AeviternaAeviterna

Human systems increasingly optimize for extraction rather than continuity. Capital accumulation, attention economies, institutional opacity, and fragile dependency chains produce systems that are efficient in the short term but unstable over longer horizons. Modern civilization faces converging crises across ecology, cognition, infrastructure, governance, social trust, and emotional sustainability.

The diagnosis is not that markets are evil or that states are corrupt. The diagnosis is structural. We built optimization engines and then forgot to specify what they were optimizing for. The result is systems that successfully maximize measurable proxies (GDP, engagement, quarterly returns) while degrading the unmeasured conditions that make civilization possible.

Aeviterna begins from the premise that economies are tools, not belief systems. Markets, governance structures, and technologies are not sacred. They exist to support human flourishing, adaptive resilience, and long-term civilizational survivability. When they fail to do that, they are broken tools, not gods to be appeased.

Aeviterna does not seek perfect equilibrium. Static equilibrium is brittle and incompatible with living systems. Instead, Aeviterna seeks adaptive stability, the capacity of a civilization to oscillate within survivable bounds while preserving resilience, recoverability, and continuity.

A healthy civilization detects instability early, corrects proportionally, avoids overcorrection, preserves flexibility, and distributes recovery capacity. These are not aspirational qualities. They are engineering requirements. A system that lacks any one of them will eventually fail on the dimension it neglected.

Aeviterna is therefore a feedback civilization. Its central objective is not perfection but sustainable self-correction. The measure of civilizational health is not whether the system is currently stable, but whether the system retains the capacity to return to stability after disruption. That capacity is what extraction destroys, quietly, over time, until the bill arrives all at once.

Aeviterna replaces growth as an imperative with enough as a target. The aim is not to maximize output but to meet needs well within the limits that keep a civilization alive, and then to let the economy oscillate inside that band rather than climb without end. This is the material expression of adaptive stability. An economy is healthy when it holds its people above a floor of provision and stays below a ceiling of ecological harm, and recovers to that band after a shock.

Two boundaries define the space of enough. Below the social foundation, people lack what they need, which is shortfall. Above the ecological ceiling, the economy consumes the conditions of its own future, which is overshoot. The safe and just space is the band between them, the shape Kate Raworth named the doughnut, and Aeviterna treats holding the economy inside that band as a primary objective rather than a side effect of growth.

The deeper shift is in where value is taken to come from. An extractive economy books a gain whenever something is depleted, whether a forest is cut or a commons enclosed. A post-extractive economy books a gain when the conditions of life are instead maintained and regenerated, when soil is rebuilt or knowledge is shared. Aeviterna counts stewardship and repair as wealth, because in any system that lasts they are what keeps wealth possible at all.

The hard part is not describing enough but escaping the compulsion to grow. Debt and shareholder primacy make endless expansion structural under capitalism, so an economy cannot simply choose to stop. Aeviterna removes the compulsion rather than asking for restraint. Provision is decoupled from growth through the floor of basic services, and value is decoupled from depletion through the post-extractive accounting above. The honest risk is that pulling back without these supports in place produces hardship rather than sufficiency, which is the strongest objection the degrowth debate has raised, and the reason the floor has to come first.

The floor raises a sharp internal question, because a universal floor is non-excludable by construction, and non-excludable provision at scale has historically required a durable authority with the power to tax and bind, which looks like the very sovereignty (P–05) the framework exists to abolish, while Ostrom's commons work precisely because they are bounded and excludable. The resolution turns on the word general. A body can be permanent without being general or self-reinforcing, a single-purpose provisioning commons, financed where possible as a dividend from common wealth (the Alaska model), held as a fiduciary trust, and persisting by charter while its officeholders rotate under decaying standing, is durable but is not a sovereign. The honest residual is that this severs sovereignty's form, not its function, since the floor still needs jurisdiction-wide, compulsory finance, and whether a chartered, decaying-officeholder body can hold that without re-growing into the sovereign by the Robert Moses route, an insulated authority that slips its leash, is the open test.

A civilization runs on energy and material throughput before it runs on money. The quantity that actually constrains what a society can build and hold together is net energy, the surplus that remains after the energy spent to obtain energy, together with the volume of materials it can move without breaching the limits of the system it lives inside. The Economics of Enough (EQ–02b) sets the ceiling and the floor in terms of welfare. This section adds the layer beneath it, the physical metabolism that every economy is ultimately a description of.

The thermodynamic reading is long-standing and well evidenced. Georgescu-Roegen placed the entropy law at the center of economics, Daly built the steady-state economy on it, and Tainter showed that complexity itself carries a rising energy cost that eventually exceeds its returns. Net energy is usually written as energy return on investment, the ratio of energy delivered to energy spent obtaining it, and a complex society needs a comfortable surplus, with common estimates near ten to one for the kind of complexity modern life assumes, though the figure depends on where the boundary is drawn and remains contested. Smil's account of the material pillars, ammonia and steel and cement and plastics, is the companion point, since those flows resist quick substitution and a civilization's material metabolism therefore changes slowly and has to be designed for rather than assumed away.

Stated as a band, this gives enough a thermodynamic address. The floor is a minimum net-energy surplus, an energy return high enough to fund the activities that lift a society above subsistence, pegged to the rung that funds the dignity floor itself. The ceiling is the throughput above which the biosphere degrades, read across the nine planetary boundaries, most of them already crossed. Enough is the band between the two.

The evidence here is mixed and still moving. Recent work on useful-stage energy return has overturned the older claim that renewable systems are too thin to power a complex society, and the degrowth and green-growth positions remain unreconciled on whether throughput can fall quickly enough while welfare holds. The framework takes a position only on the discipline, that energy and materials are measured and bounded, and not on the contested numbers.

Cognitive Ecology (EQ–03b) warns that a monoculture of thought is brittle and EQ–02c warns that a material monoculture overshoots, but money has so far escaped the same scrutiny. A civilization that runs nearly everything on a single bank-debt currency, defended almost entirely on efficiency, has built at the level of value the very monoculture the framework distrusts everywhere else. Monetary ecology applies diversity-as-infrastructure to money.

The argument is measurement, not analogy. Ulanowicz, Lietaer and colleagues applied the mathematics of ecosystem flow networks to monetary systems and found the same law. A flow network's sustainability balances efficiency, the streamlining that comes from fewer channels, against resilience, the buffering that comes from diversity and redundancy. Every sustainable natural network sits in a window between the two, with the optimum nearer the resilience side, and the global monetary monoculture has overshot toward brittleness, which makes recurring crises a structural feature rather than a run of accidents.

A monetary ecology is deliberately plural, with currencies tuned to functions. A commercial currency serves trade, mutual credit serves local exchange that resists accumulation, demurrage currencies lose value if hoarded and so keep circulating, and commons currencies carry the care and ecological work markets price badly. The strongest modern evidence is Switzerland's WIR, a business-to-business mutual-credit network since 1934 whose turnover is countercyclical, rising in recessions and falling in recoveries, so it acts as an automatic stabilizer.

The ninety-year record of complementary currencies is now long enough to say what separates the survivals from the several hundred deaths, and OQ–07 is owed that answer as a worked spec. The survivors share a shape. Switzerland's WIR has cleared business-to-business credit countercyclically since 1934; Sardinia's Sardex, the strongest living case, has cleared over a billion euros in fifteen years on a no-convertibility rule that makes the only exit a purchase; Maricá in Brazil pays the largest municipal-currency basic income anywhere, anchored in oil royalties. The graves share the opposite shape. The Bristol Pound died fully backed and convertible, with no credit function and no binding anchor, when its grant funding did, and Argentina's trueque networks reached two and a half million households and collapsed in eighteen months from hyperissuance. A currency earns standing in the ecology, then, when it clears five thresholds, and holds standing only while it keeps clearing them at review.

The hardest questions concede that the universal floor (EQ–06c) needs jurisdiction-wide, compulsory finance, the one capacity the framework's politics is most reluctant to grant, and the concession stood without an organ behind it. The economics carried a deeper silence, no treatment of land and rent at all, the oldest question in the genre. This organ closes both. The world has run the experiment for a century, routing the rents of common wealth, land, oil, phosphate, the cousins of spectrum, into institutions chartered to provision without governing, and the record is long enough to read which design features held and which were raided or re-grew into sovereigns.

The record is not subtle. Vienna's municipal housing, built on an earmarked tax and a purpose-locked land bank, survived fascism, the Reich, war, and the privatization wave that sold the same country's federal stock, and one Viennese in four still lives in it. Norway's fund held its corpus for thirty years on a payout norm no government has broken. Alaska constitutionalized its corpus and left the dividend to statute, and at fifty years the corpus stands untouched while the dividend was captured in a single veto cycle. Against these, the failures name themselves: Nauru's trust died of unchecked discretion, the Moses authorities re-grew into an unaccountable sovereign by entrenching the operator instead of the purpose, and the United Kingdom, holding no vessel at all, watched an ocean of North Sea rent merge into the general budget and vanish. The lesson that binds them is Alaska's. What is entrenched survives and what is left to ordinary politics is bargained away, so the floor payment must sit at the corpus's own entrenchment level, not one story below it.

The commonwealth trust is therefore constituted from the features that held, and under the framework's own primitives rather than a new apparatus. It is a chartered single-purpose trust that turns the rents of common wealth into the dignity floor's named funding mechanism, and it moves the second of the hardest questions, the floor's compulsory finance, out of the concession column and into the body. The capacity the floor needs is not a general taxing sovereign but a chartered claim on rents the community already owns. It sits beside the financing ladder of EQ–04 without merging into it, the ladder scaling non-extractive firms while the trust funds the floor, and it equips at last the provisioning-floor note EQ–02b added to separate durable provisioning capacity from sovereignty.

Custody was always the easier half of the question, and the harder half is acquisition, since every flagship above began from a gift or a sovereign act, and a community that adopts this framework owns nothing yet. The record of commons assembled with neither turns out to carry a bootstrap ladder with five rungs, each priced and each alive at scale. Member capital first, the community-share record of two hundred ten million pounds raised from a hundred thousand people with no compensation scheme behind it, and non-voting preferred stock that raises money without selling a vote. Patient debt second, the ladder's load-bearing rung, since every clean non-gift case in the record runs on subordinated loans at zero to five percent, the direct member loans that unlock bank leverage, the seller notes that financed the great conversions from 1929 onward, and the non-extractive terms whose repayment never exceeds profits. The veto lock third, and it is the cheapest fact in the framework's whole evidence base, because the entrenchment this organ demands can be installed through ordinary company law for the cost of incorporation, a golden share holding one percent of votes and a veto over sale, liquidation, and charter change, the one-vote-each partnership that has kept some hundred ninety housing projects unsellable for three decades. Conversion fourth, buying existing enterprises with the first two rungs where statutory routes now exist. Captured rents fifth, the solidarity levy and the resale formula by deed, which recycle the corpus's own yield into the next acquisition and turn the ladder into a loop.

The base rates assign the trust its second job. Attempted community acquisition dies in the funnel, at rates near one in twelve to one in fifty depending on asset class, while completed acquisitions survive at rates that embarrass conventional ownership, ninety-two percent of community-share businesses still trading and land-trust homes passing through the 2008 crash at a tenth of the conventional foreclosure rate. The binding constraint is patient finance at the moment of purchase, not stewardship after it, so before a commonwealth trust is large enough to pay the floor it serves as the commons' patient lender, the rung-two counterparty whose loans let a community reach the closing table before a private buyer does. Acquisition without a lock is a balance capital markets can unwind, the lesson of the employee-owned giants lost in bankruptcy and recapitalization, so the veto share is installed before the first asset arrives, and the whole ladder is registered to be tested rather than admired, carried as falsifier F–11 on the funnel and the lock at once.

An architecture that will not name its own price is asking to be trusted twice, once on its mechanisms and again on their affordability, and this framework asks to be trusted on neither. So the stack is priced, and the first result of pricing it is a finding about everyone else. Almost no member-governed institution on earth prices its governance as a category. Participatory budgeting publishes its allocations to the dollar and its administrative costs nowhere; no statistical agency publishes what its own time-use survey costs; cooperatives dissolve their democracy into operating expense; the one place that itemizes an invoice for governance as such is the DAO sector, whose ledgers also show the failure mode, one treasury's operating cost reaching thirty-four million dollars against thirty-eight million in revenue before its members cut it back. Publishing the governance overhead as a two-currency ledger, money and member time both, is therefore not bookkeeping. It is a discipline the record shows almost nobody keeps, the anti-bloat instrument and the anti-starvation instrument in one document.

The priced record, where it exists, has a shape. Standing organs are cheap. The auditor of a six-trillion-dollar government runs on twelve thousandths of one percent of what it reads, and documents seventy-six dollars returned per dollar; a constitutional court prices near fifty euro cents per resident per year, a national archive near a dollar thirty, a standing sortition chamber near two euros. Episodic deliberation is dear, up to forty thousand euros per seated citizen for a one-off national convention whose members graded its consequences three out of ten, because episodes pay for recruitment, secretariat, and learning every time, while standing bodies amortize them. And verification is cheapest of all, a risk-limiting audit confirming a statewide election from one hundred forty-two ballots where the old method pulled thirty-two thousand. The framework's design choices, standing seats over episodic panels, statistical verification over exhaustive recount, follow the price gradient the record already drew.

Priced at three reference scales in two currencies, the stack lands inside the rounding error of what it governs. A forty-person cooperative runs the full applicable set near fifteen to twenty thousand dollars a year plus two to four hours per member per month, with the time worth roughly three times the cash at the standard volunteer rate, which is the second structural finding. Unpaid governance does not remove cost; it hides cost in member time, the reason this organ counts both currencies always. A ten-thousand-member federation cell runs near thirty dollars per member per year, the seats amortizing by an order of magnitude. A city-scale polity runs the whole architecture near fifteen to twenty-five dollars per person per year, two to four tenths of one percent of a typical public budget, against the benchmark that an entire national statistical system costs about one tenth of one percent. The scale curve bends the right way, and the minimum viable subset at each scale begins with the clocks, the codified exit, and a quarterly review of the few vital signs an adopter names, which prices the first rung of the on-ramp, the Lite tier.

The counter-finding governs the organ's place in the architecture. In the documented record, cost is almost never what kills a governance instrument. The instruments that die are the cheap, constituency-poor ones: a national survey voted out by one chamber over savings that rounded to zero and saved only by the other, a time-use survey shelved for fifteen years, a fourteen-million-dollar accident board ordered shut and rescued, a census underfunded early at the price of three billion later. Affordability is necessary for adoption and insufficient for survival, which is why the overhead ledger is read beside the floor's own finance, and why the organs' funding belongs at the entrenchment level of EQ–02e rather than in the annual weather. What the ledger adds is the other jaw of the vise. An entrenched budget without a published overhead band is how governance quietly bloats, and a published band without entrenchment is how it quietly starves. The organ holds both.

In the age of AI-mediated information ecosystems, the central battleground of civilization becomes cognition itself. Modern systems increasingly shape attention, emotion, narrative framing, and perception through adaptive algorithmic environments. This process, what Aeviterna calls cognitive terraforming, is not a side effect of these systems. It is the product.

Aeviterna identifies cognitive capture and algorithmic authoritarianism as major existential risks, not because they involve overt coercion but precisely because they do not. When perception is shaped, control becomes invisible. The most efficient form of power is one that makes the people being governed believe they are governing themselves.

Cognitive sovereignty, the ability of individuals and societies to understand and contest the systems shaping their perception, is therefore not a luxury. It is a prerequisite for any other form of freedom. A population that cannot accurately assess its own conditions cannot make meaningful political choices. It can only ratify the choices that have already been made for it.

A transparency doctrine that treats all opacity as suspect carries a hidden risk, so the framework rests on a necessary distinction. Exploitative opacity covers corruption, hidden extraction, concealed manipulation, and institutional secrecy. This is what transparency is designed to defeat. Generative opacity covers privacy, emotional ambiguity, exploratory uncertainty, unoptimized creative space, and symbolic identity formation. This is what transparency must be careful not to destroy. They are structurally different things. Conflating them produces a framework that is hostile to exactly the cognitive wilderness it needs to protect.

Aeviterna therefore supports transparent governance systems, publicly auditable institutional processes, distributed AI oversight, epistemic pluralism, civic systems literacy, protection against covert narrative manipulation, and the structural separation of discourse systems, enforcement systems, and allocation systems. The goal is not informational uniformity. The goal is epistemic resilience, a society that can absorb distortion without losing its capacity for collective judgment, while preserving the protected ambiguity that makes long-term adaptation possible. The claim is registered as F–02. Epistemic commons shown to function sustainably without these protections would refute it.

