VI. The Coming Trusts: Designing for the AI Age

The architecture follows the cognitive technology

Every architecture for sustained complex thought is bounded by the cognitive technology available at its founding, and the architecture reflects what that stack could do — not what later stacks would later enable. When the cognitive technology shifts, the architectural possibilities shift along with it, and architectures founded under the older stack often look constrained or even naive in retrospect. That is not because the founders were unsophisticated. It is because they were doing the best work the stack supported, with the materials a particular century put into their hands.¹

The framers who set down the school-trust system in 1785 were working at the dawn of the industrial-era cognitive-technology stack, and the tools they had to work with were the quill, the printed statute, the paper record, the mail moving by horse, and the fragile human memory in heads spread across a continent that did not yet have the telegraph or the railroad. Their architectural choices reflected what that technology could do, and what it could not. They could not engineer institutional memory directly, because the technology of the time required all memory to live in either heads or paper, both of them fragile, both of them expensive to share across generations and across the distances of a continent. So they reached for legal rigidity as a substitute for memory: perpetual trusts, irreducible funds, written constitutions, statutory rigidity, the compact form declared “forever” in its own terms. That was the best architecture available given the cognitive-technology stack of the late eighteenth century, and it is a serious achievement under those constraints. The parts of it that have held — the rectangular grid, the section-16 reservation, the federal compact form, the trust character of the grant itself — are remarkable for any institution at any scale, let alone for a 250-year experiment built with quills and statute books carried west by wagon.

What the architecture cannot do, however, is what its cognitive-technology substrate could not do. Memory in heads and on paper, even when reinforced by perpetual statutory language, depreciates across a continent and across a quarter-millennium. The legal rigidity holds; the institutional memory drifts; and the people who would have insisted on the architecture in operational practice — the chapter members, the deputy auditors, the standing crew of the watchful — dissipate faster than the rigidity can compensate for on its own. The drift was, in this sense, foreordained by the cognitive-technology stack of 1785, not by any failure of nerve at the founding. The framers built every kind of memory their century knew how to build. The kind of memory they did not build was the kind their century did not yet possess.

The architects of the AI Targeting Authority, the perpetual climate trust, the longevity escrow, the data trust, the compute reserve — designing institutions in 2026 — have access to a different stack. They have everything the 1785 framers had, plus institutional memory amplified by AI partnership, plus searchable archives at planet scale, plus the ability to engineer feedback loops across generations that the eighteenth century could not even contemplate. They can, in short, build memory directly into the institutional design as a complement to legal rigidity, not as a substitute for it. They can engineer the eighth anchor at the founding, rather than retrofit it after two and a half centuries of drift. They can do something the framers of 1785 could not engineer for, and that no perpetual structure has yet achieved, which is to put the standing constituency into the founding compact as a structural feature rather than as a hoped-for inheritance.²

This is the rhetorical posture the rest of Section VI takes, and it is the one the rest of this paper hopes the reader will hold throughout. The case is not “learn from us; we failed.” The case is “we did the best work the cognitive technology of 1785 supported; you have a richer technology; here is what it lets you build that we could not.” Those two registers are different in their consequences for the architects whose designs will determine the conditions under which AI-era beneficiaries inherit their world. The school-trust record is offered to those architects in the second spirit — as a working manual produced by a 240-year natural experiment, not as a confession of failure by a generation whose institutional ambition exceeded its tools. The mapping the school-trust evidence makes onto the AI-age structures is, with that posture set, exact.

The mapping is also four-axis. Each AI-age structure has a characteristic primary vulnerability among the four faces of drift named at the close of Section IV. An AI Targeting Authority is most vulnerable to purpose drift: today’s elite consensus about what counts as humanity’s most pressing problems will shift, and the words “most pressing” will be filled with new content under the same nominal mandate, often without anyone in the room noticing the substitution. A perpetual climate trust is most vulnerable to corpus drift: carbon as the trust unit, atmospheric stability as the asset, may not survive the century as economically and scientifically meaningful frames; a trust whose corpus is locked to today’s units of climate accounting will become obsolete the moment those units shift, and the structure will be left holding a measure that no longer measures anything. A data trust is most vulnerable to beneficiary drift: the data subject’s own identity is unstable, given changes in privacy law, in model architecture, in the rise of synthetic data, and the perimeter of “whose data is in the trust” can be redrawn faster than the trust’s enforcement community can react. A sovereign wealth fund is most vulnerable to legitimacy drift: citizens-of-the-polity is itself a moving definition, and a fund’s purpose is hostage to whichever generation’s politics happens to define citizenship at any given moment. The seven anchors apply to all five structures. The architectural extensions named at the close of IV.F — adversarial compact review, asset-translation authority, beneficiary-class boundary provisions — apply differentially, axis by axis, to each structure’s specific vulnerability.

What follows is, instrument by instrument, the design implications. The vocabulary in each subsection is borrowed deliberately from the school-trust record. Where a school-trust mechanism transfers, I name the source case — Ohio 1803, the 1889 Omnibus Act, New Mexico–Arizona 1910, Utah 1989–94, Nebraska’s LB 1072 of 2026, Oregon’s Elliott decoupling of 2017–22 — and explicitly translate the lesson. I have tried not to write “the architects should consider” anywhere in this section. The architects do not have time for considering. They have charters to draft. The right register, on the school-trust evidence, is the imperative.

