VIII. A Letter to the Architects

If you are reading this, you are designing one of the next forever-institutions, or you know someone who is, or your work will be governed by what the institutions decide.

The institutions are arriving fast. AI Targeting Authorities — perpetual, autonomous bodies with their own resources and legal personhood and the capacity to decide for generations not yet born — are entering charter windows at the foundation level and the inter-governmental level both. Perpetual climate trusts, intended to bind tomorrow’s legislatures to today’s emissions commitments, are being capitalized in multiple jurisdictions. Sovereign wealth funds for citizens not yet born are entering the fiscal-architecture conversation in countries that did not previously have them. Longevity escrows, programmable trust instruments scoped to outlive their founders by decades, are being drafted by lawyers who do not yet know that the architecture they are inventing has run before. Compute reserves, data trusts, programmatic down-shifting authorities — the vocabulary varies, the architectural impulse does not. Each of these is a structure designed to bind future generations to commitments made today.

The architects are at the table now. The charters are still drafts.

This volume exists because the country has already run a 240-year experiment in writing the same kind of institution. The school-trust experiment is the only multi-generational fiduciary record long enough to teach what holds and what fails. This section translates the record into design implications you can use.

What the school-trust record actually is

The school-trust experiment began on May 20, 1785, when the Confederation Congress reserved Section 16 of every six-mile-square township in the federal public domain “for the maintenance of public schools.” Article III of the Northwest Ordinance, two years later, dedicated the federal compact to “schools and the means of education” forever. The Ohio Enabling Act of 1803 vested school lands in the state legislature “in trust for the use aforesaid, and for no other use, intent or purpose whatever.” That phrase — and the compact framework it sat inside — became the architectural template for forty-some state admissions across the next 156 years. The federal architecture reached its high-water mark in the New Mexico–Arizona Enabling Act of 1910, which declared the granted lands “held in trust,” defined diversion as “a breach of trust,” and authorized enforcement by the Attorney General of the United States, by the state, and by individual citizens.

The companion volume, LOOKING BACK · Schools of the Republic, is the synthesis of what 240 years of operation tell us about that architecture. This section assumes you have read it or that you will. The findings of that volume sit beneath the design implications below.

The three findings, as you need them

Finding 1. The architecture works where it has been operated as designed. Utah’s contemporary record — a school-trust program grown from $5 million annual revenue in 1990 to $150 million today, thirty times the original amount, with the architecture defended by an active constituency of parents, teachers, and legislators across thirty-five years — is the cleanest working example. The architecture does not work everywhere, but where it has been operated, the trust has held.

Finding 2. Where the architecture has not held, it has failed in two distinct modes. Gradual drift — small diversions, redefined beneficiary classes, modernized mandates, accumulating across a generation or two — runs continuously and is hard to see. Sudden directed seizure — single political moments in which the trust’s resources are taken, often under cover of crisis — runs in episodes and is hard to oppose. Both modes are predictable from the same compounding mix of forces: people discount the future steeply; mature bureaucracies acquire their own continuity interest distinct from the missions they were chartered to serve; beneficiary classes that cannot vote do not produce the political feedback that would otherwise correct the institution. Legal rigidity alone does not hold against those forces. The architecture needs a constituency.

Finding 3. Every successful school-trust enforcement case in the modern record shares one institutional feature: a non-profit plaintiff at the center of the plaintiff team. Not the state Attorney General, who is institutionally captured because she represents the trustee. Not the contingent-fee tort bar, whose economics keep it out of multi-decade injunctive practice. Not the structural-reform civil-rights bar, which has historically not owned public-lands trust law as a portfolio item. The substitute is the issue-organized public-interest non-profit, staffed by lawyers whose institutional employer is committed to the subject matter for the long arc. The non-profit forms because a constituency exists. The constituency is the upstream condition; the non-profit is the legal instrument; the suit is the visible enforcement event. The architecture, however well-written, does not enforce itself.