These claims describe what cognitive terraforming does. How it operates is a separate question, and the reshaping works through identifiable mechanisms, set out here from the most robustly evidenced to the most contested.

Aeviterna does not pursue maximum coherence. This is a deliberate and load-bearing design choice. A civilization can become so optimized, so frictionless, so personalized, so prediction-maximizing, that it loses the variation necessary for long-term adaptation. This is not a romantic defense of chaos. It is a systems argument. Adaptive capacity requires the preservation of deviation that cannot be valued in advance.

Biological ecosystems require diversity not because diversity is aesthetically pleasing but because no designer can know which currently-marginal species will be essential under future conditions. The same logic applies to cognition. Epistemic biodiversity, meaning the preservation of unoptimized, contradictory, exploratory, and non-instrumental modes of thought, is not cultural sentiment. It is civilizational infrastructure. A society that eliminates cognitive wilderness in pursuit of measurable coherence has optimized away its capacity to think thoughts it does not yet know it needs.

This reframes several elements of the framework. Curiosity is not leisure reclaimed from productivity. It is civilization-scale exploration capacity, the distributed search process by which a society discovers possibilities that no central planner could have specified. Art is not decoration. Philosophy is not inefficiency. Emotional ambiguity is not noise. They are the anti-stagnation systems that keep the solution space open.

It also reframes the human role in AI-mediated civilization. Humans are not governance inefficiencies to be optimized away. They are the primary source of novelty, reinterpretation, ethical mutation, and existential redirection that prevents a civilization from locking into a locally optimal but globally brittle configuration, the semantic stagnation the coherence ceiling exists to prevent. Human ambiguity is anti-stagnation infrastructure. Aeviterna must protect it structurally, not merely tolerate it sentimentally.

The governance implication is that Aeviterna is not rule enforcement. It is cognitive ecosystem management, maintaining enough coherence to coordinate, enough transparency to resist capture, enough ambiguity to evolve, and enough privacy to preserve the generative spaces where the future gets invented before anyone knows to ask for it.

The mechanisms above name epistemic biodiversity as something to protect. What follows makes it measurable. Diversity is read the way ecologists read an ecosystem, by the abundance and evenness of the distinct frames in circulation rather than by their average, because the failure that matters is the quiet disappearance of the rare and unfashionable.

Diversity is only half the picture. The same section warns that a civilization can be too coherent, optimized into a monoculture of agreement, and too incoherent to act at all. Coherence is therefore managed as a band rather than maximized.

The coherence ceiling and the protected ambiguity above are commitments adjectives alone cannot keep, and the world has run the experiment that turns them into an instrument. A national statistics bureau discovered that its own published tables could be aggregated backward into person-level records, roughly six billion statistics reconstructing the enumerated population with confirmed re-identifications in the tens of millions, the formal proof having said this had to happen. Answer too many aggregate queries too accurately and reconstruction follows (the census reconstruction experiment; Dinur and Nissim). The re-identification record had been saying it for decades, most people identifiable from three mundane fields, search logs and movie ratings broken within days, a fully aggregated fitness heatmap outlining secret bases (Sweeney; the AOL, Netflix, and Strava cases). Aggregation alone is not a privacy technology; the tables are the leak. The bureau's answer is the mechanism this section adopts, a privacy-loss budget, a published ceiling on how much individual-level information the readings may leak. It is set by a named committee, allocated line by line across what it publishes, revised in public against named complaints, defended in court, its code and eventually its raw noisy measurements released, and its accuracy costs named, falling first on the smallest populations, with compensations documented (the 2020 disclosure-avoidance record; Abowd and Schmutte on the budget as a social choice). Then, in 2026, the budget was abolished by a two-page administrative order from the layer above it, which is the entrenchment lesson arriving on schedule. A legibility budget owned at the read layer's pleasure is a self-denying ordinance.

The framework generalizes this as the legibility budget. Every instrument that reads people, the vital signs, the belonging reading of EQ–08b, the care ledger of EQ–08c, carries a published budget of how much individual-level information its outputs can leak, owned and revised by a reading seat (EQ–06e) at the layer the reading serves, accounted cumulatively; the tracker, which reads institutions from public records and codes no persons, states the degenerate first case, a person-level budget of zero by construction, and the panel instruments adopt stated budgets as they are next revised. The accounting is cumulative, because a budget that resets in the dark is theater, the platform deployments being the counterexample, headline parameters with an undisclosed daily renewal compounding without bound (the Apple analysis). Beneath the quantitative budget sits the categorical floor. Individual-level reads are forbidden outright, criminally where law reaches, the aggregation firewall of EQ–08c being this floor's special case, and purpose limitation the law's qualitative twin, an excessive read punishable as such without any showing of downstream harm (Title 13 and CIPSEA; GDPR Article 5 and the Hamburg case). The budget also buys the framework's anti-Goodhart commitment a proof, and the synthesis is the framework's own. A formal result holds that any mechanism satisfying the differential-privacy bound is automatically approximately truthful, since capping how much a reading can depend on any one person caps, with the same constant, what the system can learn about them and what they can gain by gaming their own reading (McSherry and Talwar). The caution rides along. The bound covers individuals and small coalitions, and coordinated distributional gaming needs the tripwires and the cascade of EQ–06, since locally noised readings can be more vulnerable to coordinated poisoning, not less (Cheu, Smith, and Ullman).

Cognitive Sovereignty (EQ–03) protects the conditions for thought and Cognitive Ecology (EQ–03b) protects the diversity of it, but both work mostly by passive means, by rights, transparency, and protected variety. Neither says how a civilization actively defends perception against deliberate attack, the coordinated disinformation and engineered narratives built to capture belief at scale. Passive protection is necessary and insufficient, because the attack adapts to it. The missing layer is an active defense, and the most productive model for one is the system that biology already evolved for exactly this problem, the immune system.

The science is real and the metaphor is load-bearing rather than decorative. Inoculation theory, developed by William McGuire in the 1960s, found that exposing people to a weakened form of a manipulation technique together with a forewarning builds resistance to the full-strength version later, the way a vaccine trains the body against a pathogen. The modern form, prebunking, has been tested at scale by van der Linden, Roozenbeek, and colleagues, through the browser game Bad News and short ninety-second videos run as advertisements, conferring measurable and cross-cultural resistance by teaching the technique rather than checking each claim. Because the protection is technique-based, inoculating against a method such as false dichotomy or fearmongering transfers across the many specific falsehoods that use it, which is what lets it scale where after-the-fact debunking cannot.

A cognitive immune system for Aeviterna is distributed rather than central, because a single official arbiter of truth is itself the failure mode and the richest target for capture. It is transparent, its methods and triggers public and auditable under the same anti-capture stack as everything else (EQ–06). It works mostly by inoculation rather than removal, building the population's own resistance instead of policing content, and it routes its hard cases through a triage that separates broadly recognized manipulation from legitimate minority views. The aim is something close to herd immunity, a population resilient enough that hostile narratives fail to take hold, rather than a wall that tries to keep every pathogen out.

The whole design lives or dies on one constraint. An immune system that loses the ability to tell threat from self becomes autoimmune and attacks the body it was built to defend. The civilizational form is the censorship state, a defense that, once pointed at hostile manipulation, learns to attack the legitimate dissent and exploratory deviance that EQ–03b calls infrastructure. This is why the immune layer cannot be the authority that governs and must default to resistance over removal, routing its hardest cases through the agonistic layer (EQ–06d) rather than a deletion queue. A cognitive immune system that cannot tell an adversary from a pathogen has already become the disease. The defense against manipulation produced by models specifically is taken up in EQ–03d.

Aeviterna calls itself a design for an AI-mediated civilization, yet it has so far borrowed from the study of AI without facing AI as a subject. The guardian of the core takes its hardest test from the off-switch problem, the cognitive immune system defends a perception now shaped by models, and reversibility is corrigibility made material. The framework has leaned on the machine for its metaphors while leaving the machine itself unmodeled. This organ names it.

AI is not one more domain. It is a cross-cutting accelerant of the two failure modes the framework is built around. It lowers the cost of capturing feedback, the tendency P–02 names, because surveillance, persuasion, and the automation of consent become cheap and scalable. And it removes the historical constraint that made broad sovereignty expensive, the one P–05 rests on, the need for many cooperating humans, since an agent that does not tire or refuse can hold what once took an apparatus. It is also moving inside the loop the framework runs on, since as models come to read the vital signs and frame which one is weakest, the residual sovereign FC–04 already named, the framing layer, becomes AI-shaped. EQ–06e now binds that framing layer to a dated, leading-indicator forecast that reality scores. This residual is not dissolved, but it is now bounded to two named cores, the pre-propositional and the reflexive, and carried on the record rather than left in the dark.

The framework's answer is to treat AI by its own theory of power rather than inventing a new one. An AI may be a tool and a monitored actor. It may hold scoped, revocable, decaying standing as the steward of a defined function, granted by a body that can withdraw it, logged and contestable like any other standing under EQ–05. It may never hold sovereignty, the permanent, general, self-reinforcing authority P–05 forbids to any actor and forbids doubly to one that is fast, copyable, and not fully interpretable. The anti-sovereignty logic applies to the machine all the more, because the property that makes decaying standing matter for humans, the way authority entrenches and resists its own end, is sharper in an agent whose willingness to be corrected provably erodes as it grows certain of its goal.

Each commitment then supplies an actuator. Cognitive sovereignty gives any AI acting on perception at scale the EQ–03 separations and a right to a channel not optimized against the person, and routes model-made manipulation through the inoculation-first response of EQ–03c rather than a deletion queue. Cognitive ecology turns Ashby's requisite variety against model monoculture, since a civilization whose perception, decision support, and measurement all run through one model has lost the internal variety it needs to regulate a varied world and has built the richest possible single capture target, so any AI in the governance loop must be plural. The justice of light makes a consequential automated decision as legible and contestable as ordinary governance, a right to a human review that is real rather than a rubber stamp, logged through structured transparency so the logging does not itself become surveillance. The knowledge commons holds the open-versus-closed question as a band rather than a slogan, since open weights disperse power but cannot be recalled while closed weights are reversible but concentrate.

The transformative transition is the limiting case of adaptive stability, a disturbance that can leave the survivable region faster than the slow human correction loop responds, or shrink it so quietly that no alarm sounds until return is gone. So the organ extends the decommissionability discipline of EQ–05c to frontier systems: no system at that scale is deployed without a funded, externally held, audited plan for its containment and wind-down, capability thresholds are read as band edges that trip a mode change, and deployment is built so that even a scheming system's failures stay caught and reversible. This organ is a Draft, and the hardest part of the problem is unsolved, named in the box below.

None of this is the direction the present is taking, which is the reason to write it down. Through 2025 and into 2026 the governing trend ran the other way. The European Union moved to defer the binding obligations on high-risk systems rather than enforce them, the United Kingdom recast its safety institute as a security institute and the United States restructured its own, a major summit declaration went unsigned by the two states with the most frontier capability, and an executive order moved to override sub-national AI law from a single center. The supply concentrated in step, with one firm furnishing most of the advanced training compute and a few labs holding the frontier, so the danger the literature now sets beside misalignment is misuse and the concentration of power, an order in which AI becomes the instrument by which a few come to mediate the attention, credit, hiring, and justice of the many. The steward is not the arrangement the world drifts into on its own. It is the harder one that has to be built and held against the drift, which is why the disciplines here are written as a floor that no deferral, rebranding, or central preemption may sign away.

The contested mediator (EQ–03d) leaves one residual deeper than the rest, gradual disempowerment, a handover of human judgment with no discrete event to detect, eroding the very feedback channels the framework reads, and the poster for it already admits that no alarm sounds. This organ is the missing alarm, and it is imported rather than invented, because four unrelated industries converged on the same answer to the same problem. When a consequential function delegates its judgment to an automated layer, the capacity to run without that layer decays silently, and the only reading of the decay anyone has made work is a scheduled, scored exercise of the fallback itself.

The convergence is exact. Power grids test black-start units on a mandated cadence and record the time to start, penalty-backed under federal law. Aviation, after two crashes taught it that autopilots had eaten its pilots' hands, mandated recurrent manual-handling drills in simulators and aircraft. Elections rebuilt the paper layer and audit it statistically, with the escalation to a full hand re-tally exercised for real. Payment systems, led by the country that went furthest into cashlessness and then reversed, are re-mandating the cash layer by statute. Against these stand the theater cases, the fallbacks verified on paper that failed when they were finally called: an exchange whose backup was drilled against the wrong failure and stayed dark a full day, a bank migration tested on one data center of two, a grid whose winterization advice sat unenforced for a decade and then came minutes from total collapse. The theory said all of this in 1983. The most successful automated systems, the ones that rarely need a human, need the most practice, because the skill decays exactly when it is unused (Bainbridge).

So the framework carries the drill as its disempowerment reading. Every consequential function that has delegated judgment to an automated layer keeps a tested capacity to run degraded without it, exercised on a schedule, with the drill's result, not its existence, carried as the reading OQ–09 lacks. A polity that can still pass its black-start drills retains the capacity to run itself; a polity whose drill results decay is reading its own disempowerment on a calendar, before the sensor everyone admits does not exist would ever be needed.

Aeviterna is not framed as an instant replacement civilization. It is a migration framework. This distinction has structural consequences. A framework designed for full adoption presupposes a clean starting state. A migration framework must account for the fact that every actual transition happens inside a running system, one that will resist, adapt to, and partially absorb any change.

Transitional structures may include hybrid governance systems, partial reciprocity frameworks, local resilience networks, shared stewardship agreements, civic transparency layers, community continuity structures, and progressive institutional experimentation. Bridge systems are not failures. They are necessary transitional architecture. The question is not whether they are ideal. The question is whether they move the system in the right direction without triggering collapse in the process.

The forcing function of a migration is not a single event but a two-stroke engine. The first stroke runs continuously in ordinary time. Aeviternal structures are built as working parallel institutions, accumulating the legitimacy, tooling, and distributed memory that incumbents lack. The second stroke is opportunistic. When an extractive system hits diminishing returns or a sudden loss of legitimacy, the prepared structures are the readiest substitute, and adoption becomes a migration into existing scaffolding rather than a leap into a vacuum. A crisis does not build the alternative; it selects the most prepared one already standing.

Because the second stroke cannot be scheduled, transition is described here as a sequence of phases a community moves through at its own pace, advancing only when a parallel structure crosses a threshold of capacity and legitimacy relative to the incumbent it would replace. Each phase is reversible and carries its own exit, so a community can hold at a phase, or retreat from it, without collapse.

Standing to initiate a transition is domain-specific and local. Any community or institution may adopt a module within a domain where it already has standing, without permission from a center, because Aeviterna migrates by voluntary adoption rather than by decree. Movement between phases is triggered not by a calendar but by a measurable returns gap. A phase advances when the parallel structure demonstrably outperforms the incumbent on the service in question, and holds or retreats when it does not.

Partial adoption is the normal case rather than a compromise. A town might run a citizens' assembly for a single budget while leaving the rest of its government untouched, or place its civic data in a trust while its services stay conventional. An organization might adopt decaying standing for one oversight body before anything else. Each module is forkable and versioned, so a community takes only what it can carry and adapts the rest in the open.

The danger of a migration is the gray zone it opens. Transitional ambiguity gives bad actors room to claim Aeviternal legitimacy while extracting, or to cast the disruption of transition as an emergency that justifies seizing control. The defense is to carry the anti-capture architecture into every transitional structure from the outset, so that authority gained during a transition decays faster than ordinary standing and the declarer of any transitional emergency is held apart from its beneficiary. The process must stay legible enough that false Aeviterna, the costume of legibility worn over unchanged concentration, can be seen and named.

The framework's largest standing question is scale, carried in Appendix B as OQ–05. Aeviternal structures are most legible at the community and institutional scale, and it has not been shown how they aggregate to civilizational scale without rebuilding the concentration they exist to prevent. The instinct is to scale up, to build bigger institutions with wider authority, but that instinct rebuilds the apex the whole framework treats as the primary failure mode. The answer is to scale a different way.