AI Targeting Authorities

Peter Diamandis and Alex Wissner-Gross, in Solve Everything, propose an AI Targeting Authority — a perpetual, autonomous body empowered to direct compute, capital, and human attention toward humanity’s most pressing problems. The Authority would have its own resources, its own decision rules, its own legal personhood, and its own capacity to act across generations. Its operational concept, “Programmatic Down-Shifting,” is the method by which the Authority can reduce its own scope as conditions evolve.³

The proposal is, on the school-trust evidence, recognizably an aspirational structure with intergenerational beneficiaries, and the pathology applies to it as cleanly as it applies to a State Land Board chartered in 1859. The seven anchors apply, and they apply at the founding moment if they are going to apply at all. The discipline below works through them in the order Section V developed them, with the eighth anchor — the watchful crew — taken up at the close.

First anchor — codified fiduciary duties. The Authority’s charter has to declare, in terms equivalent to the 1910 New Mexico–Arizona “breach of trust” language, that any diversion of the Authority’s resources from beneficiary-mission purposes is a breach of trust, voidable in court, and recoverable by independent enforcement.⁴ That is not a translation difficulty; it is a drafting difficulty, and it is the drafting move the school-trust framers got right in 1910 and got wrong in California in 1850. The lesson the comparative state record teaches is that “breach of trust” in the founding text is the doctrinal floor on which every later enforcement action stands. Without it, courts in 1876 (Sherman v. Buick), 1867 (Wyman v. Banvard), and 1914 (Alabama v. Schmidt) read the absence of the language as decisive: an “absolute and unconditional grant,” not a trust. California’s 8–10 percent retention rate of its school-grant lands, against Arizona’s 75 percent and Montana’s 85 percent, is what happens when the founding text does not name the duty.

The Authority’s equivalent: the charter must declare that compute allocations, capital deployments, and attention-directing grants made by the Authority outside the beneficiary-mission purpose are breaches of trust; that they are voidable; that the assets so deployed are recoverable; that any officer of the Authority who participates in the breach is personally liable. The duty must run to the beneficiary class — defined as the future humans whose welfare the Authority’s grants are intended to advance — and not to any of the Authority’s other stakeholders, however well-intentioned those stakeholders may be in any given decade. The Ohio 1803 for no other use, intent or purpose whatever clause is the period’s strongest statement of that closure; the Authority’s charter should contain a 21st-century equivalent in operationally specific terms. No compute hour, capital dollar, or attention-directing grant disbursed by the Authority shall be applied to any purpose other than the long-term welfare of the beneficiary class identified in this charter, nor for any other use, intent, or purpose whatever. Anything weaker than that and the doctrinal floor is missing.

Second anchor — asset restoration. Aspirational structures lose ground in increments. A small diversion here; a transfer to an aligned-with-mission affiliate there; a temporary reallocation that becomes permanent across an audit cycle; a percentage-point drop in the disbursement rate. Over decades, the increments compound. The 1889 Omnibus Enabling Act introduced the null and void clause precisely for this reason: any disposition of school-trust lands outside the granted purposes was nullified by federal text, which gave the trust a doctrinal hook to convert breach into recovery.⁴ Utah’s 1998 Schools and Lands Exchange Act is the cleanest contemporary recovery instance: 377,000 trust acres trapped within the 1996 Grand Staircase–Escalante National Monument were exchanged out for $50 million in cash and equivalent federal lands, converting what was political loss into trust gain. The mechanism worked because the underlying breach was nullifiable on the founding text.

The Authority’s equivalent: every grant cycle’s disbursements should be auditable against the founding text; any disbursement found to have violated the beneficiary-mission purpose should be nullified by operation of charter, with the resources recoverable to the corpus through an independent recovery action. The doctrinal point — that nullification operates by operation of the founding text rather than by political reversal — is the point. Political reversal is contingent. Nullification by operation of charter is structural. The school trust’s recoveries in Utah, in New Mexico, and now in active Oregon litigation have all turned on whether the founding text supplies the nullification hook. In California, where the federal grant of 1853 lacked the for no other use, intent or purpose whatever clause, the recoveries did not occur, and the corpus has run down to one-tenth of its original size. The Authority’s charter cannot afford the California omission.

Third anchor — individual and institutional accountability. The school-trust record’s hardest lesson on accountability is double. Boards have a way of dissolving responsibility — a vote is taken; the consequences are diffuse; individual trustees later report that they acted as the board acted. The trust-law tradition, older than the modern administrative board, holds that a trustee who participates in a breach is personally liable for the breach. Modern public-trust statutes have, almost without exception, blurred the rule. Margaret Bird’s correction, dated April 29, 2026, sharpened the framing: individual trustees are personally liable, and the State itself is institutionally liable. The two halves close the institutional escape hatch that diffused responsibility otherwise opens.

The Authority’s equivalent: every officer, board member, and trustee of the Authority is personally liable, by operation of charter, for any act treating Authority assets as non-Authority assets. That is the individual half. The institutional half: the Authority itself, as an entity, is liable for any institutional drift the individual votes ratify or produce — a drift the Authority owes its beneficiary class restitution for, regardless of which officer’s name is on which 2042 minute. Calling individual officers to account is a forensic question. Calling the Authority to account is a fiduciary one. Both are required, and the architecture must hold both available. The school-trust record contains, in Oregon’s post-1990 dissipation, the cleanest available demonstration of how diffused responsibility produces a forty-year drift that no individual officer’s vote, in isolation, can be held to account for. The Authority’s charter has to refuse both escape routes: officers personally liable, the Authority institutionally liable, the beneficiary class owed restoration from whichever party the recovery can be obtained from.

