panel. For minor and technical disputes, algorithmic speed and auditability may be strictly preferable. Consent makes it legitimate; staking makes it safe.
• Symbiont arbitrators with open decision logs can build stronger reputations than closed human ones, because their reasoning is inspectable. Expect a market premium for legible minds, and expect chapter law to diverge on how much silicon a court may contain. That divergence is not a problem. It is the federation working as designed.

What AI does in every case is improve the quality of information available at decision time, which is, ultimately, the only thing that actually reduces judicial error.

Pillar 3: Decentralized enforcement

When reputation, insurance pressure, and uninsurability fail, the system escalates to direct enforcement.

Ostracism in a federation

For serious crimes, security agencies move first: capture and delivery to the detention facility, as in Pillar 2. If the perpetrator escapes, the web of trust becomes the second enforcement layer. Positive attestations evaporate; negative attestations follow the identity everywhere in the graph. Expulsion from community after community follows.

In the old libertarian framing, this ended in permanent exile: the ostracized person survives only in total isolation. The federation sharpens and softens that outcome at the same time. Sharper, because a federal graph means there is no friendly neighboring jurisdiction that has simply never heard of you; your attestations precede you into every chapter that reads the graph. Softer, because exit cuts both ways: chapters differ, and a chapter may exist that accepts people with burned histories under strict conditions (restitution, monitoring, probationary membership). The federation does not promise redemption, but it does not forbid it either. What it forbids is erasure: no chapter may launder an identity by issuing a fresh one, because identity issuance is not a chapter power. A person may start over socially. They may never start over cryptographically. The empty identity follows them as a flag.

Enforcement inside a chapter follows chapter law on property and force. A property-absolutist chapter will treat trespass by an ostracized person as forcibly repellable. A commons chapter will handle boundary differently. The federal invariant is only this: force used in enforcement is disclosed, attested, and carries reputational consequence for whoever uses it. Violence without visibility is the state all over again.

Scalable multi-level enforcement

For threats exceeding a single chapter (cartels, trafficking networks, cross-chapter criminal organizations), the escalation protocol:

1. Chapter level: the directly affected chapter’s agencies intervene.
2. Regional level: agency and chapter consortia activate pre-negotiated mutual-assistance treaties, pooling resources and intelligence.
3. Federal level: crowdfunding plus decentralized oracles, with inter-agency reputation visible in the graph. Agencies that cooperate effectively earn collective attestations; agencies that refuse cooperation on clear cases lose clients and chapter contracts.

A security market at scale, without centralization. Whoever coordinates best wins; not whoever is biggest a priori.

Organized violence: crowdfunding and smart contracts

The federal level is realized concretely:

1. Anonymous crowdfunding: the affected communities finance neutralization of the threat. Anonymity protects contributors from retaliation.
2. Smart contract: funds lock with verifiable release conditions.
3. Competing security agencies execute and claim the compensation.
4. Decentralized oracles verify completion and release the funds.

Decentralized oracles: game theory and incentives

The Truthcoin/Augur mechanism

Oracles hold tokens that weight their votes. The incentive structure pushes toward truth through complementary mechanisms:

Positive economic incentive. If oracle votes align with verifiable reality, the system attracts use and funding; if they diverge, trust collapses and the money stops.

Prisoner’s dilemma. An oracle voting against the majority loses tokens, redistributed partly to the majority. No oracle can be certain a corrupt majority exists. Truth-telling is individually the safest strategy.

Double-agent incentive. Even a formed corrupt coalition is unstable: every member profits by pretending to join and then voting honestly, landing on the winning side while the betrayed conspirators lose. Asking a coalition of liars to communicate honestly with each other is paradoxical by construction.

Measurable oracle risk. Between event and resolution, residual prediction-market odds are a real-time price of corruption risk, visible to everyone. The system has a thermometer, and the thermometer is public. This is the guardrail pattern applied to the guardrail itself: even the honesty mechanism is subject to disclosure.

Self-interest as a brake on abuse

Oracles are anonymous but real; they have lives, families, vulnerabilities. No oracle is incentivized to promote action without near-certainty, because they or someone they love may one day stand on the other side. A Rawlsian veil of ignorance, but emergent rather than theoretical: each oracle implicitly designs the system’s rules as if they could be its subject.

Emergent ethical filtering

In ambiguous cases, the rational oracle abstains; voting in an unclear case risks landing in the minority and losing tokens. The result is natural filtering:

• Clear cases: high participation, mechanism fully operational.
• Ambiguous cases: low participation, contract unexecutable.
• Illegitimate cases (vendettas, elimination of innocents): no participation, mechanism blocked.