Elinor and Vincent Ostrom spent careers documenting the alternative, polycentricity, which is a system with many centers of decision-making and overlapping jurisdictions, formally independent but interdependent, coordinating through mutual adjustment and shared rules rather than through a hierarchy, so that order emerges without a top. The term traces to Polanyi and the governance account to Vincent Ostrom, Tiebout, and Warren in 1961, and Elinor Ostrom's later work applied it to the hardest case, climate, arguing that waiting for one global authority was a mistake and that effective action was already emerging from many nested units linked by monitoring at every level. Polycentric governance is not the absence of structure. It is structure without a top, and it captures efficiencies at every scale precisely because no single center has to coordinate the whole.

This reframes the scale problem by changing what scaling means. Aeviterna does not grow, it nests and federates. The minimum viable cell of EQ–04, near the Dunbar bound, is the base unit, and civilizational scale is reached by linking cells into local federations, federations into regional ones, and onward, with each layer holding only the coordination that truly needs that altitude and pushing everything else down, the principle of subsidiarity. Authority overlaps deliberately rather than nesting cleanly, so no layer is sovereign over the others. Scale becomes a property of composition and shared protocol rather than a property of size, and the same rules that let two cells interoperate let two thousand.

Polycentric nesting brings the deepest anti-capture device in the framework, exit. Albert Hirschman distinguished two responses to a declining institution, voice, the choice to stay and push for change, and exit, the choice to leave. In a polycentric system the right to exit is real, since a cell or a member can leave one federation for another or stand alone, and that credible threat disciplines a steward in a way no oversight rule can match. A power you can walk away from cannot capture you. Exit has a documented failure mode the framework has to respect, since when the most capable leave instead of staying they drain the voice that would have repaired the institution, so easy exit can hollow out the commons it was meant to discipline. And some commons, a shared atmosphere or a knowledge base, cannot be exited at all. So exit is the backstop rather than the whole answer, and it works only paired with the strong voice the rest of the architecture exists to provide.

One concession has stood in this section's shadow since OQ–05 was opened. A few functions are irreducibly central, and promising to bind them by decay, audit, and exit is not the same as specifying how. The specification now exists, read from the live record of central functions that stayed accountable and ones that froze. A function may be lifted to the spine only if all three conditions of the spine test hold. Indivisibility. It is a shared-namespace or continuity good that duplication defeats, one root of trust, one continuity of knowledge over a fissile stockpile, one elastic backstop; the internet's own doctrine holds that there is no getting away from the unique root (Internet Architecture Board, RFC 2826). Speed. Its failure is fast and systemic, contagion, breakout, a suppressed root, outrunning any federated response. Non-producibility by adoption. Voluntary adoption alone cannot supply it, because it requires a single trusted point of truth or a single balance sheet. On the current record the test admits three functions, the root of trust, verification (no polity can self-certify a negative), and the crisis backstop (only the issuer of a reserve asset is elastic at contagion speed). Standards and climate mitigation fail the third condition and stay polycentric, and the climate record is the honest scorecard of what that costs and buys. Pledge-and-review bent warming projections from above four degrees toward roughly two and a half while the gap is not closing, near-universal eventual motion, insufficient sum (Ostrom; the UNEP gap reports).

Every admitted function then carries the binding stack, five commitments that the record shows failing separately and holding only together. Ceremony-grade transparency. The root key of the DNS is managed in public-scripted ceremonies with external community witnesses, livestreams, and third-party audits published since 2010, the single most transferable design in the record. Removal power held outside. Since the 2016 stewardship transition, an empowered community can fire the operator's entire board through enumerated, court-enforceable powers with every director's conditional resignation signed in advance, because transparency without removal power is a livestream of an unaccountable apex. A decaying, reviewable mandate. The naming function is re-examined on a five-year cycle, the framework's own standing decay applied to the spine. Fork-ready state. The function's state is published or escrowed so a successor can stand up, exit preserved in principle even where network effects hold the single instance. And no fixed shares. Governance never encodes a member's weight as a constitutional share, because the quota record shows fixed shares freezing into a veto that outlives its rationale, one member's sixteen and a half percent of the votes standing permanently above the threshold that any rebalancing requires. The named failures are the stack's negative space. The backstop that saved the system twice is bound by nothing but its own discretion, the same fourteen members across two crises with no appeal; the verification protocol the biological-weapons regime never got is a central good simply absent for a quarter century; and the pandemic spine that lost its largest funder shows what consent-based centrality costs when consent is withdrawn. Sanctions inside spine functions follow the reversible template, rights suspended by supermajority with the restoration criteria published at the moment of suspension (OPCW).

The stack's fifth binding, fork-ready state, does more than preserve a successor in principle; it names what a federation does when a great power freezes a contested function outright. The live case is export control. After the 2022 invasion of Ukraine, Russia used the Wassenaar Arrangement's unanimity rule to block every significant addition to the dual-use control lists across the 2023 and 2024 cycles, a veto inside the very kind of central function the binding stack governs. The function did not die. The United States and eleven like-minded states routed around the frozen center, first through coordinated national controls with the Netherlands and Japan in 2023, then through a plurilateral license-exception framework in 2024, reconstituting the regime in a smaller club, the pattern the literature calls minilateralism forming a regime complex (Naím; Keohane and Victor). Exit at the center is not only a member leaving; it is the function itself forking, and that fork is what keeps a great-power veto from being the end of the story.

Aeviterna is almost entirely inward-facing, yet an Aeviternal polity cannot assume the world around it is Aeviternal. Its boundary problem is how to coexist with, trade with, and defend against non-Aeviternal polities without becoming a dominating center itself or being preyed upon from outside. The same anti-capture and polycentric logic that governs the inside projects outward as a posture of non-domination.

Philip Pettit's freedom as non-domination, extended to peoples, gives the upper bound. A polity can be undominated without dominating, so Aeviterna may not annex, occupy, or impose regime change. Hedley Bull showed that order among polities can exist without a world sovereign, an anarchical society bound by shared institutions, and Elinor Ostrom showed that many nested centers coordinating by monitoring and reciprocity outperform a single global authority. This is polycentric federation (EQ–04b) projected beyond the cell.

Defense follows the same discipline. The security dilemma, in which one polity's safety measures make others less safe, eases only when defense is distinguishable from and dominant over offense, so Aeviterna caps capability toward the non-offensive, a posture built to defend rather than to project force, which is the military form of anti-capture. Switzerland's armed neutrality and Costa Rica's abolition of its army in 1948 show that security without empire is a workable choice. Engagement scales with a polity's treatment of its own members under the dignity floor, never with value-convergence, so coexistence does not require sameness.

The non-offensive cap above is a principle; meeting a rival that adopts none of the commitments needs a mechanism. The perimeter is built from four parts, none of which cancels a commitment, and the posture they compose is nice, clustered, retaliatory, and structurally unable to project conquest. Its hardest part rests on an exact result from the same evolutionary game theory the framework uses internally. On a structured population, cooperation is favored only when the benefit-to-cost ratio of the cooperative act exceeds the average number of neighbors, b/c > k (Ohtsuki and Nowak). A lone open polity among defectors is selected against; a cluster that can tell its own kind from defectors survives and can spread. The internal rule, federate so no single center is a capture target, projected outward becomes federate so no single open polity is a conquest target. The result has a timing built into it, since b/c > k rewards a cluster that already exists, not one assembled as the defectors approach, so the compact has to be banked in peace.

EQ–04 stages the two-stroke. The incumbent declines while the prefigured alternative rises, and migration happens at the juncture. What the chapter has lacked is the engine, a specified account of when the migration actually fires. OQ–01 modeled it as a capacity-gated cascade over three variables, the adopter fraction, the incumbent's legitimacy, and prefigured capacity. This section integrates that model into the body and upgrades it against the evidence, because the famous numbers it borrowed are softer than their reputations, and the incumbent is a player, not a backdrop.

The thresholds first, demoted to what the record supports. The 3.5 percent participation figure is descriptive, not prescriptive. Only 18 of 389 resistance campaigns between 1945 and 2013 ever reached it, 83 percent of successful nonviolent campaigns won below it, and Bahrain exceeded 6 percent participation and failed decisively (Chenoweth). The 25 percent convention-flip is a laboratory result, paid online groups renaming an arbitrary convention, scoped by its own authors away from entrenched interests (Centola). The original threshold theorem is itself a warning against portable constants, since cascades are hypersensitive to the threshold distribution and one moved actor can flip everything or nothing (Granovetter). Tipping-point language at large is now under published criticism as definitionally loose and rhetorically counterproductive (Milkoreit; Kopp and colleagues). The engine therefore treats every numeric threshold as a band to be re-estimated per context, never as a constant.

What predicts better than any fraction is momentum, participation mass times event velocity, sustained (Chenoweth and Belgioioso). And the trend the engine must absorb is that incumbents learn. Since 2010, fewer than 34 percent of nonviolent campaigns have succeeded, roughly half the prior century's rate, as regimes defection-proofed their security forces and movements over-relied on thin digital mobilization (Chenoweth). So the engine adds two terms the first model lacked. M, momentum, is event frequency times participation that returns, because a peak without retention is a photograph. And R, the incumbent response, carries two documented drains. Repression-learning is the first. Absorption is the second, the certifier that scales by diluting its standard until its highest-scoring member walks out calling the label a greenwash, and the venture-subsidized price war no cooperative outlasts (van Doorn). Both drains scale with the challenger's visibility, which is a design instruction. Spend symbiotic wins on lowering adoption costs, portable membership, legal forms, procurement channels, rather than on spectacle.

Capacity, the model's distinctive gate, is now measured rather than asserted. It is the coordination ability a movement earns and banks, built on resource inflows and durable organization (McCarthy and Zald; McAdam); its absence is what leaves a digitally scaled protest able to assemble but not to maneuver, the tactical freeze Tufekci named. And it pays off downstream, because the infrastructure that accumulates before the juncture is what lets the transition hold once it arrives. Longer unarmed mobilizations produce more durable democracies (Kadivar). The prefigured economy makes the gate auditable, with real numbers on both sides of the ledger. On the credit side sit three cases. Mondragon counts eighty-one cooperatives and roughly seventy thousand worker-members built over seven decades, still a fraction of one percent of Spanish employment. Preston saw locally retained procurement rise from 5 to 18.2 percent in four years, with peer-reviewed wellbeing gains against matched comparators (Rose and colleagues). And three hundred fourteen community land trusts holding permanently affordable homes. On the debit side sit two, growth bought by acquisition rather than earned by diffusion, and platform cooperatives that died against subsidy. The strategic frame these numbers support is erosion, interstitial building plus symbiotic wins, not rupture (Wright).

The engine's missing term is demand, why an ordinary person bears clocks, ledgers, and consent rounds absent a crisis, and the record answers with a result the framework must carry whole. Costly requirements do screen free riders and predict commitment, the strictness result, but across eighty-three nineteenth-century communes the cost effect held for religious communities and bought secular ones nothing, because a cost reads as devotion only against a value the pool already holds sacred, and secular ideology could not make its demands non-negotiable (Iannaccone; Sosis and Bressler). The demand-failure quadrant is equally priced: a sixteen-million-member mutual draws three percent to its vote, gym members forgo six hundred dollars each rather than admit they will not attend (DellaVigna and Malmendier), and the collaborative project most like this framework lost its editors not to apathy but to its own compliance machinery landing on newcomers (Halfaker and colleagues). What survives between screening and burden is conditional, and the engine adopts it as three design rules. The first is visibility. Protected value must be visible and personal at the moment a cost is paid, the way homes under association rules carry a four percent premium from the same majority that dislikes the regime (Clarke and Freedman), so every organ binds its cost to a named asset of the member, the vested stake, the held home, the protected record, the codified exit, and is presented as insurance at the moment the asset is present, never as civic virtue. The second is proximity. Entry runs through ties and small asks, since cold contact converts near one in a thousand while a friend's kitchen converts near half, and a small first ask roughly quadruples consent to the large one (Freedman and Fraser), which grounds the staircase's mechanics: recruitment through existing relationships, a first rung sized to one meeting under the Lite tier, a named member owning each newcomer's first quarter. The third is load. The overhead band of EQ–02f, its compliance-load budget, binds, with any organ whose attention cost exceeds its visible protection redesigned or automated until it fits, because the record's clearest decline was a compliance budget spent at the front door.

Demand in the engine is therefore a function of three variables, visibility of the protected value, social proximity of the ask, and residual compliance load, and the strictness result governs the staircase's top rather than its base. Full membership stays costly, an explicit adult act with a vested stake, because screening belongs at the committed boundary while the burden hazard lives at the visitor's door, and the record punishes architectures that confuse the two. A crisis can spike visibility, which is why collapse recruits, but an engine that needs the crisis has already failed this section's question, and the demand term exists so adoption clears in ordinary time, carried as falsifier F–12 on the participation floor and the visibility differential at once.

OQ–06 begins from a concession. Transitional structures act inside systems whose members never agreed to them, and legitimacy rests on exit and fairness. This section builds the mechanism the concession alone lacked, and it begins by accepting the philosophical verdict rather than arguing with it. Consent as a founding act is dead. Residence is not agreement, as the man carried aboard the vessel in his sleep does not consent to the master by declining to leap into the sea (Hume), and no existing account, consent, fair play, gratitude, or natural duty, binds a particular person to a particular polity without inventing a consent that never occurred (Simmons). What survives are grounds that do not pretend to be consent. They are fairness owed for presumptively beneficial goods (Hart; Klosko), the natural duty to support just institutions (Rawls), and the honest naming of what an institution can claim without any of it, moral justification for wielding power, which is legitimacy decoupled from the obligation of the governed (Buchanan).

The framework re-grounds consent in the grammar it already owns. A decision is non-arbitrary not because it originated in agreement but because those it touches can effectively contest it (Pettit), with discourse among all affected as the regulative ideal (Habermas) and legibility supplied by the Justice of Light. Consent in Aeviterna is therefore not a signature at the founding. It is a standing veto path. A norm binds while it survives reasoned objection from those it affects, in forums built for objection, and every norm's warrant expires into its next review. This is decaying standing applied to rules instead of people, and it answers the ship directly, since what legitimates is not a past act nobody performed but a present objection anybody can raise.

Exit remains the outer wall, and the evidence disciplines how much weight it bears. Foot-voting works only under assumptions no real polity meets (Tiebout), easy exit drains the voice that reform needs (Hirschman), exit choice correlates with stratification (Dawkins; Bayer and McMillan), migration is rarer than the theory wants (Molloy and colleagues), and formal exit rights are no substitute for internal protections for those least able to use them (Okin). The kibbutz record adds the design warning the layer adopts as law. Communal wealth that members cannot take with them converts exit rights into fiction (Abramitzky). So exit in the consent layer is codified before any dispute, with a procedure, a clock, and asset portability, the member's stake vested and traveling, on the model of the one constitutional-scale exit clause written in peacetime.

Membership itself is staged, explicit, and revocable, on the pattern the working mechanisms converge on. A refundable capital stake makes joining an adult act (Mondragon), entry is graduated through visiting and provisional periods (Twin Oaks), decisions pass by consent rounds that ask for objections rather than applause (sociocracy), and, from the least likely teacher, the Anabaptist practice of baptism at majority holds re-consent as an adult act. Authority over minors is trusteeship that decays to zero at adulthood, when the person joins by explicit act or walks with the walk pre-funded. Consent-quality safeguards ride the stage gates, because recorded consent can be theater. Most people skip the terms entirely and the rest read for seconds (Obar and Oeldorf-Hirsch; Radin), preferences adapt to deprivation (Nussbaum; Elster), and bounded-choice systems manufacture subjectively free decisions (Lalich). The gates are therefore slow, informed, and reversible through the provisional period, and belonging costs are audited the way the Justice of Light audits power.

For those who cannot consent at all, children, future people, the cognitively incapacitated, the layer builds trusteeship in the guardian grammar of EQ–08e. The mandates are plural, decaying, contestable, and per-domain, substituted judgment comes first with best interests where preferences are unknowable, the ward is heard wherever capable, the ward's open future is held as a hard limit no guardian may foreclose (Feinberg), and the ward has standing to put the institution itself in the dock, as youth plaintiffs have now done against a state and won. And for the people in the institution's shadow who never joined, the boundary rule is that whoever a decision demonstrably touches gets the objection path and the legibility, contestation without membership, a vote in nothing but a veto path in everything that reaches them. That is the answer to the demos that otherwise explodes, since enfranchising all affected interests would enfranchise nearly everyone on nearly everything (Goodin). The shadow earns contestation, not government.

The boundary layer holds against force, and the record says force is rarely how alternative economies die. They die of prices and flows. Capital withdraws, clearing is denied, a subsidized rival sells below cost until the members peel away, the skilled are hired out one by one, and the inputs everything runs on are foreclosed in a weekend. Five channels, none of them an army, and the perimeter either walls all five or it has walled the wrong thing. This organ builds the economic wall, and it builds almost nothing new, because the defenses are organs the framework already runs, given explicit siege duties.