Fourth anchor — direct distribution. The Utah model that Margaret Bird’s work has documented at length is the cleanest available demonstration of this anchor in operation. Under the Utah School LAND Trust Program, established in 1999, a portion of the trust’s earnings flows on a per-pupil basis directly to individual public schools — visible to teachers, parents, and principals at the building level, governed by a School Community Council at each school that by statute must have a parent majority and that adopts an annual academic plan. The visibility produces an enforcement constituency at scale. When a Utah governor moves to raid the fund, the political response comes not from a single advocacy office but from every school in the state whose annual disbursement is on the table.

The Authority’s equivalent is harder to specify and more important to get right. The beneficiary class — future humans — does not receive grants in any direct sense; the Authority directs compute, capital, and attention toward problems whose solutions the future will inherit. What can be made visible is the grant-by-grant decision record itself. The Authority’s grant decisions must be visible to its stakeholders in a form analogous to Utah’s classroom-by-classroom transparency, which means: every grant, named and dated; every decision rule that produced the grant, traceable; every dissent recorded; every grant’s downstream consequences, audited at intervals the charter specifies. If a grant of compute to a longevity-research consortium is the Authority’s equivalent of a section-16 disbursement, then the population of practitioners and researchers and their broader civic communities must be able to see it, scrutinize it, and argue with it. The visibility is the prerequisite for the distributed defensive perimeter the seven anchors together compose. Without visibility, no constituency forms; without a constituency, the architecture, however well drafted, is decorative.

The wallet metaphor matters here as it matters in the school trust. In states without direct distribution, the schools are not beneficiaries in any operational sense; they are payees. The state hands the schools their own money back from the general fund, and calls the transaction a distribution. That is what handing a wallet to its rightful owner, after first removing what was inside the wallet, looks like; it is not what a fiduciary distribution looks like. The Authority’s equivalent failure mode is the closed-doors grant cycle that publishes only summary metrics — $X disbursed across Y programs — without grant-by-grant accountability. That is handing the wallet back without the contents. The Authority must publish the contents.

Fifth anchor — independent beneficiary advocacy. Utah’s Land Trusts Protection and Advocacy Office, established in 2018 under Title 53D Chapter 2, is the clearest operational instance of this anchor. The office is statutorily separate from the management agency. It has its own director, its own budget, and its own mandate: it advocates for the beneficiary against the competing pressures of the state treasury, the legislature, and other state agencies whose interests sometimes conflict with the trust’s. Section 53D-2-201 expressly directs the office to represent beneficiary interests across land management, fund investment, and distributions, and to advocate against state uses of trust assets that conflict with trust purposes.

The Authority’s equivalent: an Office of the Future-Beneficiary Advocate, statutorily separate from the Authority’s management, with its own director, its own appropriation, and its own mandate to argue against any of the Authority’s deployments that the Office judges inconsistent with the long-term welfare of the future-human beneficiary class. The 2055 humans whose welfare is in the Authority’s portfolio cannot show up in 2030 to argue their own case. Someone has to show up for them, and that someone has to draw a paycheck from a different appropriation than the Authority’s operating budget. The Office’s directorship should be a fixed-term appointment with statutory protection against removal except for cause, on the model the Utah LTPAO uses. The Office’s annual report should be published independently of the Authority’s, and should be required by charter to contain a section enumerating the Authority’s deployments the Office has flagged as inconsistent with beneficiary welfare during the reporting year. The Office’s job, formally and visibly, is to argue against the Authority on the beneficiary’s behalf. Without that structural separation, the advocacy collapses into another function of management, and the future-beneficiary class loses its only voice in the room.

Sixth anchor — specialized legal defense. State Attorneys General, as the school-trust record demonstrates over and over, are politically appointed officers whose duties run, in most states, to multiple state clients with conflicting interests. When the trust’s interests conflict with the state’s general-fund interests, the Attorney General is structurally compromised. The 1910 New Mexico–Arizona Enabling Act got this right by authorizing enforcement by the Attorney General of the United States as well as by the state and by individual citizens — building plural enforcement directly into the federal text.

The Authority’s equivalent: when the Authority’s interests conflict with the interests of the entities that fund it, host its compute, or supply its political legitimacy — and they will, on the historical evidence, conflict early and often — the Authority must have its own counsel, selected by the Office of the Future-Beneficiary Advocate, paid by the Authority’s own resources, accountable solely to the trust. The plural-enforcement principle should also be present: charter language should authorize enforcement by the Authority’s counsel, by independent member-state attorneys-general acting as third-party enforcers, and by individual citizens with statutory standing to sue on behalf of the future-beneficiary class. The redundancy is the design feature. Redundancy lets one path lie dormant while another path carries the enforcement work, which is roughly what has happened across the school-trust record’s last century. If the Authority’s architecture has only one enforcement path, that path will eventually be captured, and the Authority will have written an institution that cannot enforce itself.

Seventh anchor — mandatory fiduciary education. The State Land Boards in most public-land states are composed of officials whose primary professional formation is political rather than fiduciary. They serve as trustees because the constitution or the enabling statute names them as trustees. Most have no specific training in trust law, the rule of undivided loyalty, the prudent-investor standard, or the operational meaning of the sole-interest rule. The seventh anchor is the cognitive recalibration the political mind otherwise misses.

