In the long catalogue of governance problems that the twenty-first century is producing — climate adaptation funds intended to outlast every politician who votes them into being, sovereign wealth vehicles charged with husbanding finite mineral wealth across the centuries of its depletion, foundations endowed by technology fortunes whose donors have explicitly named the year 2200 as the horizon of their concern, and the first tentative AI-governance trusts now being drafted in capitals and university foundations on three continents — there is one problem that recurs in every variant and that no contemporary draftsman has yet solved. It is the problem of how to bind a trustee, across timeframes longer than any human lifetime, to honor an obligation owed to beneficiaries who cannot be in the room because they have not yet been born.
The problem has a name in the older legal vocabulary. It is the forever-fiduciary question, and the institutions that are now scrambling to answer it for the first time in the AI era are not, as it happens, the first institutions ever to confront it. The American school trust lands, established beginning with the Land Ordinance of 1785 and the Northwest Ordinance of 1787 and rolled out across thirty-five state admissions over the next 174 years, have been running the experiment continuously for two and a half centuries. They have not been running it as a thought experiment. They have been running it on the ground, with real land, real money, real beneficiaries, and real political pressure. They are the only data set the human race has of any meaningful length on what happens to a perpetual fiduciary obligation when the trustee is a sovereign and the beneficiaries cannot vote.
The data are not encouraging. They are, however, instructive — and the institutions arriving now, which face a much harder version of the same problem with much less margin for error, will be reading them whether they know it yet or not.
The Eighth Anchor frame
The argument for taking the school trust lands seriously as a working laboratory of forever-fiduciary practice has been called, in the wider literature this Library helps gather, the Eighth Anchor thesis. The thesis is that the American constitutional tradition has produced seven enduring institutional anchors — separation of powers, federalism, judicial review, the Bill of Rights, the public trust doctrine, the corporate form, and the regulated administrative agency — and that the eighth, less recognized but no less foundational, is the multi-generational fiduciary trust. The school trust lands are the proving ground because they have run long enough, in enough jurisdictions, under enough variation in political and economic conditions, to surface the structural features of the design that work and the structural features that do not. They are not a parochial issue. They are the most-tested working specimen the country has of an institution built deliberately to outlast its makers.
The relevance to the new forever-trusts of the AI era is not metaphorical. The duties the school trust imposes on a state — undivided loyalty to beneficiaries who cannot vote, prudent management of a corpus that must remain intact across generations, productivity of the assets in real terms, full and faithful accounting to parties who cannot demand it themselves, and a prohibition on diverting the trust to purposes more politically congenial in the present — are, structurally, the same duties that any AI-era forever-trust will impose on whatever entity is designated as its trustee. The difference is that the AI-era trusts will impose those duties under conditions of much greater technical complexity, much faster institutional turnover, and a beneficiary class — future human generations, or in some framings the long-run flourishing of humanity in relation to non-human intelligence — that is even less able to assert itself than schoolchildren.
The question for the new trusts is not whether the school-trust pattern of doctrinal commitment and political slippage will recur. It will. The question is whether the institutional design of the new trusts can be improved, in light of the school-trust experience, to make the slippage less severe.
Two contrasting cases
The school-trust record is not uniform. Different states, working with similar federal grants and similar enabling-act language, have produced strikingly different outcomes — and the contrast among them is the most useful evidence the Library can offer the architects of the new trusts about what the structural features of success and failure look like in practice.
The state whose trajectory is most often cited as the success case is Utah. Utah was admitted to the Union in 1896 with a federal grant of more than five and a half million acres, set aside under the Utah Enabling Act of 1894 (28 Stat. 107) on terms substantially the same as those in every other public-land state admitted during that era. For most of its first century of statehood, Utah managed those lands the way most western states managed theirs — as ordinary public lands inside a general-purpose state agency, with proceeds dropped into a permanent fund that earned essentially what a state checking account earned, with no separate professional staff, no separate fiduciary mandate, and no separate accountability to the trust’s beneficiaries. By 1990, the permanent fund stood at roughly $41 million, and the per-pupil annual contribution was somewhere between negligible and embarrassing.
What changed Utah was a combination of doctrinal pressure and structural reform. The Utah Supreme Court in National Parks and Conservation Association v. Board of State Lands, 869 P.2d 909 (Utah 1993), held that the school land trust was real, that it attached to the use of the land and not merely to the disposition of proceeds, and that the way Utah had been managing its trust lands was unlawful. The decision did not itself design a replacement. That work was done by the Utah legislature in 1994, with the creation of the School and Institutional Trust Lands Administration (SITLA) as an independent agency with a dedicated fiduciary mandate; and again in the years following, with the creation of the School and Institutional Trust Fund Office (SITFO) to manage the permanent fund as a real investment portfolio rather than as a state savings account. Three decades later, the permanent fund stands at nearly four billion dollars — a hundredfold increase, achieved not by any windfall of resource extraction but by the disciplined operation of professional fiduciary management against a beneficiary-aligned mandate. [CITE PENDING — SITFO Annual Report; comparative permanent-fund figures.]