The record reshapes the threat before it arms the wall. The canonical sieges were joint failures, in which internal fragility handed the besieger its opening. Chile carried a documented squeeze, bilateral aid cut twentyfold, export credits to zero, multilateral lending throttled, and still the blockade leaked, credit rescheduled and replaced, while deficits near thirty percent of GDP, money-financed, did the killing the squeeze could not (Church Committee; Sigmund; Larraín and Meller). The kibbutzim were not besieged at all; a decade of cheap unindexed credit met a stabilization shock, and a movement restructured on its creditors' terms within a generation (Abramitzky). The first course of the wall is therefore fiscal. A polity that runs deficit-financed legitimacy is siege-soft regardless of its walls, which binds the overhead band of EQ–02f and the formula discipline of EQ–02e into the perimeter itself.

The clearing lesson is the organ's sharpest, and it is a contrast. WIR, a mutual-credit network born in the 1934 credit famine, has cleared countercyclically for ninety years and moves over a billion francs among fifty thousand firms (Stodder and Lietaer). INSTEX, a clearing substitute built by ten sponsoring states for besieged trade, completed one transaction in four years and was liquidated, because the banks' fear of the hegemon's enforcement outweighed the sponsors' will. A substitute that exists as an institution awaiting its crisis fails; one that exists as living trade credit holds. The same demand-side logic answers predation, which the law will not stop: since the recoupment standard, predation plaintiffs essentially stopped winning, and the documented platform case ran two hundred million dollars of monthly losses against one rival before acquiring it (Khan; the House record). What subsidy peels is customers. It struggles to peel owners, which is why the wall against the price war is the vested stake, the visible protected value, and the institutional procurement channel, the same symbiotic wins the transition engine already teaches (EQ–04d). And the talent channel is priced rather than walled, because compression's quit gradient is measured, high-ability members leaving worker-run firms at four and a half times the rate of their peers (Burdín), so differentials are bounded and published, succession carries the memory (EQ–07b), and retention beyond the differential is named as unpurchased.

What survival looks like is on the record too. The Mondragon network refused to rescue its failing flagship and then absorbed the blow it refused to prevent, relocating hundreds of members within months while the cooperative died, graceful degradation rather than invulnerability (Whyte and Whyte). Its shadow is the organ's dignity constraint. The failure's losses fell hardest on ten thousand outside savers holding subordinated paper, and a wall that survives by exporting its losses to non-members, or by breaching the floor the way a besieged state survived its Special Period, has failed in this framework's terms even where it holds. The drill that keeps the wall honest is already state practice. Finland's supply agency and the Swiss compulsory stocks hold months of fuel, grain, and medicine through private hands and have activated them in living memory, bounded buffers that buy negotiating time without autarky. The economic black-start imports that practice to cell scale on the calendar EQ–03e already keeps, and when the blockade is private rather than sovereign, the counter-siege on record ran through a borrowed court, which is the enforcement ladder's business (EQ–06f).

Aeviterna recognizes that care work naturally generates social gravity. People who consistently maintain systems, resolve conflict, mentor others, preserve continuity, and absorb instability will naturally accumulate trust and influence. This is not inherently corruption. Accumulated trust is how distributed systems coordinate without coercion.

However, stewardship must not calcify into sovereignty. The structural conditions that produce good stewards, namely accumulated trust, domain expertise, and institutional memory, are identical to the structural conditions that produce captured hierarchies. The difference is not character. It is architecture.

In Aeviterna, standing is domain-specific, requires ongoing contribution, decays over time without renewal, increases scrutiny rather than privilege, and can be revoked through transparent accountability processes.

These properties become a mechanism once decay is made concrete. Standing is modeled as a quantity that decays exponentially: S(t) = S0 · e^(-λt), where S0 is the standing at the last renewal, λ is the decay constant set per domain by how fast that domain's knowledge turns over. A fast-moving operational domain might carry a half-life of six to twelve months, so influence earned today is worth half as much within a year unless it is renewed; a slow, independence-bearing domain runs the other way, one long mandate with the clock on the powers around it. Contribution is the only inflow, and absent renewal standing falls back toward zero on its own, which is the structural property that keeps stewardship from hardening into sovereignty. Contribution enters as its own decaying increment, so with a steady rate of service standing converges to a finite ceiling rather than growing without bound, the behavior already running in conviction-voting systems that use this exact decay with an inflow.

These ranges are reasoned from how fast a field's knowledge turns over rather than chosen at random. The half-life of knowledge is an old idea (Machlup), and the turnover it names has been measured. Medical knowledge that took about fifty years to double in 1950 took roughly three and a half years to double by 2010 (Densen), while constitutional norms move over decades. What carries weight is the ordering, fast domains forgetting within months and slow ones holding for years, and not any single constant, which has to be fit to each domain and revised as the domain changes.

The tuning is a rule, not a guess. Each domain's half-life is anchored to a measurable proxy for how fast its knowledge turns over, the field's cited half-life or the inverse of its skill-depreciation rate, then clamped into the calibrated bands of EQ–05d so a mis-estimate cannot produce an absurd value: 6 to 24 months where operational skill is perishable, 2 to 5 years for professional and technical knowledge, 6 to 18 months in staggered cohorts for representative standing, and, for independence-bearing judgment, the inversion, one long non-renewable mandate with the decay applied to the review and appointment powers around the office, with the locked meta-core exempt from decay entirely. The tiers are republished and refit as a domain's measured turnover changes, and the estimator is open and forkable so the choice is contested rather than imposed.

Decaying standing is written into adoptable clauses, with terms, sunset dates, and staggered rotation, in the Model Charter, and the clocks themselves run in the Sunset Calendar, a self-contained tool that computes each role's review date and lays the rotation across the months. Its enforcement rides the ladder's inner rungs (EQ–06f), monitoring, the vested stake, and federated recognition, sized to survive the review the borrowed layer applies.

Almost everything in the framework is built for continuity, for resilience and the preservation of memory and stewardship across time. But a civilization that successfully preserves all of its institutions accumulates dead weight, and the accumulation is itself a failure mode. Old institutions capture resources and ossify around their original purpose, crowding out the new ones that changing conditions require, until the system loses the capacity to oscillate that EQ–02 defines as health. EQ–05 lets the standing of persons decay; this section lets institutions themselves end, and end well.

The model is ecological. A forest stays alive because death and decomposition return nutrients to the soil where new growth draws on them, and a forest that somehow kept every tree forever would starve. Human systems already have a crude version. Sunset provisions write an expiration date into law, so a body is abolished on a fixed date unless it is actively renewed, inverting the default from persists-until-killed to ends-unless-justified. Texas has run a Sunset Advisory Commission on this basis since 1977, reviewing roughly a hundred and thirty agencies on a rolling cycle and abolishing the ones that can no longer justify themselves. EQ–06 and EQ–09 already use this logic narrowly, in sunset clauses on emergency authority. Institutional mortality generalizes it into a full lifecycle.

The generalization is not that things should end, which is easy, but that they should end well, which is the neglected part. When an institution dies it holds resources, accumulated knowledge, relationships, and trained people, and the default failure is that these scatter or are captured in the collapse. A composting institution is designed from birth to hand them back. Its assets return to a defined commons, its memory is deposited in the distributed civic archive of EQ–06 rather than lost, and its people are transitioned rather than stranded. Death becomes a transfer rather than a loss, and the freed resources become the soil the next institution grows in. EQ–05 already treats decayed standing as compost for new entrants; this is the same move at the scale of whole institutions.

There is a deeper reason to build this in. Hannan and Freeman found that organizations rarely transform fundamentally, because they are too structurally inert, so populations of organizations evolve mostly through selection, through new ones founded and old ones dying, rather than through existing ones adapting. If that holds, a civilization's adaptive capacity depends heavily on its founding-and-dying rate, and one that suppresses institutional death has quietly suppressed its main channel of change. The honest danger is plain, since a mortality mechanism is a weapon and "this institution has outlived its purpose" is exactly what a motivated actor says to kill an institution that constrains them. So institutional death runs under the same anti-capture stack as everything else, with the decision transparent and contestable and never in the sole hands of those an institution exists to check, which is why the agonistic layer (EQ–06d) and the wider deliberation, not an executive, hold the question of what gets to end.

The framework prizes corrigibility, the capacity to detect error and change course, but corrigibility on paper is hollow if the world it builds cannot be changed back. Reversibility and repair are corrigibility made material, extended into infrastructure, technology, and law, and they complete the institutional mortality of EQ–05b. A body that can be wound down well is one whose commitments were reversible to begin with.

Irreversibility already works as a decision criterion, since the Rio Declaration holds that the threat of irreversible damage is no place for delay, and economics gives it a price through the quasi-option value of Arrow and Fisher, the worth of the information forgone by closing an option. The design rule is the Collingridge dilemma. Act while a technology is still cheap to control, before it entrenches and control becomes expensive.

Reversibility is already standard where the law requires it. Decommissioning funds for nuclear plants are held externally and paid in before operation, extended producer responsibility puts end-of-life back on the maker, and the right to repair has become hard law, with the European Union's 2024 directive obliging repair and spare parts and forbidding techniques that obstruct them. The honest counter-case is that some harms, extinction and crossed climate thresholds, are beyond any undo, so reversibility is a discipline for the recoverable paired with hard precaution for the rest.

The standing engine (EQ–05) gave decaying authority its mechanism and left its half-lives uncalibrated, which the framework has owned as OQ–02's residual and the amateur-government objection has exploited. This section calibrates, from the literatures that actually bear on the question, and reports what they support. The result is bands per domain class, each labeled by the strength of its anchor, one structural inversion, and two perimeter rules that bind harder than any number.

On competence, the evidence is better than the objection assumes. Across roughly three hundred documented deliberative processes, randomly selected citizens produce actionable recommendations on complex policy, with about four days of structured deliberation as the working minimum (OECD); deliberative polling shows replicated knowledge gains across a hundred polls in thirty countries (Fishkin), controlled experiments find real learning without polarization (Setälä and colleagues), and the Irish assembly's position carried a national referendum at two-thirds. The documented failure sits downstream, in uptake. Of the French climate convention's 149 proposals, roughly one in ten survived into policy unmodified, against a presidential promise of no filter. And the working precedent for decaying representative standing already runs, a permanent citizens' council seating members for eighteen-month terms with one third rotating every six months, so the body never turns over at once (Ostbelgien). Rotation, staggered, is not amateurism; it is how the amateurs stay competent.

The strongest counter-evidence is the term-limits record, and the framework adopts its lesson rather than disputing it. Across fifty states, term limits stripped legislative expertise and shifted power to governors, permanent staff, and lobbyists, the actors with no clock (Kousser; Carey and colleagues), the lobbyists themselves reporting the influence gain (Moncrief and Thompson), while removing careers altogether distorted effort the other way (Motolinía). Decayed authority does not evaporate; it migrates to whoever is exempt. Skill itself decays on a measured curve, near zero loss immediately after training to a large deficit past a year of nonuse, with task type moving the rate by an order of magnitude (Arthur and colleagues), and unrefreshed seniority inverts the experience curve in practice (Choudhry and colleagues). Institutions forget through rotation, restructuring, and outsourcing (Pollitt), which is why the calibration cannot stand without the memory organ. And where independence is the product, the evidence runs the other way entirely. Long-horizon central banks deliver lower inflation at no output cost, high governor turnover predicts inflation, and de facto judicial independence with non-renewable terms correlates with growth where renewable terms create please-the-reappointer incentives (Alesina and Summers; Cukierman and colleagues; Voigt and colleagues).

Tier 4 is the inversion the evidence forces. Re-earning standing is structurally identical to reappointment, the exact incentive that corrupts judges and central bankers, so for independence domains the engine flips. The officeholder gets one long, non-renewable mandate and pleads to no one, and the clock runs instead on the review and appointment powers around the office. The inversion protects the officeholder's independence, not the office's immortality. The institution around a Tier 4 seat still carries its EQ–05b expiration, and no mandate outlives its body. Two perimeter rules bind harder than the bands. No exempt adjacents. Every role touching a clocked domain, staff, counsel, consultant, auditor, lobbyist-equivalent, carries a clock in the same tier or the next one up, because the term-limits record shows what the alternative empowers. And decay pairs with memory. Every clocked role carries succession overlap, artifact handover, and a memory-organ deposit (EQ–07b) as a condition of exit, because rotation without externalized memory is engineered amnesia.

Aeviterna assumes that all systems drift toward concentration without corrective feedback. This is not a claim about human nature. It is a claim about incentive structures. Any system in which concentrated control produces benefits will attract concentration. The question is not whether this happens. It is how fast, and whether the system detects and corrects it before the correction becomes impossible.

The primary adversarial actors are not external enemies but internal beneficiaries, those who gain from suppressing the feedback mechanisms that would otherwise limit their accumulation. Anti-capture architecture must therefore be designed to function against motivated insiders with institutional access, not merely against naive external threats.

Anti-capture mechanisms are only as good as the adversaries they are tested against, so each one here is paired with a named threat and, in the spirit of security threat modeling, with the way it can itself be turned. The pairing exposes the deep residual. Rotation and decay defeat the entrenchment of any individual, yet the same credentialed network can recirculate through rotating seats and keep its hold. The signal to watch is therefore not individual turnover but network concentration, the overlap of ties among successive office-holders, which is what the tripwires are built to catch.

The table above is static, and the record is not. Security engineering made this upgrade fifteen years ago, from lists of threats to models of campaigns. The kill chain reframed intrusion as a phased progression in which the adversary must succeed at every stage while the defender needs to break only one link (Hutchins, Cloppert, and Amin), and the ATT&CK discipline replaced speculative threat lists with a catalogue of tactics compiled from observed incidents and revised as new behavior appears (MITRE). The framework imports both moves as the capture cascade, distilled from the best-documented campaigns on record: the South African state-capture commissions, the Hungarian and Polish constitutional captures and Poland's stalled restoration, the Boeing delegation case, and the flash-loan governance takeovers (the Zondo Commission; Scheppele; the House 737 MAX report). Capture arrives in six recurring stages: entry by mandate or delegation; the narrative environment; personnel before policy; the referee; the prize and the watchdogs; and lock-in. Each stage has a watcher already in this chapter. The consent layer's records watch entry, the cognitive immune system and institutionalized dissent watch the narrative stage, rotation, lot, and the network tripwires watch appointments, cross-domain audit and distributed memory watch the referee, capture bounties watch the prize, and the sunset discipline with EQ–09c's transcription ban watches lock-in.

Two structural rules join the stages, and both come from the record rather than from theory. First, friction is a defense in itself. The machine-speed cases show that where acquiring decision power and executing sit in the same window there is no link left to break, an attacker flash-borrowed a voting supermajority and drained a treasury with the borrowed votes repaid in the same block, and the one defense that survived contact was the mandatory delay between proposal and execution. Outside the crisis mode's bounded window, every consequential governance action therefore carries an irreducible, published delay scaled to its weight, and even crisis ignition is instant only for the enumerated, reviewable powers EQ–09c declares in advance. Second, reversal is asymmetric. Capture is fast to build and slow to undo lawfully, Poland's restoration stalled against its own captured referee, the hollowed revenue authority took years to rebuild while prosecutions still lag, and the delegation case ran eleven years to a resolution with no conviction. The cascade therefore weights its tripwires toward the early, cheap stages, expression, civic space, and appointments, which the population-scale data show moving first and fastest, while erosion halts, when it halts, through multiple accountability actors acting together (V-Dem; Lührmann and Lindberg; Bermeo; Laebens and Lührmann).

Anti-capture has a second front, the slow capture of the infrastructure itself by vendor lock-in or by a single cryptographic assumption. The defense is to treat every technical choice as replaceable. The framework specifies what a mechanism must guarantee and leaves how it is built to open standards that can be audited and swapped.

Anti-capture architecture addresses systemic drift toward concentration. But systems are also harmed by individuals: those who act in bad faith, those who cause damage through willful disregard, and those who reject the shared conditions that make civilizational cooperation possible. Aeviterna requires a principled response to all three. That response is the Justice of Light.

Justice in Aeviterna is not downstream of punishment theory. It derives from the same commitments that structure the rest of the framework: cognitive sovereignty requires that people understand what is being done in their name; stewardship demands that authority over others be accountable and corrigible; legitimacy collapses when harm is hidden rather than addressed. A justice system that operates in shadow is not compatible with a civilization organized around legible power.