The Authority’s governance committee will be composed, predominantly, of officials whose primary professional formation is technical or political — AI researchers, philanthropic executives, government appointees. Annual training on the fiduciary duties of trust officers — taught by independent practitioners, not by the Authority’s own legal staff — should be a charter-mandated condition of remaining in office. The training is the cognitive recalibration the role requires, because the trustee disposition is not the natural disposition of the technologist or the elected official, and the difference has to be taught and re-taught at intervals that match the pace at which it is forgotten. The training’s content should include the school-trust record itself, treated as a worked case-study of what fiduciary failure across two centuries actually looks like.

Eighth anchor — the watchful crew. This is where the design problem is hardest. The school trust acquired its enforcement community late and only partially, after two centuries of drift, and the Authority would need its enforcement community in place before the first grant cycle.

Diamandis and Wissner-Gross’s concept of “Programmatic Down-Shifting” implicitly recognizes the problem; what it does not yet specify is who has the authority to do the down-shifting, on whose behalf, with what enforcement mechanism, paid out of which line item. The school-trust evidence answers part of that question by elimination: the answer cannot be the Authority’s own staff. The answer has to be a community external to the Authority, structurally invested in the Authority’s compliance, and resourced enough to keep showing up for as long as the Authority itself exists. The Authority’s charter should include, by name, the community-of-practice institutions the founding generation expects to carry the watchful-crew function: an academic-and-policy infrastructure on the model of the Utah Foundation’s school-trust-program work; an issue-organized public-interest non-profit with statutory standing on the model of the Advocates for School Trust Lands family; a Library — institutional locus, public substrate, multi-generational continuity — on the model of America’s School Trust Library at schooltrusts.net. The Library is the existence proof: a 2026 institution can be built, at workable cost, that holds the kind of multi-generational civic memory the framers of 1785 could not engineer for. The Authority’s charter can name a successor institution as the watchful-crew locus from day one, fund it from the Authority’s corpus on a statutorily protected appropriation, and insulate it from removal except for cause.

That is the eight-anchor mapping, in operational form. Every anchor has a school-trust precedent, every anchor has a translation move, and every translation can be specified at charter-drafting depth. The architects have the vocabulary they need.

Perpetual climate trusts

Climate trusts under construction in 2026 — the New Zealand Climate Change Commission; the proposed UN climate trust funds; the various national sovereign carbon funds; the European Green Deal’s intergenerational-equity provisions — face the structural problem the school trust faced. The beneficiary class is the silent class. The institution administering the trust will, over a century, optimize for its own continuity. The cognitive bias of the people running it will favor the present, because that is the cognitive bias of the human animal regardless of which century the human happens to be working in.

The international-law literature on intergenerational equity has been working the same problem the school trust embodies, in a different domain, for more than thirty-five years. Edith Brown Weiss’s 1989 In Fairness to Future Generations is the canonical text.⁵ The legal-philosophical question Weiss poses — whether and how a polity can incur enforceable obligations to a beneficiary class not yet alive — is the same question the 1785 framers answered in their own period vocabulary by reaching for the charitable-trust form. Weiss’s answer is similar in shape: a trust-like fiduciary obligation, supplemented by representational mechanisms (a future-generations ombudsman; standing for future-generation advocates), all operating under explicit principles of intergenerational equity. The Pope’s 2015 encyclical Laudato Si’ gives the climate trust a moral vocabulary the school trust acquired only late in its own history.⁶ What both still lack, and what the school-trust record can supply, is the operational architecture — and what that architecture’s actual performance shows about the limits of formal language alone, however eloquent the formal language happens to be.

The climate trust’s primary vulnerability, named at the head of this section, is corpus drift. Carbon as the trust unit, atmospheric stability as the asset, may not survive the century as economically and scientifically meaningful frames. A trust whose corpus is locked to today’s units of climate accounting will become obsolete the moment those units shift, and the structure will be left holding a measure that no longer measures anything. The school trust’s response to the corpus-drift problem was the geographic embedding of the asset itself: section sixteen, on the plat, before the township was settled. The schoolhouse parcel was a particular piece of a particular place, and it could not be redenominated without visible alteration of the survey itself. Geographic embedding does not transfer cleanly to atmospheric carbon, but the principle — that the corpus must be defined in terms that resist political redenomination — does.

The design implications, anchor by anchor:

Codified duties. The 1910 New Mexico–Arizona breach-of-trust language, scaled up to international compact: any redirection of a national or international climate trust’s corpus from emissions-reduction or atmospheric-stability purposes is a breach of trust, voidable and recoverable. The phrase for no other use, intent or purpose whatever should appear in the climate-trust charter in updated language that names the specific deployments the founding generation intends — emissions reduction, adaptation infrastructure, intergenerational equity disbursements — and excludes by operation of charter any other deployment.

Asset restoration. Where carbon budgets are exceeded, recovery mechanisms must exist, and exist as a matter of binding instrument rather than political aspiration. The 1889 Omnibus Act’s null-and-void clause is the textual model: every overshoot creates a legally cognizable claim against the trust’s deployable resources, recoverable to the climate-protection corpus through independent enforcement. The trust must be designed so the overshoot triggers the recovery; political responsiveness to the overshoot is the school trust’s California failure mode, not its New Mexico success mode.

Individual and institutional accountability. Board members of climate trusts personally liable for breach in the same sense charitable-trust trustees are at common law; the trust itself institutionally liable for institutional drift its individual votes produced. The Margaret-Bird sharpening of Anchor 3 — both halves required — applies to climate trusts with the same force it applies to school trusts.