Eight design implications for what you are building

These are the design moves the school-trust record validates. They are stated in the order in which an architect should think about them.

1. Specify the beneficiary class concretely enough that defending it produces a constituency.

The school-trust beneficiary class — children, present and future — is specific. Disputes about who the trust is for are rare. The class is also small enough at any given moment that the people closest to it (parents, teachers, principals) are an identifiable set. AI-era trusts often have vaguer beneficiary classes: “future generations,” “the public,” “humanity.” Vague beneficiary class produces no constituency. There is no one whose self-recognition as a beneficiary motivates them to defend the institution. If the architecture you are writing names a beneficiary class so abstract that no one wakes up in the morning thinking that means me, you have made the constituency-formation problem structurally impossible. Specify the class concretely. Name the rooms in which the beneficiary’s circumstances are decided. The constituency will form around those rooms.

2. Route the institution’s consequential decisions to the local sites where deliberation already happens.

This is Margaret Bird’s mechanism, drawn from her forty years inside the Utah school-trust system: Increase the revenue to schools, get it directly to every single school where the parents, the teachers, and the principal are deciding how the money is implemented. And all of a sudden, you have built a huge constituency. Translate “revenue to schools” into “consequential work to named local sites where deliberation is already organized.” The principle generalizes. Climate trust funds routed to local watershed councils, to municipal climate-adaptation committees, to indigenous-nation governing bodies — you have built constituencies. Climate trust funds routed to a centralized agency for nation-state-level disbursement — you have built nothing the agency cannot dissolve. The same rule holds for every other class of forever-trust the next generation is building. Routing decisions to local deliberation sites is the operational design choice that produces the political defense the architecture will need a generation from now.

3. Build redundancy into the enforcement architecture. Do not rely on a single enforcer.

The school-trust experiment’s clearest doctrinal lesson is that the natural enforcer the architecture presupposed — the state Attorney General acting as parens patriae for the beneficiary class — turned out, in the school-trust context, to be institutionally captured. The AG works for the trustee, because the trustee is the State and the AG is the State’s lawyer. The doctrinal apparatus the framers inherited from English chancery courts presupposed independence between enforcer and trustee that the school-trust context does not produce.

You have to design for the possibility that your designated enforcer will be captured by the institution being enforced against. Plural enforcement paths matter. The 1910 New Mexico–Arizona Enabling Act got this right: it authorized enforcement by the federal AG, by the state, and by individual citizens. The redundancy is the design feature. Redundancy lets one path lie dormant while another paths carries the enforcement work, which is roughly what has happened across the school-trust record’s last century. If your architecture has only one enforcement path, that path will eventually be captured, and you will have written an institution that cannot enforce itself.

4. Make the corpus visible. Make diversion legible.

The school-trust corpus is land. Land is geographic, surveyed, mapped, and durable. Its identity is preserved even when individual parcels are sold, because the proceeds are tracked as a perpetual fund and the fund’s principal is publicly accounted for. Diversion of the school-trust corpus is, in principle, visible — when it is not visible, the lack of visibility is itself a clue.

Many of the assets you are working with are less visible. Compute is fluid. Data is fluid. Financial instruments are abstract. Fluid asset forms make drift harder to detect, which means drift will compound longer before anyone catches it. The design move: build legibility into the corpus from the beginning. Public reporting on the corpus’s composition. Periodic audits with public release. Visible incompleteness — empty cells in the report where data has not yet been collected — as a credibility signal. Margaret Bird’s working principle, which the school-trust accountability movement has settled on across the past decade, is that visible incompleteness with correction pathways is the fastest path to credibility, and is the only way a public ledger of public assets can earn its name. The principle applies whether the asset is land, capital, compute, or data.

5. Engineer for the politically-silent beneficiary.

The school-trust beneficiaries — schoolchildren, present and future — cannot vote. They do not write op-eds. They do not contribute to political campaigns. They are, in political terms, silent. The institution that is supposed to protect their interests has to protect them against the natural tendency of every legislative body to redirect resources toward beneficiaries who can vote.