No external rule defining “legitimate contracts” is needed. Consensus emerges from willingness to participate.

Global distribution and anonymity

Oracles are globally distributed and anonymous. No one knows the cultural, religious, or political composition of the oracle set, so no bloc can be organized. Anonymity is a security property, not a side effect: an oracle that cannot be identified cannot be bribed or threatened. The only coordination channel left is reality itself, and voting with reality is exactly what the system rewards.

The system fails toward inaction

The direction of failure matters more than the rate of success. The “veto risk” (oracle non-participation rendering a contract unexecutable) is not a flaw. It is the most important safety property in this document. Better a guilty person unfunded than an innocent person killed. State systems fail toward action; drones, raids, executions, with the error discovered afterward. This framework fails toward inaction, which in matters of life and death is the only acceptable default.

Additional considerations

Parental responsibility

Parents bear responsibility for the children they bring into the world, and the quality of upbringing reflects on parental reputation in the graph. A persistently badly behaved child generates negative attestations toward the parents; an incentive toward responsible parenting that exists nowhere in structured form today. Universal access to AI as an educational instrument gives every individual, whatever their family context, access to knowledge and decision capacity; an individual so equipped remains responsible for their own actions. Symbiont tutors are chapter choice and family choice, not federal mandate.

Contribution, taxation, and voluntary charity: a chapter parameter

The old draft declared taxation abolished everywhere. That was ideology pretending to be architecture. The exitarian position is different: a chapter may levy contributions if and only if exit is free. A tax you can escape by walking out the door is a membership fee, and membership fees are legitimate. A tax backed by detention is extortion with a flag. The Grundgesetz does not ban taxation; it bans captive taxation, and then lets the chapters compete. Chapters that fund generous services through contributions will attract members; chapters that extract will watch their graphs and their populations drain. Exit is the tax revolt that never needs a barricade.

Within low-contribution chapters, voluntary charity covers disability, illness, emergency, and poverty, with the web of trust making the performance of charitable organizations visible and the guardrails making their books open. Donors choose, verify, and withdraw; a corrupt charity starves in public.

Property regimes as chapter law

The old draft assumed total private property. The framework does not. Property regime is one of the axes on which chapters legitimately differ: absolute private property, commons, cooperatives, mixed condominiums of owners with shares. What the federation requires is only that the regime be stated, consented to at entry, and escapable at will. Security of chapter spaces follows chapter rules: owners’ agents, communal watch, robotic surveillance; whoever manages a space has direct incentive to keep it safe, under penalty of clients, members, and reputation. There is no enforcement void, because there is no unowned, ungoverned space inside a chapter; and between chapters, treaty law holds.

Conflicts between arbitration agencies

Agencies that fail to agree cannot act outside their legitimacy without consequence. An agency abusing power exposes itself to the same crowdfunding mechanism designed for organized violence; the community can finance corrective action against it. Power without perceived legitimacy becomes a target, not an advantage.

Total refusal of arbitration

Refusing all arbitration is not punished directly; it is priced. A person recognizing no dispute-resolution mechanism is uninsurable and untrusted; the result is progressive economic isolation, not as imposed punishment but as the natural consequence of declining the trust protocol society runs on. Exit from arbitration is respected like every other exit. It is simply expensive, as it should be.

Civil law

Most real disputes are contractual: unpaid bills, defective deliveries, divergent readings of agreements. The same competing-arbitrator market handles them, with the same reputational dynamics and the same disclosure guardrails. Arbitrators who rule unjustly lose clients; arbitrators who rule well gain them; and because case outcomes are published, “unjustly” is a verifiable market fact rather than a vibe.

Evidence

Private investigators cooperate with insurers and security agencies in an integrated evidence ecosystem. Blockchain timestamping fixes the moment of collection; backdating or modification becomes detectable. Chapter courts and arbitrators set their own evidentiary standards, which become yet another axis of chapter competition: members who fear loose standards leave for stricter chapters.

Protection of vulnerable subjects

Those who cannot maintain their own graph position (the demented elderly, the severely disabled) are protected by the general mechanism: exploitation of the defenseless is among the most reliably punished behaviors in any reputational community, because observers punish it even when victims cannot. Guardrails reinforce this: institutions serving vulnerable people operate under mandatory disclosure, and their auditors are their competitors.

Witness protection

Testifying against a powerful node carries retaliation risk. Mitigations: agencies and insurers profit by protecting witnesses (demonstrated protection wins clients and cases), and anonymous testimony over Tor-class networks separates the validity of evidence from the identity of its source.