Direct distribution. The climate trust’s outputs visible to local stakeholders, in the form of climate-resilience disbursements analogous to Utah’s school-by-school distribution. Where climate-trust funds are routed to a centralized national agency for nation-state-level disbursement, the political defense the architecture will need a generation from now is structurally absent. Where climate-trust funds are routed to local watershed councils, to municipal climate-adaptation committees, to indigenous-nation governing bodies, to regional resilience boards — every disbursement at the local level builds a constituency at the local level, on Margaret Bird’s mechanism: increase the consequential decisions reaching the local sites where deliberation already happens, and the constituency forms around those sites. The principle generalizes from schools to watersheds, from teachers and parents and principals to the local civic infrastructure that already exists wherever climate adaptation is consequential.

Independent advocacy. A future-generations office, structurally separate from the trust’s management, drawing on a separate appropriation, with statutory standing to litigate against the trust on the beneficiary’s behalf. Weiss’s ombudsman is the operational form. The climate-trust ombudsman should publish an annual report enumerating the trust’s deployments inconsistent with intergenerational equity during the reporting year, and should have standing to sue the trust’s management in international or domestic forum.

Specialized legal defense. The trust’s own counsel, independent of the political branches of the parent state. Plural enforcement: by the trust’s counsel, by international intergenerational-equity standing, by domestic citizen-suit provisions. The redundancy is the design feature.

Mandatory fiduciary education. Annual training for the trust’s governing officials, year in and year out for as long as the trust exists, taught by independent practitioners on the school-trust and intergenerational-equity records.

The watchful crew. The climate trust’s watchful-crew problem is the global analog of the school trust’s. The school trust’s recovery, where it has occurred, has come through state-level non-profit infrastructure with national reinforcement. The climate trust will need analogous infrastructure at international scale: a community of practice across nations, sustained by institutional infrastructure with its own continuity, holding the trust’s substrate as a public commons, capable of refusing to let drift compound across cycles. The Library at schooltrusts.net is the worked example at one nation’s scale; the climate-trust analogue would be a Library at international scale, built into the founding compact rather than retrofitted. The architects of the climate trust who do not name their watchful-crew institution at the founding will discover, in the 2080s, that the founding text alone is not enforcing itself. The school-trust record is unambiguous on this point.

Sovereign wealth for the unborn

Sovereign wealth funds chartered for the benefit of citizens not yet born — Norway’s Government Pension Fund Global as the resilient case, the various Gulf-state funds and proposed African and Latin American funds as the variable cases — face a vulnerability the school trust faced in compressed form: legitimacy drift. Citizens-of-the-polity is itself a moving definition, and a fund’s purpose is hostage to whichever generation’s politics happens to define citizenship at any given moment.

The Norwegian fund is the cleanest contemporary working example of an architecture consistent with the seven-anchor formulation: a transparent management entity statutorily separated from the political branches; explicit ethical guidelines codified in the Ministry of Finance’s Guidelines for Observation and Exclusion; an external advisory council — the Council on Ethics — with independent standing; codified duties to the beneficiary population; and a culture of professional fiduciary identity at the management entity that has, over the fund’s three decades of existence, resisted political diversion through several governments of different parties. The fund’s size, its transparency, and its institutional separation from the political branches are the structural defenses that have held.⁷

The school-trust record predicts that the variation in sovereign-wealth-fund architectures will track the variation in their long-run performance, fund by fund and decade by decade. Funds with strong codified duties, independent advocacy, and active enforcement constituencies will hold their ground. Funds without will drift on the predicted timeline; and the drift will not announce itself as drift while it is happening, because drift never does.

The drift mode here is specific. The fund’s earnings get redirected to current operations; the corpus is “preserved” but the income is captured. Oregon’s Common School Fund record, as documented in the companion volume’s Chapter 5, is the school-trust analogue: the principal preserved (and even modestly grown) through the period 1990–2017, while the income’s direction and adequacy drifted away from the beneficiary’s interest. The structural defense is irreducibility clauses tested against political pressure across multiple cycles, and the test the school-trust record offers is unambiguous: an irreducibility clause untested by political pressure is decorative; an irreducibility clause that has held against three or four serious pressure events across multiple political generations is structural.

The design implications:

Codified duties. The fund’s charter declares that the corpus is held in trust for the population, present and future, of citizens-as-defined-at-the-founding-and-as-amended-by-supermajority-only. The Oregon Constitution’s Article VIII, Section 2 separate, and irreducible fund language is the period’s strongest single-state statement; the sovereign-wealth-fund equivalent should appear in the founding constitutional or charter document, not only in implementing legislation that the same legislature can amend. Nebraska’s 2026 LB 1072, which constitutionalized state-level school-trust protections after a 1965 fiduciary-duty erosion, is the contemporary template for what testing-against-pressure produces when the architecture survives the test.

Asset restoration. Any redirection of the corpus from the founding purpose creates a recoverable claim against the redirecting political body. Norwegian-fund precedent: when proposed redirections have arisen, the institutional separation between Norges Bank Investment Management and the political branches has functioned as the de facto recovery mechanism — the fund’s management refuses the redirection, citing the founding text. The structural defense is the institutional separation; the doctrinal hook is the for no other use equivalent in the founding text.