The design implications are several. Standing provisions that let parents, guardians, advocates, and non-profit organizations sue on behalf of the silent class. Mandatory advocacy infrastructure inside the institution itself — a fiduciary advocate’s office, separately funded, structurally insulated from the institution’s operating leadership. Public-record obligations that force the institution to disclose its decisions in real time, so that drift cannot compound across years before anyone notices. None of this substitutes for an active constituency. All of it is what the architecture has to provide so the constituency can organize when it forms.

6. Mandate the cognitive infrastructure for the watchful crew.

The school-trust framers could not engineer for the constituency that would defend the architecture across two and a quarter centuries because the cognitive technology of 1785 could not hold institutional context across generational time. A citizen-trustee in 1885 had no way to know what the trustee of 1885 inherited from the trustee of 1825. Memory had to live somewhere, and the available somewheres were inside individual heads — where it died with the head — or inside paper records that depreciated faster than the asset they described. The framers used legal rigidity as a substitute for institutional memory because legal rigidity was the only memory technology of their century.

Your century’s cognitive technology is different. Frontier AI systems can hold context across decades. Public databases can hold figures and citations indefinitely. Networked infrastructure can let citizens in 2125 audit decisions made in 2025 with the same ease that citizens in 2025 can audit decisions made in 2024. Use it. Build the watchful-crew infrastructure into the institution’s charter. Mandate the public-ledger publication. Mandate the editorial-and-citation discipline. Mandate the substrate accessibility that lets future watchers do the work the founding generation could not do for itself. The cognitive technology the school-trust framers needed and could not engineer for is, accordingly, becoming feasible. Build it in.

7. Sequence carefully. Routing first, recognition second, ceremony third.

Margaret Bird’s sequencing rule for the Utah school-trust reform — first increase revenue from the lands, then save the fund — generalizes to the institutional design problem. Spending the principal before the revenue is in place is a fiduciary error. Building the symbolic vesting layer (the seal, the certificates, the ceremonial recognition) before the routing-deliberation-recognition chain has produced visible artifacts is an institutional error of the same shape. Without artifacts to vest, the symbols are decoration. With artifacts, the symbols are wayfinding.

The rule for your institution: build the routing layer first; let the deliberation produce visible artifacts; recognize the artifacts through a public record; only then add the ceremonial recognition layer. Reverse the sequence and you produce a fraternal order before you produce the work. Constituencies form around work. They do not form around ceremony.

8. Write the institution’s own architecture into its public substrate.

The school-trust framers wrote the architecture in statutes and constitutions but did not document the architecture’s operation in any sustained way. The result is that the operational know-how of the trust — how to administer Section 16 lands, how to compute the income equity from a permanent fund, how to read the historical breach-of-fiduciary doctrine — has had to be reconstructed in every generation. The reconstruction is expensive and the institutional memory it produces is fragile.

Your institution can do better. Publish the architecture itself as part of the institutional substrate: the charter, the implementation procedures, the audit standards, the editorial methodology, the relationship between the institution and its watchful crew. Make the architecture as public as the corpus. Future generations will not have to re-invent what you have built; they will inherit a working pattern they can extend. The Library at schooltrusts.net is the school-trust experiment’s belated attempt to do this. Your institution has the chance to do it from the beginning. The Library at schooltrusts.net is one worked example of this implication in operation: its five-layer Knowledge Stack — primary sources, canonical claims, audience models, toolchain, deliverables — is described in Section V.5 above. The point is not that every long-horizon institution should adopt this Stack literally; the point is that long-horizon institutions need some working architecture for substrate, canonical claims, audience-aware rendering, partnered toolchains, and deliverables — and this volume’s Stack is one specific instantiation a future architect can extend.

Three failure modes the school-trust record warns against

These are the moves the record predicts will produce drift, capture, or seizure. Avoid them.