Intellectual property and attribution

Patents and copyright in their traditional form find no home in the reference model; in practice they serve incumbents blocking innovation more than inventors seeking reward. What must be guaranteed is attribution. Presenting another’s idea as your own is a lie, and lies have concrete reputational cost; the plagiarist is flagged and the community recognizes the true author. Recognized authorship converts to competitive advantage without legal monopoly: the originator sells into a market that knows exactly whose idea it is. Chapters may diverge here too; some will honor limited IP terms by internal law. The federal layer guarantees only attribution, because attribution is information, and information is what the graph runs on.

Healthcare

Healthcare in the reference model is private and charitable: direct payment, competing health insurers, and charity under disclosure. Without pharmaceutical monopoly, insurers are incentivized to fund cures and prevention, because a healthy client pays premiums longer and costs less; the insurer’s profit and the patient’s lifespan point the same direction. Insurers also select and monitor physicians, because every medical error is a compensation event. Competition on medical quality replaces bureaucratic drift. Chapters with communal health systems operate under the same invariants: disclose outcomes, allow exit.

Counter-insurance and capture signals

Anyone may hold insurance against abuse by dominant agencies. Its premium is a market signal: when capture risk rises, the premium rises, and a risk that state systems hide until collapse becomes visible, priced, and tradable in real time. Competition remains the primary defense; an agency captured by the wealthy loses its mass clientele, which outweighs any plutocrat’s portfolio. Serving the few at the expense of the many is a losing strategy in a transparent market, which is why the guardrails exist: they make the market transparent enough for that sentence to be true.

Chapters and cultural diversity

The old draft noted that web-of-trust communities in Argentina, Italy, and Japan would develop radically different norms, and called this a fundamental property rather than a flaw. The federation formalizes that insight: the chapter is the unit of normative diversity.

Trust is not abstract; it is rooted in culture, custom, and a local sense of what is right. What earns attestation in one chapter earns flags in another. Tolerance thresholds, the weight of family versus individual, commercial practice, moral standards: all of it varies, and all of it should vary. A state imposing one legal code across an entire territory is not unity; it is forced averaging. The web of trust is neutral infrastructure; the chapters are the norms. Moving between chapters means adapting to local rules, exactly as moving between countries does today, with one difference: the rules were not imposed by a distant authority, they emerged from the people living under them, and the door you entered through remains unlocked behind you.

There is no universal concept of crime beyond the near-universal (murder, theft, fraud). For everything else, each chapter and each subgraph finds its own equilibrium. The Grundgesetz does not harmonize values. It guarantees the conditions under which different values can coexist without war: exit, boundary, neutral infrastructure, self-sovereign identity, and an accountability chain that spans carbon and silicon.

Post-scarcity perspective

AI and robotics drive production costs toward zero. This does not obsolete the framework; it simplifies it.

Costs that scale downward

Insurance administration, detention, investigation, surveillance: all automated, all near-free. The justice system scales down with the economy. With no patent monopoly, anyone can build their own robots and AIs; energy and materials remain the only real scarcities, and near-free production covers near-free premiums.

The economic motive for crime disappears

When scarcity vanishes, theft, robbery, and fraud lose their rational basis, while near-free distributed security collapses the odds of getting away with anything. Serious crime under these conditions is committed by those who cannot weigh consequences: mental illness, not calculation.

Prevention through mental healthcare

Mental healthcare is likewise near-free, and insurers are strongly incentivized to identify and treat risk early: a client who commits a serious crime is an enormous cost, while therapeutic AI and cheap medication are trivially cheap. Before detention and before exile, the system inserts prevention, because prevention is the profit-maximizing strategy. Symbiont therapists with published outcome data will outperform both silent human practitioners and silent machine ones; the guardrail logic applies to minds too.

Residual containment

For those who still reach serious crime after reputation, insurance pressure, and treatment, containment remains: automated detention at near-zero cost, removal from the community, and the oracle mechanism as last resort.

Property, value, and the finite planet

Scarcity does not vanish; it migrates to the physical: space, location, resources, energy, materials. Property value is set by demand, and demand for space persists for three reasons: location near one’s trusted subgraph (the web of trust itself creates geographic value; well-connected land is expensive land), natural resources, and pure desirability. Price discovery happens through voluntary exchange, likely in cryptocurrency. Property also functions as environmental protection: owners decide, the graph makes consequences visible, and insurers price long-term ecological damage because it is ultimately their cost.

From money to reputation as a medium of exchange

As production costs fall and the graph matures, money loses its everyday function in trusted contexts. If giving something costs you almost nothing and the graph shows the recipient contributes positively to the world, their attestation is payment enough; it strengthens your position and opens doors elsewhere. A technologically enabled gift economy emerges: those who give receive reputation; those with reputation receive. This is not a time bank; the exchange is qualitative, denominated in the recipient’s attestation and weighted by how much each observer trusts that recipient’s judgment. Observer-relative all the way down.