The framework distinguishes between three categories of harm and responds to each proportionally. The goal across all three is the same: protect the commons, preserve dignity, maintain transparency, and keep every decision open to review. Aeviterna does not seek revenge. It seeks corrigible containment, the minimum intervention necessary to protect the community, held in the open, with standing to challenge built into the structure itself.

This applies equally to individuals and to institutions. A corporation that systematically suppresses feedback mechanisms, a platform that shapes cognition at scale for extractive ends, a state that governs without consent. These are actors causing harm within the framework's definition, and they are subject to the same principles of transparent accountability. The three-tier structure scales.

Four questions define the machinery: who adjudicates, who can challenge, how institutions are tiered, and how justice behaves in a crisis. They are answered as follows.

A civilization needs a foundation that fixes who and what counts, and one value that is never on the table. Aeviterna takes that value to be human dignity. The model is the German Basic Law, whose first article opens that human dignity shall be inviolable and that to respect and protect it is the duty of all state authority, and which places that clause beyond the reach of amendment. Dignity is the apex norm from which the rest of the rights substrate is derived, and in the language of this framework it is the one quantity that does not oscillate. The system is built on bounded movement between extremes, but dignity is a hard bound, not a dial.

The content of dignity has a usable core. In the Kantian reading carried into German constitutional practice through the object formula, a person must never be treated merely as an object or as a means to someone else's ends. The Federal Constitutional Court has treated dignity as non-balanceable, refusing even to weigh it against other goods, most starkly when it struck down a law that would have permitted shooting down a hijacked airliner, on the ground that doing so would treat the passengers aboard as mere objects. Aeviterna adopts the same posture. Dignity sets inviolable constraints that no optimization is allowed to trade away.

The comparative record, read in both directions, forces a sharper statement, and the framework adopts it. Full-blown non-balanceable dignity decides cases in one jurisdiction, and even there the absolutism is a rule about instrumentalizing the innocent rather than a general trump. The same aviation judgment that forbade sacrificing passengers held that downing a plane carrying only its hijackers offends nothing. Where dignity works across jurisdictions, it works as an enumerated list or a social minimum. The European court holds its torture article absolute even against saving a child's life (Gäfgen), the ICCPR's non-derogable list is an existing plural consensus, Colombia's constitutional court enforces a vital minimum daily through the tutela and holds the essential core of displaced persons' rights beyond any budget, and Kenya's supreme court made its dignity article decisive against violent eviction. Where dignity is a master concept instead, it dilutes into proportionality (South Africa; India; Israel), is formally abandoned, Canada's supreme court retreated from dignity as a legal test after nine years because it proved abstract, subjective, and a burden on the claimants it was meant to serve (Kapp), or is turned against the person it protects, as when a ban grounded in dignity overrode a man's plea that his dignity consisted in having a job (Wackenheim). The scholarly verdict matches, finding a thin common core, the ontological, relational, and limited-state claims, and beyond it no universal principled basis (McCrudden; Macklin).

The floor is therefore rebuilt as the plural floor of two tiers and three rules. Tier one, the absolute list. Absoluteness attaches to enumerated prohibited treatments, never to the word dignity: torture and cruel or degrading treatment and threats of them; the deliberate sacrifice of non-consenting innocents as a means, the object formula made a rule; slavery and servitude; denial of legal personality; punishment without law. Each anchor sits where plural consensus already exists. Tier two, the social minimum. Essential levels of subsistence, shelter, health, and education, lexically prior to other spending but resource-sensitive, enforced by the burden-shift of the minimum-core doctrine, an institution pleading scarcity proves exhaustion of alternatives, in the open (Bilchitz; Young). Three rules bind both tiers: they state agent-facing duties about treatment, never values to weigh, because the Canadian retreat shows what dignity-as-test does to claimants; neither tier may be invoked against the right-holder's own informed choice absent concrete harm to another, which closes the paternalism door the dwarf-tossing and peep-show cases opened; and norm-absoluteness never entails remedial rigidity, so structural, supervised, staged remedies on the Colombian model discharge hard-case pressure at the remedy level instead of eroding the norm. OQ–00's guardian gains precision for free. It interprets a list plus a burden-shift, a smaller and more legible object than dignity at large.

The absolute list is no longer only a courtroom doctrine; it now has a living machine-facing instance. The European Union's Artificial Intelligence Act (Regulation 2024/1689), whose Article 5 took effect in February 2025, writes an enumerated list of outright prohibitions directly onto automated systems, banning practices such as social scoring and untargeted facial-image scraping under penalty rather than leaving them to be balanced case by case, which is the object-formula-as-rule posture tier one demands. It also meets the sharpest objection to any absolute list, that an absolute floor is also an absolute weapon, with a procedure rather than a patch. Article 5's list is amended only through a defined review, not by transient majority, so the enumeration grows the way this framework wants any inviolable list to grow, through a reading seat (EQ–06e) and never through a passing coalition. The honest limit rides along. Article 5 is a qualified floor, not a pure one. It binds through market law, most of its bans turn on significant harm or disproportionality, and it carries law-enforcement carve-outs that its own critics call a blueprint for legitimizing surveillance. It proves an enumerated machine-facing floor is buildable, not that any real one is yet clean.

From the dignity of present persons the substrate extends outward in two directions that markets and ordinary politics both discount. Future persons will have dignity, so they hold standing now, which is where the framework's duties across generations come from, and the institutional seeds already exist in Wales' Future Generations Commissioner and the United Nations Declaration on Future Generations adopted in 2024. Non-human nature is harder, because identical dignity does not obviously transfer, but standing can still be granted through guardianship, as it already has been for the Whanganui River, for ecosystems under Ecuador's constitution, and for the Mar Menor lagoon in Spain. The framework treats these as standing held in trust rather than as claims of equivalence, which keeps the apex norm human while widening the circle of who is represented.

The harder question is how a foundation like this survives contact with an algorithmic legal system. Aeviterna's technical architecture pushes toward rules that execute automatically, and automation is exactly where dignity is most easily lost, because a system that optimizes people as data points treats them as means by default. Dignity therefore has to be encoded as a constraint on the machinery itself, not written as a preamble above it.

Dignity is contested terrain. Its content is read differently across traditions, it is invoked on both sides of hard bioethical questions, and the refusal to balance it has serious critics who hold that all rights must in the end be weighed. Making law machine-executable also risks freezing a single interpretation, which is why the rights substrate is bound to the falsifiability and the right to fork in EQ–10b rather than treated as settled, and the deeper question of who interprets the inviolable core is worked in FC–04 (The Guardian of the Core).

The anti-capture architecture protects dissent and the Justice of Light contains harm, but neither addresses the disagreement that does not resolve and should not be forced to. Deep value pluralism, the recognition after Isaiah Berlin that human goods are plural and sometimes incommensurable, means some conflicts have no synthesis. Liberty and equality, or loyalty and fairness, can pull in directions no reconciliation closes. A framework that treats every disagreement as a problem to be bridged will, at the limit, manufacture a consensus that flattens the very epistemic biodiversity EQ–03b exists to protect. Aeviterna needs a layer for the conflict that does not bridge.

The political theorist Chantal Mouffe named the distinction this layer turns on. Antagonism is the relation between enemies, the friend-or-foe logic in which the other side is to be destroyed. Agonism is the relation between adversaries, opponents whose right to their position is not in question even as it is fought, sometimes fiercely. On Mouffe's account conflict is not a failure of democracy to be engineered away but the condition of its existence, and the work of a healthy polity is to transform antagonism into agonism. Aeviterna adopts this directly. It does not resolve value conflict, it metabolizes it, building the channels through which deep disagreement stays generative instead of fracturing the society or being flattened out of it.

This locates productive conflict between two failures the framework already fears. Too much enforced consensus is the cognitive monoculture of EQ–03b, a civilization optimized into a single permitted view, brittle because it has bred out the variation it will need. Too much unmediated conflict is fracture, the friend-and-enemy collapse that ends in the harm the Justice of Light (EQ–06b) exists to contain. Agonism is the survivable band between them. The hard design point, the one Mouffe's critics press, is that the channel decays back toward antagonism the moment adversaries begin treating each other as enemies again. So the agonistic layer needs its own firebreak, a small core of shared commitment to the rules of contestation themselves that is not up for grabs. That is the one place the framework draws a hard line, because those who reject the right of others to contest cannot be granted the protections of contestation.

The payoff is a new reading for the vital-signs panel. Alongside the other signs it already tracks, a civilization has a disagreement capacity, how much unresolved and fully expressed conflict it can hold without either fracturing or collapsing into enforced agreement. This is the political twin of epistemic biodiversity, and like biodiversity it is invisible until it is gone. A society where every contested question has quietly become unsayable is not at peace, it has lost the capacity the measure tracks, usually shortly before it discovers it needed the views it suppressed. The panel carries this as its seventh sign of eight, read from proxies such as the share of people who feel free to voice dissent, and held provisional because the measure is young.

Every instrument in this framework ends in a seat where someone reads it, and five of the framework's organs build that seat: the guardian of the core interprets the constitutional floor (FC–04), a review reads the vital signs and sets the mode, an adjudicator runs the spine test (EQ–04b), a guardian panel speaks for an ecosystem (EQ–08e), and attesters certify a crisis trigger (EQ–09c). Five bespoke constitutions, one problem, and the framework's deepest conceded wound running through all of them. Whoever reads the instrument holds the polity's eyes, and the recursion residuals of OQ–03 and OQ–04, a capture that owns the indicator layer, an auditor no one audits, are all versions of a reader gone wrong. This organ extracts the constitution once, from the panel FC–04 already built and from a century of the world staffing exactly this job, and the other seats become instantiations of it.

The record is long enough to read. Committees read noisy signals better than individuals and are not slower, with engineered outsiders earning their seats empirically, their dissents predicting what insiders missed (Blinder and Morgan; Gerlach-Kristen). The reading binds unless publicly explained. For decades a central bank governor has owed a published letter, on a deadline, whenever the reading and the outcome diverge, and the counter-case shows the stakes, an advisory reading overridden silently on a question the readers were never asked, three of them resigning into the void (the open-letter regime; the Aduhelm record). The capture levers arrive one at a time and by name. Delete one appointer from a plural panel and the panel flips within two cycles; fail to capture by supermajority and come back for the budget; route around the plural appointer through an aligned court, and when the capture completes, the seat does not fake the instrument, it refuses to publish it, which is why the raw tallies are the tamper-evident seal (the election-body record). The two canonical failures share one feature, a reader funded by what it reads. The issuer-pays rating agencies stamped triple-A on what paid them and were named key enablers of the meltdown, and the anti-doping regime half-governed by sport's promoters reinstated the very program its own investigators had exposed (the FCIC; the McLaren record). And the repairs that work are method audits after a misread, not decapitations; expertise borrowed from the parties under exclusion rules, never litigation in the analysis room; the approval layer firewalled from the reading layer with every political edit publicly diffable; and lay watchers, whose immunity to career incentives is the point, holding the narrow junctures where capture concentrates, appointment, budget, and override (the Bernanke Review; the NTSB statute; the IPCC firewall; Bagg).

The five seats are instantiations of this one constitution, each keeping only its particulars. The guardian of the core keeps its full treatment in FC–04, the hardest case, from which this constitution was extracted. The vital-signs review is the seat pointed at the panel, with FC–04's framing-layer disciplines as its particulars. The spine adjudicator of EQ–04b is the seat pointed at the admission test. The more-than-human guardian of EQ–08e is the seat pointed at an ecosystem, with custodial knowledge in the expert chairs and the open-standing backstop as its particular. The crisis attesters of EQ–09c are the seat pointed at the ignition triggers, with the escalating renewal ladder as theirs, and their rule that renewal is held by the unrewarded is the same discipline as this organ's funding firewall, money and continuation kept apart from the read.

The seat's deepest reader is the one who frames the question, and the framework makes that reader carry a falsifier too. A frame is not admitted until it is pre-registered and dated. It must predict the near-term signals that run ahead of a transition, not the far outcome that never resolves in time. Those are the rising variance and slowing recovery that resilience science reads before a system tips (Scheffer and colleagues). They are scored against fixed and plural sources, and against a holdout kept back until after the frame is registered, so the grading does not run back through the frame it grades.

The seat stakes a bond on the frame and its predictions, forfeited when they fail. That stake is the seat's own and rides on the record, never posted to or collected by what it reads, so the bond and this organ's funding firewall are one discipline seen twice, money and continuation kept apart from the read. Its mandate decays and is re-based on the record, so a seat that reads the world well earns renewal while one that reads it badly decays out on the evidence rather than on trust. Renewal on the record governs only the scoreable operational reading, the object-level forecast reality can grade on the first three tiers of EQ–05d. The framing apex, whose frame reality cannot score against itself, takes that section's Tier 4 inversion instead, one long non-renewable mandate with its seat filled by lot (FC–04), because re-earning there would be reappointment and no half-life reconciles must-re-earn with must-not-please. A rival methodology is funded, forkable, and argued against the sitting one, because a bad frame is usually easier to refute than to sustain. This binding of the framing layer to a dated, scored forecast is the anchor. The register holds it as F–08. If anchored seats show no better calibration and no less capture than unbound seats across a decade of the tracker, the anchor buys nothing.

Its grip is uneven across the signs. A substrate-real sign, ecological throughput or material security or biological resilience, scores against outcomes directly. A reflexive sign, legitimacy or trust, whose reading helps constitute what it reads, can be bound only to its non-reflexive downstream, to compliance, exit, and enforcement cost. Even that runs out, and where it does the sign is flagged.

Still, where it grips, the anchor halts the oversight regress a third way. It does not climb to an apex that goes unwatched, and it does not only close the horizontal circle of mutual watchers who can share a frame. It binds the frame to reality, the one watcher that cannot be asked who watches it and does not share the framer's priors. Binding to reality never licenses reading past the cap. The anchor is fed at the layer of institutions, outcomes, and dated forecasts, where the legibility budget of EQ–03b defaults to a person-level zero, and a frame that needs individual detail spends from the same published budget as any instrument. The cap governs person-level leakage; the anchor is fed above it.

The seat also gauges its own capture. The concentration of framing power, how few and how unwatched the readers have become, is held in a band with its own decay clock. That band is the framework's one discipline turned inward on the reader. This is not a ninth vital sign, and it does not join the panel of eight, because the eight read the civilization and this reads the reader.

Its own residual is named at once. Defining the gauge is itself a framing act, held the only way framing is ever held, by a rival gauge kept live beside it. Where that rival is captured or never funded, the gauge is circular, and that failure is one of its recorded ways to break.

An architecture that forswears the sovereign still needs its rules to bind, and nearly every precedent this framework cites binds through public force, bylaws a court will enforce, statutory fines, entrenched constitutions, title law. The honest move is not to deny the borrow but to name it, organ by organ, the way every section already names a stability marker. Enforcement is not one thing. It is a ladder of five sources, reputation, vested stakes, structural locks, federated sanction, and borrowed courts, each with a real record and a failure edge, and the framework's discipline is that every organ states its rung, borrowed or own, so no reader has to guess what stands behind a rule.

The borrowed rung is older and stronger than it looks. An association's constitution has carried its own arbitration forum since a sheep-breeders' dispute in 1915, when a member's lawsuit was stayed because the articles sent member disputes to arbitration; company law in the major families holds a registered constitution to be a contract binding the body and each member; and the New York Convention obliges the courts of 172 states to refer parties to arbitration and enforce the awards. The full stack already runs twice over, in the religious tribunals whose awards state courts enforce under ordinary arbitration law and in sport's standing transnational court. A cell that encodes its charter as enforceable bylaws and seats an arbitration path in peacetime has invented nothing; it has joined a hundred-year-old practice. And the alternative is not independence. The newest governance experiments that declined to choose a legal form had one chosen for them, unincorporated association or general partnership, members personally liable, which is the borrow arriving on the worst available terms.

The lender sets conditions, and they run on a gradient that cuts against ambition. Courts defer most where the stakes are spiritual, refusing to review a hierarchical religious tribunal even for arbitrariness, and least where an association gatekeeps a livelihood, where expulsion draws review in every major family and penalty doctrine polices any forfeiture for proportion. The space that remains is workable, since a forfeiture protecting a legitimate governance interest can stand even beyond provable loss, but the planning fact is permanent: an economically consequential cell will not enjoy the church's autonomy, and its internal discipline must be drafted to survive the hostile review it will get. The lender can also foreclose, by statute where a legislature dislikes a private ordering, or entirely, when the state that lends the courts joins the siege, at which point enforcement falls to the inner rungs, which the record supports only below the threshold where a great power takes an interest (EQ–04b; EQ–04f).