Failure mode A: The architecture relies on a single legislative moment. The most fragile institutions in the school-trust record are the ones whose architecture rests on a single piece of state legislation, alterable by the same legislature that wrote it. New Jersey’s perpetual school fund — strongly protected, constitutionally vested, with an anti-borrowing provision phrased as strongly as anything in the country — has nonetheless been altered by amendments to the same constitution. The lesson is that the strongest formal protections, written into the most senior state law, can be revised by the same political process that created them. If your institution can be undone by the next legislative session, it has not yet been engineered against the political-cycle problem. Build provisions that require multi-stage amendment — federal compact, constitutional ratification, supermajority — and insist on supermajority requirements for any reduction in the institution’s commitments to the beneficiary class.

Failure mode B: The architecture treats the constituency as a downstream result rather than an upstream input. The school-trust record’s most damaging assumption — implicit in the framers’ design and unaddressed for two centuries — was that the constituency would form on its own once the architecture existed. It did not. It formed in places where someone organized to build it. Where no one organized, the architecture drifted and the corpus rotted. If your institutional design assumes that the constituency is automatic, you have already designed for failure. The constituency must be an explicit operational object of the institution, with budget, staff, and named institutional carriers. The non-profit substitute that has carried the school-trust enforcement work for the past forty years was an emergent response to the framers’ missing step. Your institution can build the equivalent explicitly.

Failure mode C: The architecture lets the trustee define its own performance metric. Many institutions in the school-trust record have, across time, redefined the metric by which the trust’s performance is measured. The “permanent fund” gets accounting redefinition. The “income” gets reclassification. The “beneficiary” gets expanded or contracted. The trustee, in each case, was the entity defining its own performance metric. The result is that drift can occur entirely within the metric, with the metric itself migrating to accommodate the drift. If your architecture lets the trustee define the audit standard, it has not yet been engineered against this. The audit standard, the disclosure obligations, and the public-reporting format must be set externally — by the constitution, by an independent board, by mandatory standards-organization governance — and must be amendable only by the same multi-stage process that amends the institution’s charter.

What this section is asking you to do

The school-trust experiment is the only multi-generational fiduciary record long enough to teach what holds and what fails. The findings are now in: the architecture works where someone shows up to defend it; legal rigidity alone is not enough; the constituency that defends the architecture is built by routing consequential decisions to local sites where deliberation already happens. Translated into the design language of your institution: the trust survives where its watchful crew is built into the architecture, not where it is left as a downstream wish.

Your charter is still a draft. The institutional architecture you write in 2026 will operate, if your design holds, for centuries. The school-trust record is on your desk now in the form of this volume. The companion volume — LOOKING BACK · Schools of the Republic — has the case-by-case evidence. This section has the design implications.

The country owes the next generation of trust architects what we learned. We are not the people who will live to see the institutions you are building reach their tenth, fiftieth, two-hundredth year of operation. The schoolchildren of 2186, like the schoolchildren of 1859, are the beneficiary class your architecture will have to outlast. Whether the architecture holds depends on whether the watchful crew is in place when the architecture is later attacked — and whether the watchful crew is in place depends on whether your design produced the constituency that produces the crew.

We do not know whether the AI-era trusts now arriving will hold across the time horizons their charters set. The school-trust record suggests that the architecture, properly designed and operated by a constituency that treats it as theirs, can hold. It also suggests that the architecture, even strongly written, drifts and is seized in the absence of the constituency. The active variable, after two and a quarter centuries, is what it has always been: the people willing to show up.

That is the lesson, and it is also the question. Where, in your institution, is the watchful crew? What sentence in your charter tells the next generation how to find them? Whose names will be on the public ledger when your founding generation is gone, and what artifacts will those people have produced that establish their standing in the institution you are about to launch?

The architects of 1785 did not have the answers to those questions. They wrote what they could write. We have the answers now. Use them.

— Dave Sullivan, Oregon Advocates for School Trust Lands and the Library, 2026.