Individual and institutional accountability. Officials who vote for redirections personally liable; the state institutionally liable. The Oregon Common School Fund record demonstrates that institutional liability outlasts individual liability across sixty-year cycles, and that the recovery doctrine has to name both halves to be operative.

Direct distribution. This is the anchor most easily missed in sovereign-wealth-fund design. The Norwegian fund disburses to the national budget; the disbursement is, in the school-trust vocabulary, a payee transaction rather than a beneficiary transaction. The 4 percent rule — the fund’s disbursement limit — has held under political pressure, but the disbursement is not visible to local stakeholders in any operational sense. The architectural extension the school-trust record suggests: at least a portion of the fund’s disbursement should reach local civic infrastructure on a per-municipality or per-region basis, visible to the local population, governed by local councils with statutory mandates that mirror the Utah School Community Council’s parent-majority requirement. The constituency forms around the local visibility. Without it, the fund is defended by the management entity’s professional culture alone — a culture that, as Section IV’s eusocial-pathology analysis warns, can drift over generational time even where the formal architecture is strong.

Independent advocacy. A future-citizens advocate’s office, statutorily separate, with the mandate to argue against the present-day political process on behalf of citizens not yet born. The Council on Ethics is the precursor; the future-citizens advocate is the next step.

Specialized legal defense. The fund’s own counsel, plural enforcement paths, citizen standing to litigate in defense of the corpus on the beneficiary class’s behalf.

Mandatory fiduciary education. For the management entity’s officers and the political-branch officials who oversee them.

The watchful crew. The fund’s watchful-crew institution should be named at the founding: a public-substrate Library on the schooltrusts.net model, capable of holding the fund’s institutional memory across generational cycles, with statutory protection of its appropriation and its independence. The Norwegian fund has not yet built this; its watchful-crew function is presently distributed across journalism, academic monitoring, and the Council on Ethics. The school-trust record predicts that distributed monitoring without an institutional locus is fragile across multiple political generations. The architects of new sovereign-wealth funds should build the locus from day one.

Longevity escrows

Longevity escrows — programmable trust instruments scoped to outlive their founders by decades, dedicated to research and treatment access for beneficiaries on fifty- and hundred-year horizons — have the perfect-host beneficiary problem in clinically pure form. The hundred-year beneficiary cannot sue, vote, or testify, and the present-day institution can therefore do whatever it likes with the beneficiary’s nominal share.

The drift mode is specific: the escrow’s purpose is reframed by successor trustees; the founder’s intent dilutes; the original “longevity research and access for the founder’s descendants and the broader population” mandate becomes “biomedical research broadly defined” becomes “health-related deployments at trustee discretion” becomes a general charitable fund whose connection to the founder’s intent is rhetorical only. The same trajectory has played out in the school-trust record, in compressed and elaborated forms both. New Jersey’s perpetual school fund, constitutionally vested with an anti-borrowing provision phrased as strongly as anything in the country, has nonetheless been amended by amendments to the same constitution. Mississippi’s sixteenth sections drifted from “for the maintenance of public schools” toward “for state purposes broadly construed” across a generation of legislative reinterpretation. The pattern is the same: founding intent diluted by successor reinterpretation, in the absence of a structural defense that refuses the reinterpretation by operation of text.

The structural defense the school-trust record names is explicit purpose-binding language with breach-of-trust nullification. The 1803 Ohio implementing act’s for no other use, intent or purpose whatever clause is the period’s strongest such language; the 1910 New Mexico–Arizona Act elaborates it into express trust language with breach-of-trust nullification. The longevity-escrow analogue: the founding instrument should declare the purpose in operationally specific terms, name the deployments the founder intends (research; treatment access; specific population categories), and declare any deployment outside those purposes a breach of trust voidable by operation of the instrument itself.

The design implications, anchor by anchor:

Codified duties. The escrow instrument names the duty as fiduciary, the standard as the punctilio of an honor the most sensitive, the rule as the sole-interest rule. Modern professional-trust statutes have, almost without exception, blurred this. The longevity-escrow instrument can refuse the blur: undivided loyalty to the named beneficiary class, with no political-discretion exception, on the model of Utah Code § 53C-1-102.

Asset restoration. Any disbursement found to violate the founding purpose is voidable and the assets recoverable. The 1889 Omnibus Act null-and-void model.

Individual and institutional accountability. Successor trustees personally liable; the escrow institutionally liable for institutional drift. The board cannot dissolve responsibility into procedural diffusion.

Direct distribution. The escrow’s deployments visible to a community external to the escrow. The hundred-year beneficiary cannot show up; the people who can show up are the descendants of the founder, the affected medical practitioner community, the broader population of beneficiaries-by-class. The escrow’s grants should be visible to those communities at grant-by-grant resolution, not only in summary metrics.

Independent advocacy. A future-beneficiary advocate, statutorily separate from the escrow’s management, with standing to litigate against the management on the beneficiary’s behalf.

Specialized legal defense. The escrow’s own counsel, plural enforcement paths, descendant or class-member citizen standing to litigate in defense of the founding intent.

Mandatory fiduciary education. For successor trustees, on the school-trust record and the longevity-research record both.

The watchful crew. The longevity-escrow’s watchful-crew institution should be named at the founding: a research-and-policy substrate analogous to the Library, capable of holding the escrow’s institutional memory across the founder’s intended time horizon, with statutory protection. The Library at schooltrusts.net is the worked example for civic memory at one nation’s scale; the longevity-escrow analog is the founder’s institutional successor — a research foundation, a public-substrate library, an academic chair endowed with the fiduciary obligation to monitor the escrow’s compliance — named at founding, funded from the escrow’s corpus on a protected appropriation, insulated from removal except for cause.