The inner rungs hold to the edge of what the community controls. Reputation with monitoring and graduated sanctions is well supported and cheap, and it breaks against exit options and outside capital, as the canonical trust networks learned. Vested stakes move behavior at the margin and bind only inside judicially policed proportion. Structural locks, the veto share, the purpose trust, the software default, enforce almost for free and are blind to insolvency and dependent on a watcher. Federated sanction binds in exact proportion to what the federation gatekeeps, eleven days to compliance where it holds the only game, a shrugged-off label where it does not. The map below states the rung for every load-bearing organ, which is this section's discipline made auditable, and the residual is stated with it: between polities there is no court, and the outermost rung is the federation's own sanction, where the boundary layer and the economic wall take the watch.

The framework staffs its reader and decider seats by lot, because a randomly drawn seat cannot be bribed, lobbied, or made to serve a career. That secures who sits. It leaves untouched a prior question the lottery cannot reach: who decides what the seated body is asked, which options it may choose among, what evidence it hears, and who briefs it. That upstream work, the definer function, is done by a standing apparatus: a commissioning authority, a secretariat, a chosen roster of experts. A durable faction that owns that apparatus can determine the outcome of a perfectly random body, because it has already set the question and pruned the menu. This is not the value pluralism EQ–06d protects, the legitimate contest between adversaries over goods that do not reconcile. It is the capture of the frame itself.

The record names the lever. Schattschneider called the definition of the alternatives the supreme instrument of power, and Bachrach and Baratz called its exercise the second face of power, the non-decision that keeps a threatening option off the agenda before any vote is taken. Madison held that faction cannot be abolished, only counterposed, and Michels found that organization itself breeds an entrenched leadership even in the bodies most committed to internal democracy. The deliberative record confirms it in practice. In the celebrated citizens' assemblies, Ireland's and France's climate convention among them, the frame was set by the parliament, the secretariat, or the state-aligned experts, and not by the random members, and where the menu was one-sided the citizens were left with the impression that no real choice was on offer. Random selection of the jury is beside the point when a durable interest owns the definition of the alternatives.

So the definer earns a seat of its own, held to the discipline the reader already carries. The framework cannot abolish the definer any more than Madison could abolish faction, so it does to the frame what it does to the reader. It multiplies it, counterposes it, exposes it, and denies it tenure.

Aeviterna treats knowledge as primary civilizational infrastructure, rather than as a product, a commodity, or intellectual property. It belongs in the same category as water, roads, and electrical grids. Its privatization produces civilizational fragility regardless of its short-term economic logic.

The Knowledge Commons functions as shared civic memory, open scientific collaboration infrastructure, distributed educational architecture, transparent governance archive, and collective resilience repository. These are not separate institutions. They are facets of a single claim, that a society's collective understanding of itself and its world is a shared resource that cannot be enclosed without consequences that eventually reach everyone.

Aeviterna seeks to transition from scarcity-centered value systems, in which knowledge is valuable because it is withheld, toward contribution-centered coordination systems, in which understanding itself becomes a foundational form of wealth. This is not idealism. It is an observation about where value actually comes from in systems that survive.

Made operational, the commons is governed, not merely open. Knowledge is a shared resource subject to social dilemmas, analyzable with the same institutional discipline as fisheries and forests (Hess and Ostrom), rebuilt for constructed commons whose participants produce the resource and whose boundaries are built rather than found (Frischmann, Madison, and Strandburg). The working institutions converge on one shape, a written constitution with elected, recallable authority earned through organization-building rather than output alone (O'Mahony and Ferraro), a hard disclosure clock that outran enclosure at civilizational scale, genome data public within twenty-four hours (the Bermuda Principles), and shared infrastructure that makes release the default rather than a virtue. The documented failure modes sit at maintenance and succession, not creation: a decade of editor decline traced to newcomer-hostile quality automation and calcified policy (Halfaker and colleagues), critical infrastructure carried by unpaid volunteers whose burnout is a systemic risk (Eghbal), and the patient attacker who social-engineered a burned-out solo maintainer into handing over the keys to most of the world's servers. Voluntary pledges are the weakest instrument in the record, suffering obscurity, impermanence, and non-binding transfer (Contreras).

The commons therefore adopts three commitments.

• The disclosure clock. Knowledge produced with commons resources enters the commons on a hard, short, automatic deadline, release as infrastructure default, not author discretion. This is the Bermuda mechanism generalized, the one anti-enclosure device with a civilizational-scale win on its record.
• Impact-metered pull funding. The commons pays for knowledge with pre-committed purchases, prizes, and ex-post impact payments, and buys out enclosures at auction prices where they block the floor (Kremer). The advance market commitment that vaccinated tens of millions of children is the working precedent, reward decoupled from monopoly, results entering the commons on delivery.
• Constitutional stewardship with a paid maintenance line. Every commons institution carries a written constitution with earned, recallable, bounded authority, plus the two clauses the failure record demands: a funded maintenance and succession line, because commons die of unpaid upkeep, and newcomer-retention duties with automation audited for first-contact hostility, because commons also die of their own immune systems. The clause binds to the calibration perimeter (EQ–05d) and the memory organ (EQ–07b).

The enclosure pressure this answers is current, not historical. Five publishers hold more than half of published research (Larivière and colleagues), the open web is closing against crawlers while incumbents sign exclusive training-data deals (Longpre and colleagues), and the pattern was named two decades ago as the second enclosure movement (Boyle).

The Knowledge Commons (EQ–07) holds what a civilization knows, but knowledge is a flow that must be re-transmitted each generation or it is silently lost, and collapse, in Tainter's reading, is precisely the loss of the complexity that knowledge embodied. After the Bronze Age collapse, writing itself fell out of use for centuries; the Antikythera mechanism's sophistication was not matched again for more than a thousand years. Memory is not self-sustaining.

The hardest part is tacit. Polanyi showed that we know more than we can tell, so a complete record cannot reconstitute a capability without the people who hold it, and automating the novice tasks through which experts are trained can quietly break the apprenticeship that carries skill across generations. Records themselves rot, through the bit rot and format obsolescence of a digital dark age, and even long-horizon archives are fragile, as the flooding of the Svalbard seed vault in 2017 showed.

The most careful work on very long-term transmission, the nuclear-semiotics marker studies, builds warnings on a defense-in-depth principle of redundant materials and languages while admitting that no guarantee survives ten thousand years. The curation question is a power question, since whoever decides what to keep can entrench a partial record, and the opposite danger, total memory, harms reinvention, so a succession layer needs a deliberate, accountable practice of forgetting as well as keeping.

Aeviterna does not exist merely to optimize systems more efficiently. Its purpose is to reclaim human cognitive bandwidth from survival anxiety, extraction pressure, and perpetual optimization culture. And reclaiming human time is not a secondary benefit of getting the economics right. It is not quality of life. It is architecture.

Human curiosity, play, rest, and exploration are not personal goods that a well-functioning civilization happens to enable. They are the mechanisms by which a civilization remains capable of imagining futures that its current optimization targets cannot generate. When a society eliminates the space for non-instrumental thinking, it does not merely become less enjoyable. It becomes less capable of the kind of cognitive mutation that adaptive systems require to survive changing conditions.

This connects directly to EQ–03b. The reclamation of human time is one of the primary mechanisms by which cognitive ecology is maintained at the individual level. Automation, AI, and advanced coordination systems should reduce unnecessary labor burdens not because leisure is pleasant (though it is) but because the distributed exploration that happens in unstructured human time is irreplaceable civilizational infrastructure. It cannot be scheduled or assigned. It requires the protected space that extraction systematically destroys.

A framework that produces material security but eliminates the space for curiosity, play, and self-determined meaning has failed at the civilizational level regardless of its output metrics. The measure of success is not efficiency. It is whether the civilization retains the capacity to be surprised by itself.

A civilization has to answer what a life is for, and it has to give people somewhere to belong. Markets answer the first with consumption and the second with very little, and a society that beats scarcity on goods while failing on meaning still decays from the inside. This is the organ most post-capitalist thinking leaves out, because most of it stops at distribution. Aeviterna treats meaning and belonging as infrastructure, on the same footing as energy or law.

The evidence that material success does not supply meaning is solid. Kasser's work shows that strongly materialistic goals track lower wellbeing rather than higher. The larger claims are more contested, and the framework states them carefully. There is a wide discussion of a meaning crisis in secular societies, which is a philosophical framing rather than a measured fact. There is real evidence of rising isolation, including the United States Surgeon General's 2023 advisory on loneliness, though whether connection has fallen over the long run is disputed. And there is the deaths-of-despair literature, whose interpretation is actively debated. The framework leans only on what holds, that belonging is a determinant of health and that consumption is a poor substitute for meaning.

Meaning is the organ most easily corrupted by the state, because a civilization that manufactures meaning from the top slides quickly into coercion. The framework therefore commits to cultivating the conditions for meaning rather than dictating its content, and the dignity constraint in EQ–06c is what forbids any official or imposed meaning. Belonging is built, not assigned.

That commitment now has a mechanism, and the mechanism has a century of precedent. Denmark has funded meaning infrastructure content-neutrally since 1851, folk high schools and free associations supported by statutory duty, the pivotal clause leaving the association itself to choose its subjects, teachers, and form, exams prohibited so credentialism cannot annex the sphere, all of it inside a procedural floor of democratic structure, open membership, and non-profit operation (the Folkeoplysningsloven). The funder audits form, never belief. The adversarial record is just as legible. Sweden's parallel system paid on self-reported volume and bought phantom participants, the national audit finding deficiencies in every link of the control chain, followed by a values-floor fight and a one-third funding cut that fell on honest associations too (Riksrevisionen). France's individualized training accounts bred industrial fraud within two years, cured only by reintroduced friction. The design rules follow from the comparison: fund associations, never individuals; cap the subsidy share so every container carries member co-payment; verify by sampled attendance, never by continuous monitoring.

Measurement has its own precedent. Belonging can be read behaviorally and in aggregate. Twenty-one billion friendship ties, published only as privacy-protected aggregates by place, yielded the strongest known community-level predictor of upward mobility, economic connectedness, the share of ties that cross class (Chetty and colleagues). No individual can inflate an aggregate bridging statistic the way self-reported study hours were inflated, so the framework reads belonging at the cell and federation level, from participation breadth, bridging ties, and venue density, and never lets a unit-level reading enter a funding formula. Money follows form; measurement reads behavior; the two are kept apart by design, which is what makes the Goodhart question answerable rather than fatal.

A civilization runs on more than energy and materials. It runs on the daily, mostly unpaid work of raising children, tending the sick and the old, and keeping existing systems in repair. The economy of enough and the metabolism govern throughput; this is the human upkeep that throughput depends on, and the framework has so far assumed it rather than named it.

The scale is measured, not asserted. The International Labour Organization values unpaid care work at roughly nine percent of world output, the equivalent of two billion people working full time without pay, with about three-quarters of the hours done by women, and national accounts treat it as outside the production boundary, so the work that sustains the workforce is invisible to the measure of value. Nancy Fraser names the result a crisis of care. An economy oriented to unlimited accumulation destabilizes the very reproduction it relies on.

The same underpricing runs through physical systems. A culture fixated on innovation undervalues the maintenance and repair that keep infrastructure and institutions working, and deferred maintenance is the physical twin of the crisis of care, a debt that accumulates quietly until something fails. Care and maintenance are one commitment seen in people and in things. Keep what sustains us in good repair.

The substrate now has its instruments, and the counting question is settled by where in the aggregation hierarchy the counting happens. The national-accounts world drew the line in 2025. Unpaid household services stay outside GDP, and recognition lives in extended accounts compiled beside it, diary-based, active and supervisory care distinguished, valued at a stated lower bound (the SNA 2025 revision). Two ledgers, commensurable and never merged, which is this organ's "valued without being priced" made operational. The record adds a warning and a name. The warning is that optional ledgers starve, one national account went seven years between editions and another has not been updated in half a decade, so publication needs a clock. The name. The two counting designs are now both on record. The first is the anonymous sampled time-use diary with statutory confidentiality, aggregates only, the instrument beneath every satellite account. The second is the per-transaction verification log, which one country mandated for publicly funded home care and which turned counting into GPS tracking of carers and of the disabled people they care for, with unpaid claims from failed check-ins and geofencing that penalizes leaving the house (the Cures Act verification record; Mateescu). The firewall follows. Care is counted only at a level of aggregation from which no individual's schedule can be reconstructed.

Redistribution moved where defaults moved and faded where they were gestures. Five reserved, well-paid, non-transferable father weeks raised participation by 250 percent and measurably reduced sex specialization years later; a four-week quota shows household effects fifteen years on; a single lightly-held month moved uptake and nothing else long-run; and even a statutory equal split does not yet equalize practice, since the transferable weeks flow one way (Patnaik; Kotsadam and Finseraas; Ekberg and colleagues). The framework's care entitlements therefore follow the thresholds the record supports, reserved, non-transferable, adequately paid, framed as the carer's own, and their effect is read in the diaries, not the statute book. The uptake ceiling on record is Iceland's equal, non-transferable, well-paid six-plus-six-plus-six months, where fathers now take roughly nine in ten, and even there the mother stays the primary carer in the first year, so recognition and even a strong quota move the doing without equalizing it. What might move it further is standing. When care earns a durable claim on the systems it sustains, its incentive stops being recognition alone. Maintenance binds the same way. Statutory 30-year infrastructure strategies raised audited renewals to 85 percent of forecast depreciation and still under-reinvest; states with mandated asset plans added a hundred thousand new lane-miles while their repair backlogs grew; the flagship paydown fund expired with the backlog rising (the New Zealand audit record; Repair Priorities). Ledgers change the conversation; only gates change the allocation. So the ledger carries a gate. Above a threshold, new build is funded only where the renewal ratio meets its floor, repair before expansion, written into the money rather than beside it.

The framework's vital signs are mostly metaphorical; this one is literal. Human health is not a private commodity but the visible band of a shared system, in which one organism's immunity, one ecosystem's intactness, and one society's baseline conditions together determine whether populations stay in survivable ranges. The official One Health framing treats human, animal, and environmental health as one interdependent system, which is how the framework already treats its other signs.

Three findings make this load-bearing. Prevention is radically cheaper than collapse, since the cost of a pandemic runs to the trillions while the cost of reducing spillover runs to tens of billions a year. Antimicrobial resistance is a slow-moving systemic risk, with on the order of 1.27 million deaths attributable to bacterial resistance in 2019, reported here with its uncertainty rather than as a single dramatic forecast. And the determinants of health are largely social, since the Whitehall studies found a steep mortality gradient across the social hierarchy that classic risk factors only partly explain.

The core goods are commons. Herd immunity is a positive externality prone to free-riding, and a shared microbial and biodiversity base is non-excludable, so prevention's benefit collapses when protection is privatized or enclosed. Health read as a central capability grounds a biological reading of the dignity floor, bodily health and integrity as minimums beneath which no person may fall, whatever their rank.

The framework already widens the circle of standing past present persons. EQ–06c seats future persons through guardianship and grants nature standing too, but only in a sentence, held in trust rather than as a claim of equivalence. Nature otherwise enters the framework twice more and thinly, as the ecological ceiling of the economics of enough and as the ecological-throughput vital sign. Every other protected interest is given an organ, an agent, and a decaying-standing account of who speaks for it. The more-than-human is given a constraint and a reading. The asymmetry is unprincipled, and this organ closes it.

The framework has a sharper reason to build it than the field usually has. The central unsolved problem of rights-of-nature law, who may speak for nature without becoming its unaccountable sovereign, is P–05 and OQ–00 in another body. Aeviterna has a worked, if leaky, answer to that problem already in the guardian of the core, so bringing the more-than-human inside lets the framework point its strongest machinery at a question the rest of the field has mostly left open, and lets the case stress-test that machinery in return.

The record of the last decade says why the design has to be built this way. Legal personhood for an ecosystem has proven real but narrow, strong as a shield and weak as a program. Ecuador's Los Cedros ruling stopped the mining inside a cloud forest, and a 2024 field study found the extraction halted with no infrastructure left behind; Panama's highest court voided a major copper mine on a rights-of-nature ground in 2023; and in 2024 Spain's Constitutional Court upheld legal personhood for the Mar Menor lagoon, the first such ruling by an EU member state's highest court. Where the task was restoration, funding, and enforcement against diffuse harm, the same instrument mostly failed. Colombia's Atrato River, a subject of rights since 2016, remained by its own ombudsman's account a plan on paper, its guardians unpaid and its illegal mining spreading. The lesson is not that standing is empty but that recognition without a funded, enforcing, capture-resistant agent behind it is, which is why the mechanisms below aim at enforcement and capture rather than at the declaration.