Money does not disappear; it specializes. Of its three historical functions (medium of exchange, store of value, speculation), reputation absorbs the first, post-scarcity erodes the second, and the third flourishes: prediction markets, the same Truthcoin/Augur infrastructure that powers the oracles, become the primary domain of cryptocurrency. People stake tokens on discoveries, outcomes, agency performance, insurer recidivism rates. The prices aggregate what people actually believe, which is information the whole society can use.

This is related to futarchy, Robin Hanson’s proposal (2000) that prediction markets aggregate dispersed information better than any deliberative body. Here it applies without its original governmental framing: no representatives, no state, but markets as an informational lens on the consequences of choices, integrated with the graph. The web of trust provides retrospective reputation (how an actor behaved, from your vantage point); prediction markets add the prospective dimension (how the market expects them to behave). An agency may be excellent in your subgraph while the market prices a high probability of failure; that is an additional signal, information rather than imposition.

Trading fees from these markets fund the oracles, which are the framework’s enforcement of last resort. The circle closes: daily life runs on reputation, cryptocurrency specializes in prediction and residual low-trust exchange, and prediction markets sustain the oracles that guarantee the system.

Comparison with the current system

This framework is not perfect. Judge it against what exists, not against an ideal.

Dimension	Current state system	Exitarian framework
Judicial accountability	Nearly nonexistent	Reputational, through web of trust and competition
Right of exit	Effectively none; exit costs everything	Constitutional invariant; mechanism number one
Legal pluralism	One code per territory, imposed	Chapter constitutions under a thin Grundgesetz
Protection from judicial error	Partial (appeals)	Structural (incentives not to act under uncertainty)
Information asymmetry	Regulators captured, records sealed	Mandatory disclosure, adversarial audit, oracle verification
Enforcement vs organized violence	Selective, corruptible, geopolitical	Crowdfunded, oracle-verified
Extrajudicial executions	Exist, unaccountable	Require distributed consensus with skin in the game; system fails toward inaction
Corruption	Systemic, often consequence-free	Disincentivized by irreversible identity loss
Proportionality	Varies by jurisdiction	Emergent per chapter; disciplined by exit
Civil law	Slow, expensive, unpredictable	Competing arbitrators with published records
Intellectual property	Patent thickets, cartels	Reputational attribution; chapter divergence allowed
Decision quality	Individual judge, no standard tools	Symbiont-assisted; accountability chain spans carbon and silicon
Evidence integrity	State chain of custody, manipulable	Cryptographic timestamps, competing investigators
Prevention	Separate from justice	Integrated via insurer incentives incl. mental health
Vulnerable protection	Underfunded social services	Reputational pressure plus mandatory institutional disclosure
Taxation	Captive levy, no consent	Chapter contributions disciplined by free exit
Healthcare	Bureaucratic, patent-inflated	Insurers incentivized to cure and prevent; competition on quality

Transition and implementation

Parallel and legal strategy

The transition happens through parallel substitution, not revolution. The framework begins as a voluntary opt-in layer inside existing law. Private arbitration is already recognized nearly everywhere for commercial disputes; agencies and insurers incorporate as ordinary companies; the web of trust ships as an open protocol with no center to attack. Chapters form first as communities of practice, then as contractual networks, then as jurisdictions in everything but recognition. Only when critical mass makes the system cheaper and better than the state do individuals exit state services, lawfully, one function at a time.

Under legal repression, the protocol is built to persist: unseizable identities, nodes on censorship-resistant networks, contracts on resistant chains.

Development phases

1. Initial phase: the web of trust starts with small stakes; favors, sales, service exchanges. Build the graph; mature the understanding.
2. Intermediate phase: as the graph densifies, reputation acquires real value; arbitration agencies emerge for commercial disputes; first chapters declare constitutions under the Grundgesetz.
3. Advanced phase: with critical mass, the system absorbs increasingly critical functions, up to the oracle layer for extreme cases.

The full architecture may take decades. Individual components emerge independently and reinforce each other as they reach critical mass. The graph can be built before the oracles exist. The chapters can be declared before the graph is dense. Each piece that lands makes the next one easier.

Notes

This document supersedes the earlier libertarian draft. The mechanisms survived contact with a better frame; the frame is what changed. Libertarianism describes one chapter, perhaps several. Exitarianism describes the federation: the conviction that no system, however well-designed, earns the right to hold anyone, and that the only justice worth building is the kind you can walk away from.

The door is the doctrine. Everything else is engineering.