The longevity escrow’s particular acceleration is that the time horizon is short enough — the founder’s lifetime plus fifty to a hundred years — that the escrow’s drift can be observed by the founder’s grandchildren. The school-trust record’s two-and-a-quarter-century time scale is what makes the school trust’s drift hard to observe in any single generation; the longevity escrow’s compressed time scale means the drift, when it occurs, is observable. That visibility is an architectural opportunity. The founder can specify, in the instrument, that the founder’s grandchildren and the broader beneficiary class shall have standing to litigate any deviation from the founding intent — a citizen-suit provision on the 1910 New Mexico–Arizona model, but with the citizens being the founder’s descendants and the affected class members, and the time scale being short enough that the original founding intent is still legible to the people standing.

Compute reserves and data trusts

The smaller-scale aspirational structures emerging in the AI economy share the same pattern at smaller stakes. Compute reserves — proposals to hold AI infrastructure for public-benefit purposes on perpetual horizons — face the school-trust pathology in compressed form. Data trusts — entities holding ownership of consumer datasets across changes in corporate ownership — have the eusocial-drift problem in compressed timeline, because the data-trust bureaucracy can drift in a five-year corporate-acquisition cycle the way the school-trust bureaucracy drifted in a fifty-year political cycle.

The school-trust record’s geometric embedding — the schoolhouse parcel on the plat before the township is settled — does not transfer cleanly to compute and data, which are non-physical, non-geographic, non-surveyed. What does transfer is purpose-binding, beneficiary protection, and the watchful crew’s standing. Land was a particular kind of asset; trusts holding non-physical assets have to engineer durability through different means.

The data-trust vulnerability, named at the head of this section, is beneficiary drift. The data subject’s own identity is unstable. Privacy law evolves; model architectures change; synthetic-data substitutes blur the line between “data the trust holds for the subject” and “data the trust holds like the subject’s data”; the perimeter of “whose data is in the trust” can be redrawn faster than the trust’s enforcement community can react. A consumer dataset transferred from one corporate owner to another in a 2030 acquisition may carry obligations the trust, on its founding text, was supposed to enforce — and may have those obligations quietly redefined in the implementing documents of the acquisition itself, with the trust’s nominal continuity preserved but its operational meaning hollowed.

The structural defense is explicit beneficiary-class boundary provisions. The school-trust record’s California omission case is the cleanest counter-case: where the founding text fails to specify the beneficiary class with enough particularity that the courts can enforce its boundaries, the boundaries dissolve in the first generation of contestation. The 1910 New Mexico–Arizona Act’s specification of the public schools of the state as the beneficiary class, with breach-of-trust nullification of any disposition outside the granted purposes, is the period’s strongest beneficiary-class boundary. The data-trust analogue: the founding instrument names the beneficiary class — the data subjects whose data is held by this trust as of the founding date and any subjects added by the procedure specified in §X — with enough particularity that any redefinition of the class requires charter amendment by the procedure specified for amending the founding text, not by the implementing documents of any subsequent corporate transaction.

The design implications track the seven anchors. The compute-reserve and data-trust architects should:

• Codify duties (Anchor 1) in operationally specific terms — for the operational benefit of the named beneficiary class and for no other use, intent or purpose whatever. The 1803 Ohio clause, applied to compute allocation and data custody.
• Build asset-restoration mechanisms (Anchor 2) — overflows of compute or data outside the granted purposes nullified by operation of charter, recoverable to the trust through independent enforcement. The 1889 Omnibus Act null-and-void model, applied to non-physical assets.
• Install individual and institutional accountability (Anchor 3) — officers personally liable, the trust institutionally liable, both halves required.
• Make deployment visible (Anchor 4) — every compute allocation and every data-grant visible at grant-by-grant resolution, not only in summary metrics. The Utah classroom-by-classroom transparency, applied to compute.
• Establish independent advocacy (Anchor 5) — an Office of the Data-Subject Advocate or Office of the Compute-Beneficiary Advocate, statutorily separate, with standing to argue against management.
• Provide specialized legal defense (Anchor 6) — the trust’s own counsel, plural enforcement paths.
• Mandate fiduciary education (Anchor 7) — for the trust’s officers and the political bodies that oversee them.
• Name the watchful-crew institution (Anchor 8) — at founding, funded from the trust’s resources, insulated from removal.

The acceleration in this domain is the five-year corporate-acquisition cycle. The school trust’s drift compounded over fifty years; the data trust’s drift can compound over five. The architects must compress the seven-anchor cycle accordingly. Annual fiduciary education becomes quarterly review; annual independent-advocate reports become continuously published; annual public-substrate updates become real-time. The Library at schooltrusts.net updates its substrate monthly; the data-trust analogue may need to update weekly. The cognitive-technology stack of 2026 supports the higher cadence; the architects should build for it.

The compounding feature

The compounding feature of these structures is that they are not sequenced. They are emerging in parallel, in the same five-year window, with the same underlying technological substrate, and often with the same handful of architects working across more than one of them at a time. The aggregate institutional inheritance of the AI moment will, by 2030, include several dozen aspirational structures with intergenerational beneficiaries, varying widely in the strength of their formal architectures and in the presence of their enforcement constituencies. The school-trust evidence predicts that the variation in their thirty-year performance will track the variation in their starting conditions, the way the variation in the fifty states’ performance has tracked the variation in their starting conditions across the past two centuries.