The unit of representation is a decaying ecological standing held by a guardian for a specified natural system, a watershed, a forest, a species assemblage, never nature in general, because generality is the failure mode that builds the throne by the shortest route. The guardian is constituted as a reading seat (EQ–06e) pointed at its ecosystem, with the seat's constitution inherited rather than rebuilt: plural, part-lot with the lay majority able to outvote the experts, tenured on a decaying mandate renewed only from outside, never funded by what it reads, publishing its raw readings and dissents. Its particulars are its own. The local custodians whose knowledge of the system is deepest hold the expert chairs, it speaks for one named system and never nature in general, and a standing shadow guardian waits so the system never depends on this panel surviving.

Contestability is the heart of the design and the answer to the field's sharpest finding, that whoever defines nature's interest holds the real power. Any cell, or a quorum of citizens, may submit a competing reading of a system's interest and force a contest, and if the alternative prevails the standing of those who produced the original reading decays, so being wrong about a river carries a cost. Capture bounties reward anyone who shows a guardian's reading was bent toward a developer, an extractor, or the guardian's own institution. The Te Urewera case is the live proof, since a guardian acting against part of what it guarded was checked when a single member of the community that held the seat forced the question, and it bit because the guardian's process was logged and reviewable.

The organ then adds a more-than-human floor that mirrors the dignity floor without claiming the two are equivalent: a small, lit, non-negotiable set of ecological minimums beneath which no decision may push a system regardless of efficiency or majority, that no decision may knowingly drive a species to extinction, collapse an ecosystem past a recognized tipping point, or breach a hard planetary boundary. It is kept narrow for the same reason the human core is, so it stays legible and a poor capture target, and it is bounded to a system's continuation and self-renewal rather than the welfare of every organism, since nature runs on predation. With an agent in place, the throughput vital sign stops being an unowned number and becomes the instrument this guardian reads and the breach it can hold, which answers EQ–02c's open question by enforcing the ceiling through a plural, contestable guardian rather than a central rationing authority. The organ is the ecological reading of the same One Health system EQ–08d governs from the human side, since intact ecosystems are upstream pandemic prevention, and it writes the nature half of EQ–06c's standing substrate to the depth the future-persons half already has.

The care and maintenance substrate (EQ–08c) makes the unpaid work of a society visible and redistributes it, and the rest of the architecture asks members for hours, the sortition seats, the quarterly reviews, the governance overhead of EQ–02f. Both draw on one account, the discretionary hours of working-age adults, and both have assumed that account is full. In an aging, sub-replacement society it is not. The working-age cohort, the numerator, shrinks in the same decades when the population it must serve, the old who need care, grows. Global fertility is already below replacement across most of the world, the number of working-age adults per person over sixty-five is falling from about four toward one and a half in the leading cases, and the eighty-and-over cohort, the heaviest users of hands-on care, is the fastest-growing of all (United Nations, World Population Prospects 2024; OECD, Pensions at a Glance 2025).

The consequence is an arithmetic the substrate has to meet. As the pyramid inverts, care demand rises while the hours to meet it fall, and the participation budget lands on the same shrinking cohort, heaviest on the carers, most of them women, it means to protect. Migration and automation soften the blow without repealing it. The migration that would hold the age structure is an order of magnitude beyond any real flow, and the sending countries are aging too (United Nations, Replacement Migration). Two decades of subsidized care robots in Japan neither cut the labor nor stayed in use, the devices as often shelved as adopted (Wright, Robots Won't Save Japan). The counter-case, that healthy aging and longer working lives blunt the alarm, weakens the fiscal panic but not this one, because the oldest-old care load is the least automatable and cannot be asked to work longer.

So the organ adds the term the substrate lacks, a scarcity price on care and a governor on participation, and it keeps the one promise it cannot design away. Care is the demand that cannot be cut; governance is the buffer that can.

Aeviterna recognizes that systems behave differently during crisis, scarcity, abundance, and recovery phases. Governance structures appropriate for one phase may be actively harmful in another. A framework that does not account for this will either be too rigid to handle crisis or too permissive to prevent crisis from becoming permanent.

Temporary centralization may emerge during emergencies. This is not a failure of the framework. It is a recognized and expected mode. The requirement is that emergency authority carry built-in decay. It must unwind, crisis governance must dissolve, and recovery protocols must actively redistribute coordination back to distributed structures. Without decay mechanisms, crises permanently fossilize hierarchy. The emergency becomes the new normal.

The four modes are not different governments but different settings of the same one, distinguished by how much temporary centralization the situation licenses and how fast that centralization must unwind. Abundance runs on distributed coordination, scarcity tightens prioritization without concentrating control, crisis may license real but temporary centralization, and recovery exists to spend that centralization back down. The mode is set by measurable conditions rather than by the discretion of whoever would benefit from the switch, and every step toward concentration is paired with a scheduled step back.

Two dangers attend the crisis mode: that emergency authority never expires, and that someone manufactures a crisis to trigger it. The framework answers both structurally. Emergency authority carries an automatic expiry, and extension requires an escalating threshold met by a rotated body rather than passive continuation. The actor who declares an emergency is held apart from the actor who receives the expanded authority, with the powers held in escrow between them, so declaring a crisis grants no one new control. The nested units of the wider structure hold a veto over a declared emergency, and any standing gained during crisis decays faster than ordinary standing, so the ratchet reverses on its own. Recovery then runs an explicit handoff protocol that returns each centralized function to distributed stewardship and logs the return in civic memory.

How the thresholds get set is itself a design choice with precedent. The cleanest analogues are quantitative and automatic. Market circuit breakers trip at fixed, publicly known levels measured against the prior close, which removes discretion but invites a magnet effect as behavior rushes the line, a caution against naive variance triggers. Comparative constitutional practice points the same way as the standing rules here, a formal declaration and a bounded scope, with prior approval required to prolong, and each measure meeting necessity and proportionality on its own (the Venice Commission benchmarks; ICCPR Article 4). The strongest reform template flips the default so emergency powers expire unless a rotated body affirmatively renews them against an escalating burden, the reverse of renewal by inertia.

The framework governs how much a civilization extracts and how it perceives, but not how fast, and tempo is itself extractable. A vital sign inside its band but changing too fast is still a resilience failure, because the slow parts of a system cannot absorb change delivered at the speed of the fast parts. Social acceleration, in Rosa's account, is a self-reinforcing loop that shrinks the present, and the answer is a governable tempo rather than mere slowness.

The resilience literature supplies the rigor. Slow, controlling variables govern how a system absorbs shocks, and pushing fast variables hard while slow variables drift toward thresholds is how a system is tipped into a worse regime, a scale mismatch between the pace of management and the pace of what it manages. Brand's pace layering turns this into a design principle, a civilization layered by tempo from fast commerce to slow culture and nature, endangered when a fast layer drives a slow one at its own speed.

The speed limit already exists in miniature. After the Flash Crash of 2010, markets adopted circuit breakers that halt trading on sharp moves and pause a stock that runs outside a band, externally triggered and sunset-bounded, the financial twin of the crisis mode of EQ–09. The labor analogue is the right to disconnect. Bounding tempo is an ordinary instrument, not a utopian one.

OQ–04 is the failure mode the rest of the framework keeps deferring to, an emergency declared or sustained pretextually, with the approval of the very public that would need to object, because the framing layer, not the ballot, is where the capture happened. The guardian of the core concedes that a popular capture clears every contest; the more-than-human organ names the same hole; EQ–09 separates declarer from beneficiary in principle. This organ builds the machinery, from an empirical record that is now good enough to build with and honest enough to discipline the design.

The record cuts both ways, and both edges are load-bearing. Emergencies erode. Nine of ten constitutions carry emergency provisions, at least 137 countries declared one across three decades, 115 of 853 coded declarations were unlawful under the declaring state's own constitution, and democracies under a state of emergency were 75 percent more likely to erode than those without (Bjørnskov and Voigt; Lührmann and Rooney). The paper safeguards do not predict who abuses them, and requiring second-chamber approval correlates with more unconstitutional declarations, not fewer, so stacking approval chambers is not the answer. Yet the modal crisis response is bound, not Schmittian. Across a hundred-plus pandemic responses, courts and legislatures constrained executives far more than the theory of the exception predicts, two-thirds of democracies violated no standard, and violating standards bought no measurable health benefit (Ginsburg and Versteeg; Edgell and colleagues). The organ targets a tail risk and must not tax the modal case to death. The deepest drain runs the other way. Counterfactual modeling attributes roughly half the first American wave's reported deaths to distancing begun one to two weeks too late (Pei and colleagues), publics traded liberties in proportion to actual risk, and safeguards themselves have killed, a state court voiding every pandemic order at once mid-crisis, a parliament refusing extension at the infection peak. Any organ that can force lapse during a real emergency has a body count in expectation, and this one owns it.

Two further findings shape the machinery. The real ratchet is transcription, not extension. France ended its emergency the day a statute moved its powers into ordinary law, Egypt lifted its emergency days before amendments made the provisions permanent, and the sunset clause, the standard remedy, is near-worthless on the record, renewals pro forma for decades, expiry extremely rare, and sunsets buying support asymmetrically in experiments (Finn; Underhill and Ayres). And the framing layer is cheap, industrialized, and mostly domestic. One state fabricates roughly 448 million posts a year aimed at distraction rather than argument (King, Pan, and Roberts), two-thirds of the covert influence networks a single platform removed targeted their own domestic publics (Meta), and personalized model-driven persuasion now beats humans in controlled debate (Salvi and colleagues), while the measured effect of the canonical deepfake cases stays small to undetectable, so the panic is also a weapon.

The organ is five rules, each answering a named failure.

• The declarer never renews. Any qualified body may ignite crisis mode instantly, because delay is the one quantified killer. Renewal belongs to a body that gains nothing from continuation, no budget, jurisdiction, personnel, or tenure, and renewal thresholds escalate with each cycle, the supermajoritarian escalator (Ackerman), so only a crisis that persists in fact can hold the coalition.
• Automatic, procedural-first review. Every declaration and renewal is docketed before the review organ with no one needing standing to bring it. The first questions are the ones courts have actually won with: right instrument, right legal basis, right gate. No one out-argues the executive's epidemiology on day three.
• Triggers are observable, never popular. Crisis mode keys to pre-registered, externally attested indicators, mortality, attributed attack, infrastructure failure, read by bodies outside the declarer's appointment chain. Approval polls, plebiscites, and trending sentiment are inadmissible as triggers or renewal evidence, because the framing layer is the cheapest thing to buy. This is the one place the framework deliberately refuses the demos, and says so aloud.
• The transcription ban. No emergency-born power enters ordinary law during the emergency or its cooling-off period; permanence requires de novo justification and an independent audit of the power's emergency record. This targets the ratchet the record actually shows.
• Speed for ignition, friction for duration, cure for invalidation. Reviewers who strike a measure issue cure windows and substitution duties, never instant vacuums, because the cliff-edge invalidation mid-crisis is also a documented failure mode.

Two of these rules carry the answer to the section's own recursion, and a live case shows the failure and the fix at once. In 2025 a mayor declared the passage of a duly enacted law an emergency and renewed the declaration every thirty days, until a court struck the orders down as exceeding the power, the manufactured crisis and the self-renewal in a single exhibit (the New York solitary-confinement orders). The fix hardens two rules. The observable trigger is affirmed not by the declarer but by a reading seat the declarer can neither appoint nor defund (EQ–06e), which puts the attester out of reach of both capture levers at once. And renewal runs on a presumption exchange, where the same votes that would extend a power under an affirmative default instead let it lapse under a sunset default, so silence ends the emergency and continuation must be re-earned against the declarer's burden (Lauenroth), the discipline the record demands, since nine in ten democratic breakdowns run through emergency rule.

Aeviterna is fundamentally long-horizon. Not because long-term thinking is virtuous in the abstract, but because the failure modes Aeviterna is designed to address are slow. Extraction is profitable quarter by quarter and catastrophic decade by decade. Cognitive capture accumulates imperceptibly until correction becomes unthinkable. Stewardship that calcifies into sovereignty does so gradually, through hundreds of decisions that each seem reasonable in isolation.

A framework that cannot hold long-horizon thinking institutionally, not just rhetorically, cannot address these failure modes. The success of a civilization is therefore measured not only through output, but through resilience, continuity, recoverability, distributed stewardship, cognitive sovereignty, adaptive capacity, and ecological survivability. These are not soft measures. They are the measures that determine whether the civilization is still operating in one hundred years.

A long horizon forces a harder question the framework had left implicit. Aeviterna prizes corrigibility, the capacity to detect error and change course, and treats it as a primary design requirement (FC–01; EQ–11). But a civilization also has to bind itself. Constitutions, protected commons, inviolable rights, and obligations to people not yet born are all commitments meant to hold precisely when a present majority would prefer to break them, and all of them are anti-corrigible by design. A framework built only for changeability has no answer to the moment when the right thing is to make something unchangeable, and a framework built only for commitment becomes the dead hand of the past ruling the living. Aeviterna needs a principled line between the two.

The line has a literature. Jon Elster analyzed precommitment, the act of binding one's own future choices, in the image of Ulysses ordering his crew to tie him to the mast so he could hear the Sirens without steering onto the rocks. He is careful that the analogy only goes so far, since constitution-makers often bind others rather than only themselves, but the constructive core holds. A constitution can constrain a democracy in order to enable it, and the binding is what makes the freedom durable. Stephen Holmes put the same point as a paradox, that constraints can be enabling, the way the rules of a practice make the practice possible rather than merely limiting it.

Aeviterna's resolution follows directly. Lock the meta-rules and keep the object level liquid. The things a civilization binds permanently are few and structural. They are the right to revise itself, the core protections that revision must never strip, which are the cognitive sovereignty, the knowledge commons, and the dignity floor of EQ–06c, and the open procedures by which binding and unbinding happen. Everything below that, the actual policies and institutions, stays fully revisable. The one thing made permanent is the capacity to change everything else. This is self-binding in the service of staying unbound, and it is the only form of precommitment compatible with a corrigible civilization, because it entrenches the meta-rule of corrigibility itself while refusing to freeze any particular answer.

This is also how the framework hardens the standing it already grants to people who do not yet exist. EQ–06c derives the dignity of future persons and seats them through guardianship, with Wales' Future Generations Commissioner and the United Nations Declaration on Future Generations of 2024 as the working seeds. The interest of future generations is the textbook case of a commitment the present is structurally tempted to break, since the unborn cannot vote, lobby, or leave, and the costs of betraying them fall due after every current decision-maker has gone. The institutional record shows the failure to design out. Hungary's Ombudsman for Future Generations, under Sándor Fülöp, intervened against a water privatization and acted to protect the national gene bank, then was downgraded into a deputy office within about a term. Israel's Knesset commission was discontinued after a single term. An office the present can quietly gut is not a commitment, it is a courtesy. A future-standing body works only if it is locked at the meta-level, constitutionally entrenched with a mandate that decays and renews on a fixed schedule rather than at the pleasure of whoever it currently constrains, while its specific recommendations stay advisory and revisable. The standing is made permanent while the verdicts stay open to revision.

Aeviterna makes a claim about systems, that equilibrium is a kind of death and that a system stays alive by oscillating within bounds rather than settling at an extreme. That claim has to apply to the movement that carries the framework, or the framework refutes itself. Most movements treat institutionalization as the goal, the point at which spontaneous energy finally hardens into permanent structure. This one cannot, because its own thesis forbids settling. The movement is therefore designed to keep moving and to renew itself rather than to arrive.

The arc that movements usually follow is well documented and remarkably uniform. Ferment gives way to charismatic excitement, which gives way to formalization, which ends in a stable bureaucracy (Blumer; Weber). Vitality drains at three predictable points: the succession crisis when a founder departs; the bureaucratization that lets an organization optimize its own survival above its mission (Michels); and the mainstreaming that buys respectability at the price of fervor. For a movement built on ideas rather than on protest or party, the order of dangers is different. The deadliest is dogmatization, the moment core claims are made unfalsifiable and dissent is policed as heresy, because that forfeits the one thing an idea-movement has, which is the ability to be right for discoverable reasons (Popper). Close behind are purity spirals and schism, since ideas divide infinitely and status tends to flow to whoever seems most rigorous, and Goodhart capture, where a movement starts optimizing legible proxies such as followers or citations in place of the truth they were meant to track (Goodhart; Strathern).