That parallel emergence is also an opportunity. If the architects of the AI Targeting Authority, the perpetual climate trust, the sovereign wealth fund for the unborn, the longevity escrow, the data trust, and the compute reserve are working in the same five-year window with overlapping legal and policy talent, they can converge on a common architectural vocabulary at low marginal cost. The vocabulary is on offer in this paper: seven anchors, plus a watchful crew, with school-trust provenance for every move and a body of two-and-a-quarter-century evidence behind the diagnoses. The cost of installing the architecture once, in a charter being drafted now, is a small fraction of the cost of retrofitting it after the drift has begun. The school trust’s retrofit cost — measured in the staffing of advocacy offices, the litigation of breach-of-trust cases, the decade-long process of recovering corpus diverted in the previous decade, the reconstruction of institutional memory in every generation — is the cost the architects can avoid by installing the architecture at founding.

The window is open now. It will not stay open. The charters being drafted in 2026 will harden, on the school-trust evidence, into the form they will keep across the generations the structures are intended to bind. The architects in the room are the architects whose decisions will determine whether the institutions they are building hold across the time horizons their charters set. The school-trust evidence is on the desk now, in the form of this volume and the companion record. The architects have what they need.

The live cases: PERS and Social Security

Two structures already in operation in the United States are domestic exhibits of the same pathology under way in real time, on American ground, with American officials and American beneficiaries. They are the empirical bridge between the school-trust record’s nineteenth-century origins and the AI-era structures the rest of this section has named.

The Oregon Public Employees Retirement System (PERS) is, in the technical sense, a defined-benefit pension fund. In the structural sense, it is an aspirational structure with intergenerational beneficiaries — current public employees whose retirements depend on contributions and investment returns extending over a sixty- or seventy-year working-life-plus-retirement horizon. PERS’s funded ratio has drifted in ways that match the school-trust pattern: near-term political pressure (legislative reluctance to fund the actuarially required contribution) compounded by hyperbolic discounting (the future obligation is steeply discounted by the legislators making the decision in any given session) and beneficiary-side political weakness (public employees, while not voiceless, do not generate the institutional feedback that would correct the drift in real time, because the feedback that would correct the drift would have to come from teachers and bus drivers and child-welfare caseworkers who are too busy doing their jobs to maintain a parallel actuarial watch). The result is the predicted trajectory.⁸

Social Security operates on the largest stakes the country has. It is the most consequential aspirational structure the United States has built since the school trust, by orders of magnitude, and it faces the same pathology. The actuarial reports document the drift.⁹ The political response — periodic “reform” packages that paper over the drift without correcting its causes — is the eusocial-drift signature in classical form. The bureaucracy administering Social Security has not failed; its operational competence is high, and the checks go out on time. What is failing is the political infrastructure around it: the cognitive bias against valuing future beneficiaries at par; the absence of a structural advocacy office for the future-beneficiary class; the absence of personal accountability for the political actors whose decisions consume the fund’s actuarial soundness across the time horizon during which the consequences will arrive.

These domestic cases are not anecdotal asides, and they are not intended as a partisan indictment of either party’s stewardship of either fund. They are the closest contemporary American examples of the school-trust pathology in operation, and they show that the diagnosis is not specific to nineteenth-century land grants. It is a diagnosis of how aspirational structures behave under intergenerational pressure, full stop, and it applies to PERS and to Social Security in the same form it applied to Mississippi’s sixteenth sections in the 1820s and to Oregon’s Common School Fund in 1907. The same anchors apply. The same watchful-crew problem applies. The same architectural extensions — adversarial compact review, asset-translation authority, beneficiary-class boundary provisions — apply.

PERS and Social Security are the empirical bridge for the AI-era architects in the room. They are running the school-trust pathology now, on American beneficiaries, with American officials, in real time. They are also the cases on which the architects of the AI Targeting Authority and the perpetual climate trust can test their architectural intuitions before the AI-era structures harden. If a proposed charter provision would not have prevented the PERS funded-ratio drift, or would not have prevented the Social Security actuarial drift, the provision is not yet strong enough.

The mapping closes

The closing point is the simple one. The school-trust evidence and the AI-age design problem are not separate problems; they are the same problem, observed at different stages of its institutional development.

The school trust has 240 years of running data, partial recoveries, and a slowly arriving enforcement constituency.¹⁰ The AI-age structures have a charter window of perhaps five to ten years before they harden into the form they will keep. The window is when the architecture can still be specified, while the founding generation is still in the room and still thinks the founding moment matters.

What the school-trust evidence offers is a working manual: the pathology, the seven anchors, the eighth element. The architecture worked where it was installed, and it failed where it was not installed; the difference, state by state and structure by structure, is the difference in their architectures and in the constituencies that did or did not assemble around them.

Section VIII — A Letter to the Architects — is where this volume’s accumulated argument speaks to the architects directly. Section VI’s job has been to do the per-instrument translation: to take the school-trust record and lay it down, anchor by anchor and instrument by instrument, in operational form. Section VIII does the rhetorical work the operational record cannot do on its own. The two sections are the volume’s hinge — VI as the working manual, VIII as the address — and they are intended to be read together. The architects who have read this section have the vocabulary they need. What Section VIII asks is whether they will use it.

Footnotes / source notes