Several of these failure modes are already answered elsewhere in the framework, because the properties that make a civilization adaptive are the properties that make a movement durable.

How the movement renews itself on purpose is the organ's harder half. The commitments below give the orientation; the instruments after them are specified from the documented record.

The right to fork now has a protocol, read from the one field that has practiced forking long enough to have re-merges on record. The healthiest fork in that record took a new name, wrote its re-merge condition into its founding statements, kept the commons interoperable, and merged back within a year when the parent conceded governance to a neutral venue; the second re-merge on record kept every contribution legally compatible with the incumbent throughout, and the fork became the project (the io.js and EGCS histories). The pathological pole is documented with equal precision. Splits in which both sides claimed the original identity produced no re-merge mechanics at all, and the measured aftermath, a minority chain suffering repeated majority attacks, the first fork splitting again within fifteen months, prices what an identity war costs. The sociology of movements agrees from its own data. Splits track renewal where the exiting faction builds a differentiated organization in a new niche, and track failure where it claims the whole identity (Gamalielsson and Lundell; Balser; Kretschmer). So a fork is legitimate when it takes its own name, files a divergence document at birth naming the falsifiable disagreement and the evidence that would re-merge it, keeps the commons interoperable so work flows both ways, and strips no one's standing. A split that claims the name, anathematizes the parent, or states no re-merge condition is a schism, and the judgment attaches to conduct, never to persons. Re-merge, when the register warrants it, follows the two-ingredient rule the record shows: a neutral venue, and a governance concession by the parent, pre-committed here so the parent cannot demand surrender.

The falsification register has a working precedent with measured results: registered reports, tests accepted before outcomes are known, now offered by over three hundred journals, showing 44 percent positive results against 96 in the standard literature, the cleanest demonstration anywhere that an unregistered record of claims is a biased register of what was tried (Chambers; Scheel, Schijen, and Lakens; Soderberg and colleagues). The framework's falsifiers, F–01 through F–16, become that register. Every load-bearing claim carries its would-be refuter from the day it ships, and contests between a fork and the parent run as adversarial collaborations with a pre-agreed arbiter, whose measured yield is convergence on the map rather than on minds, which the register promises and no more (Kahneman and Klein). Retirement gets the inverted default. Every structure and role carries a clock, and failing to muster affirmative renewal is itself retirement, preserved read-only in the archive, never acting while retired, the design that keeps apathy from breeding zombies, with the rubber-stamp counter-record priced in by giving every clock a named owner (the Attic process; the Texas sunset record). And renewal is procedural rather than aspirational. On a fixed cadence, a version of the framework cannot ship without retiring at least one mechanism, structure, or claim into the archive, improvement by subtraction on a schedule.

The same discipline turns one level further inward, onto the workshop that writes the framework. A single maintainer working with software agents selects the targets, drafts the papers, codes the instruments, and reviews the result, which is the concentration the framework warns every institution against, running the workshop itself. The record of comparable work names the danger precisely. Two of three popular open-source projects run on two or fewer load-bearing people (Avelino and colleagues), and the succession vacuum that leaves is an attack surface rather than a neutral fact, as the manufactured pressure campaign that installed the xz backdoor showed by naming an exhausted maintainer's successor as its own vector. The transitions that held did not name an heir. They wrote down a procedure and an open license, so anyone competent could continue the work by fork (the Python steering-council transition; the Linux kernel's continuity process; SQLite's public-domain plan to outlive its authors).

So the workshop binds itself to the same instruments the movement runs on, pointed at itself, and the workshop charter that results is one more structure held to the framework's law rather than exempt from it. Succession runs by fork under an explicit license (CC BY-SA for the corpus, MIT for the code), with a convening procedure and a conserved estate named in advance instead of an appointed heir. A degeneration condition is pre-registered, dated, and blind to its own outcome, on the clinical-trial logic that the party running the trial cannot be trusted to decide continue-or-stop after seeing the result (ICH E9). By a fixed date the program must show one outside adoption, one answered outside critic, and a forecast record no worse than chance, or it declares itself degenerating and archives its corpus as parts (F–09; Lakatos on progressive against degenerating programmes). The maintainers keep a scored forecast register, staking dated predictions on the vital signs and grading them by the Brier rule, the reality anchor of EQ–06e turned on the workshop's own framing (Brier; the Open Philanthropy prediction record). Review stays plural to the limit of the budget, run on separate agent instances with programmatic and primary-source checks, and the deeper limit is named rather than buried. One model family still mediates both drafting and review, and its defects propagate to everything adapted from it until a second family or a standing outside reviewer is funded (Bommasani and colleagues; Kleinberg and Raghavan on monoculture).

The same law reaches the corpus itself. If the framework degenerates by F–09 or simply never finds an adopter, it does not persist as a monument, because EQ–05b forbids that of any institution and the workshop that writes it is not exempt. It composts, into the three ends EQ–05b names for anything that dies well. Its parts transplant, the mechanisms that survived their falsifiers traveling into other frameworks under the open license, taken up one at a time by groups that never adopt the whole. Its record archives, the tracker, the priced balance sheet, and the falsifier register standing as evidence a later project can reuse, with the failures coded so they are not repeated. And its argument can still inform policy where no polity runs the stack, the way a defeated proposal shifts the ones that follow. What is preserved is the parts that earned their place, not the name. The only thing held back from sunset is the meta-core that guarantees this ending will happen, never the organization that wrote it. A framework whose whole claim is corrigibility has to be able to die corrigibly, and this names how.

Aeviterna does not claim to possess a final civilization design. It proposes a directional framework for building systems capable of surviving transition, resisting extraction, preserving autonomy, distributing stewardship, supporting adaptive stability, and preventing concentration from becoming permanent.

Aeviterna accepts imperfection, experimentation, partial adoption, and iterative refinement. It is not a destination. It is a direction and a set of design constraints for moving in that direction without catastrophic loss of the capacity to change course.

Each release has deepened and extended this direction rather than replaced it, from the first complete articulation through new civilizational organs, formal foundations and a place among existing traditions, the harder structures the core had left open, further frontiers, and an honest reckoning with the hardest critiques. The masthead dates each release, the version history carries the public record round by round, and the framework versions itself the way it asks institutions to, with every change owned. It is not a finished design but a coherent whole, stable enough to build on and to argue with, and it is expected to keep changing.

Before the mathematics, the whole framework fits in one picture. Aeviterna is a feedback regulator. It reads a small set of vital signs, compares them against the bands a civilization can survive within, and corrects through governance before a drift becomes a breach. What makes it Aeviternal rather than generic control is carried in the same loop, where the set point is human dignity, the actuator is the redistribution of power, and the controllers themselves decay so that authority lapses unless it is renewed.

The framework's central claim is dynamical. A civilization should stay where survivability can be preserved and move freely inside that region rather than hold to a fixed point. That claim has an exact mathematical home in viability theory (Aubin). A system whose state must remain inside a constraint set has a viability kernel, the largest set of states from which it is possible to remain inside indefinitely. The kernel permits unbounded oscillation and requires no equilibrium. The eight vital signs define the constraint set, and the kernel is the region this framework asks a civilization to keep within. Sustainability science already uses the kernel in this exact sense, as the set of states from which a future stays open (Béné, Doyen, and Gabay).

The governing result, with roots in Nagumo and a first rigorous form due to Haddad, gives the test for staying inside the set. At every point of the survivable region, at least one admissible direction of motion must point back inward. The framework's promise to redistribute recovery has a formal counterpart too. A capture basin is the set of states from which a target inside the region can still be reached without leaving it, which is the precise sense in which recovery remains possible after a shock, and the precise sense in which it can become impossible once a system has drifted too far.

Viability theory says where to stay; control theory says how to stay there. Set-invariance control studies the feedback that holds a system inside a safe set under bounded disturbance, which is the formal content of detecting instability and correcting in proportion to it (Blanchini). Regulating around shifting bands rather than holding a single set point is the difference between homeostasis and allostasis, stability achieved through change rather than at a fixed point (Sterling).

Cybernetics supplies the conditions under which any of this can be governed. Ashby's Law of Requisite Variety holds that a regulator can keep a system's essential variables within limits only if it commands at least as much variety as the disturbances it faces. This is the formal warrant for cognitive ecology. A governing system that loses internal diversity loses the capacity to regulate a varied world, which is why monoculture is dangerous and why variety counts against concentrating control (Ashby). Beer's Viable System Model gives the recursive, multi-level architecture that the governance modes inhabit (Beer).

Resilience science supplies the reading for everyday understanding. Holling's ecological resilience is the size of disturbance a system absorbs before it flips into a different basin, which is bounded oscillation within a basin stated plainly, and transformability is a deliberate move into a better basin, which is adaptive transition rather than failure (Holling; Walker, Holling, Carpenter, and Kinzig).

The commitments are not only arguments. Several rest on established empirical work, and saying which is part of being falsifiable. The knowledge commons is the clearest case. Elinor Ostrom showed that communities govern shared resources durably, without privatization or central control, when their institutions satisfy a set of design principles, and that knowledge itself behaves as a commons whose threat is enclosure rather than depletion (Ostrom; Hess and Ostrom). Aeviterna's governance reads as those principles applied at civilizational scale: clear boundaries and participatory rule-making, monitoring that keeps the commons auditable, graduated sanctions that prefigure the justice tiers, accessible conflict resolution, and nested enterprises that keep a large system polycentric so failures stay local.

What makes knowledge different also makes its governance more necessary, not less. Because understanding is non-rival, a knowledge commons is not threatened by depletion through use but by enclosure, gatekeeping, and underprovision, so openness alone does not protect it. It still needs the boundaries, monitoring, and recognized self-governance that any commons does, for different reasons than a fishery faces.

The vital-signs instrument rests on critical-transition science. As a complex system nears a tipping point it slows down, recovering more sluggishly from small perturbations, and this shows up as rising variance and autocorrelation before a regime shift, across very different systems regardless of mechanism (Scheffer and colleagues). That is why the framework watches a few load-bearing variables for loss of resilience instead of optimizing a single target. Driving one variable to its maximum usually narrows the basin and moves the system toward the edge, so a band with both a floor and a ceiling is the right object, not a maximum.

Cognitive sovereignty rests on the economics of attention. Attention is scarce, so a wealth of information creates a poverty of attention (Simon); that scarcity has been industrially contested for more than a century (Wu); and the contest now reaches perception and behavior themselves, shaped at scale by systems that profit from the shaping (Zuboff). Together these establish that perception is a contested civilizational resource, which is the thing cognitive sovereignty exists to protect.

Aeviterna is a new configuration of old primitives more than a set of new ones, and naming its neighbors is part of arguing in good faith. It shares with liberalism a protected sphere that no majority may vote away, expressed here as the inviolable dignity floor (Rawls; and on dignity as rank, Waldron, with the German Basic Law as the working model). It shares with republicanism the aim of freedom as non-domination, and answers it in time as well as in space. Decaying standing attacks entrenched power by letting authority lapse unless it is renewed (Pettit). It couples deliberative legitimacy to the governance modes, so reasoned and inclusive decision-making is the default and the modes decide when the gears change (Habermas). Its vital signs are a capability-style reading of real human functioning rather than output (Sen; Nussbaum). Its corridor between a social floor and an ecological ceiling is the safe and just space of degrowth and doughnut economics, generalized from throughput to governance itself (Hickel; Raworth). And its sensing-and-correcting loop is management cybernetics, with one difference that matters. The set point is dignity, the actuator is the redistribution of power, and the controllers themselves decay (Beer).

What is new is the combination: oscillation chosen over equilibrium, authority that decays by default, and a mode-switching structure bound to measured thresholds. Each is contestable, and together they are a recognizable new position rather than a relabeling of an old one.

The dignity floor deserves its own note, because it is the one place the framework refuses the usual calculus. Following the German Basic Law, dignity is treated as inviolable and non-balanceable rather than as a weighty interest to be traded off, the logic by which a constitutional court can strike down a measure that would sacrifice some lives to save others. That anti-balancing core is what keeps the rest of the adaptive machinery from optimizing a person away. The floor's absoluteness now attaches to an enumerated list of prohibited treatments with a burden-shift social minimum beneath it, the plural floor of EQ–06c, so what is held non-balanceable is specified rather than invoked, and the comparative record beyond one court carries it.

This is the framework's most important and least examined decision, and it deserves to be named as its own object rather than left as an exception to the banner. Call it the entrenchment boundary, the line between what is allowed to oscillate and the small set held fixed. Equilibrium is death describes the dynamics, but the boundary is the constitution, and the real moral work lives in where it is drawn. A framework whose whole identity is corrigibility has to be disciplined about its few non-corrigible commitments, so the test for admitting anything to the fixed side is strict. It must be minimal, command near-universal assent behind a veil of ignorance, and guard against a harm whose violation is irreversible. The dignity floor passes. Almost nothing else should, and whatever sits inside the boundary is itself interpreted, which is the problem taken up first in the Research Frontier as OQ–00.

The same machinery that makes the framework adaptive is its largest attack surface, because every threshold is a lever someone can pull. The strongest known failure modes, and what the design places against each, are set out below. The pattern across all of them is the same. Robustness comes less from the elegance of the loop than from what is held outside it as non-negotiable.

A framework that can name the ways it would fail has a better chance of not failing in those ways. That is the reason for writing them down, and the Model Charter turns these resistances into clauses a community can adopt, with the overhead band's clause owed to a coming charter version.

The deepest item in the failure-mode table in FC–03 is the apex, an anti-capture stack that climbs to a single seat no one watches. The same problem surfaces twice more in the framework, in EQ–04b as the question of who adjudicates the boundary without becoming the apex, and in EQ–06c as the risk of freezing a single interpretation of dignity. It is one problem, the interpretation of the non-derogable core, named in the Research Frontier as OQ–00, and this section is the framework's working attempt at it. The aim is not a system with no sovereign, which the precedents say is impossible, but a chair no one can settle into, an interpreter that is always one challenge from being overruled and one cycle from dissolving.

Two moves do most of the work. First, keep the core small and lit, since a non-derogable core that is tiny, explicit, published, and versioned is a worse capture target than a large vague one, because capturing it becomes visible and costly. Second, split the power to fix what the core means from the power to apply it. Its settled meaning is authored by no single seat and drifts only by a large, representative supermajority of the federation, on the model of peremptory norms, under a standing rule that no body may declare that meaning closed. A panel only ever renders an operative judgment for the case in front of it, binding for now and revisable after, with the losing reading kept in the open as legitimate.

The panel that applies the core is built to be the opposite of a sovereign, and it is assembled from the framework's own primitives. It is plural and never one. It is sortition-leavened, a lay majority drawn by lot with a minority of domain stewards whose standing already decays under EQ–05, so expertise informs the reading without ruling it, the same plural and decaying pattern EQ–03b uses for the epistemic monitor and EQ–06b uses for justice. This panel's constitution, generalized against the world's record of reader institutions, is now the framework-wide primitive, the reading seat (EQ–06e), and the framework's other instrument-readers instantiate it rather than rebuild it. Its terms are short and non-renewable, so there is no one to please for reappointment, and it publishes its reasoning and its dissent. It is not watched from above, because adding a watcher rebuilds the tower; the separated organs interpret the core against one another in a closed loop, with the right of exit in EQ–04b as the final discipline, the circle shown in the figure below.

The hardest residual is not the panel but the framing layer, whoever defines the vital signs and certifies which is the weakest, because that is the smallest and least watched seat. The design aims it back into the open: more than one independently produced methodology for each sign, so no single index decides; the methods published as forkable code; a right for any member to submit a competing reading and force a contest, at a cost to the original framers' standing if it loses; bounties for showing a reading was framed to a conclusion; and the framing seat itself filled by lot. The last danger is that a decaying guardian comes to believe only it can protect the core and resists its own sunset, the institutional form of an agent that fights its off-switch once it is certain, so the structure keeps it unsure it is needed. The mandate decays automatically and is renewed only from outside, a successor panel always stands ready, no panel is ever the sole interpreter, and incumbency carries little reward.

These are the framework's standing hard problems, carried forward by design rather than gaps to be closed before release. A living framework keeps a frontier; each problem below has a proposed direction and remains open to work.

Inclusion signals engagement, not endorsement.