The state whose trajectory is most often cited as the failure case in the modern record is Oregon, and the most visible single instance is the Elliott State Forest. Oregon was admitted in 1859 with a grant of two sections per township for the support of common schools. For most of the twentieth century the trust forest portfolio produced meaningful timber revenue for Oregon schools. Beginning in approximately 2017, however, the Elliott — a roughly 82,500-acre block of the trust forest, located in Coos and Douglas counties — was effectively withdrawn from timber production and transitioned, through a series of legislative and executive actions, into a research-and-conservation forest under the management of Oregon State University. The transition was framed as a “decoupling” of the Elliott from the trust, with the state purchasing the trust’s equity in the forest at appraised value and applying the proceeds to the Common School Fund. Whether the appraised value reflected fair market value, whether the appraisal was a purchaser-supplied appraisal of the kind the Utah Supreme Court held was suspect on its face in National Parks, whether the substitution of research-and-conservation uses for income-producing uses constituted a Skamania-type breach, and whether the entire structure constituted a diversion of trust value for non-trust public purposes are the questions raised by current Oregon litigation. The Library takes no position on the merits of that litigation. What can be said, on the public record, is that the Common School Fund’s distribution rate has run far below its portfolio return rate over the past two biennia, and that the comparison between Utah’s permanent fund growth and Oregon’s per-pupil distribution numbers is not subtle. [CITE PENDING — DSL biennial reports; comparative per-pupil figures.]
The two states had similar grants. They have similar populations. They have similar federal-land contexts. They have produced trust outcomes that differ by an order of magnitude or more, and the difference is traceable, with reasonable precision, to design choices about who the trustee is, what the trustee’s incentives are, and how the trustee is held accountable.
What forever-stewardship requires
The Utah-Oregon contrast is useful precisely because it isolates the structural variables. When the same federal compact, the same general kind of land, and broadly similar political environments produce such different outcomes, the cause of the difference is necessarily institutional. Five features distinguish the design of the successful version from the design of the failing one, and each of the five has direct application to any AI-era trust that will be expected to honor its obligations across centuries.
The first is institutional separation. SITLA exists as a freestanding agency with a fiduciary mandate written into its enabling statute, separated from the general natural-resources department, with its own director, its own staff, and its own budget. Oregon’s trust-lands function, by contrast, has lived inside the Department of State Lands, which is itself a subordinate body of a state government with many competing priorities — economic development, environmental protection, public recreation, local government support — none of which is, properly speaking, a trust purpose at all. The lesson for the AI-era trusts is direct. A trustee whose primary loyalty runs to a general-purpose government, a foundation board with mixed objectives, or a corporate parent with shareholders to whom it owes other duties will, over the long run, slip its fiduciary moorings. The trustee must be a dedicated entity with no other masters.
The second is investment professionalism. SITFO manages Utah’s permanent fund the way a major university endowment manages its corpus — diversified across asset classes, benchmarked against market returns, governed by an investment policy statement and overseen by a board of investment professionals. Oregon’s Common School Fund is professionally managed by the Oregon Investment Council and the Treasury, but the distribution policy that determines how much of the corpus actually reaches the schools each year operates under different constraints, and the gap between portfolio return and distribution rate has been large and durable. For the AI-era trusts, the lesson is that fiduciary discipline at the investment level is necessary but not sufficient; the distribution rule matters as much as the investment rule, and a corpus that grows faster than the beneficiaries can claim is, after a certain point, a corpus that has slipped its purpose.
The third is transparency. The successful trust regimes — Utah’s, Texas’s, New Mexico’s at its best — produce annual reports that publish, in disciplined and comparable terms, the size of the corpus, the asset allocation, the return rate, the distribution rate, and the per-pupil contribution to the beneficiaries. The reports are externally audited. They are written for the beneficiaries’ representatives, not for the trustees’ convenience. The failing regimes, by contrast, produce reports that are difficult to compare across years, that mix trust accounting with general-fund accounting, and that obscure rather than clarify what the beneficiaries received. The lesson for the AI-era trusts is that accountability requires not merely a duty to account but a discipline of accounting, written into the trust instrument and enforceable by some institutional party who actually reads the report each year and is empowered to act on it.
The fourth is doctrinal anchoring. The successful school-trust regimes operate under a body of state-supreme-court doctrine — Pettibone, Skamania, Idaho Watersheds, Ebke, NPCA, Kanaly, and the federal triad of Vincennes, Lassen, and Andrus — that names the duties precisely and that has been built up over decades through litigation. The doctrine is not always followed, as the recurrence of the breaches shows, but its existence creates a stable reference point against which any given administrative action can be tested. The AI-era trusts will not have the benefit of two centuries of common-law development by the time they are most needed. They will have to build the doctrinal anchor explicitly, in the trust instrument itself, with reference to the fiduciary tradition the school-trust cases have already mapped.
The fifth is beneficiary-aligned constituency. The Utah reform of 1994 was not produced by the trustees acting in isolation; it was produced by a coalition — including the National Parks and Conservation Association, the LDS Church, and beneficiary-aligned legislators — that had organized itself to insist that the trust be managed as a trust. The Oregon Common School Fund has had, for most of its history, no organized beneficiary-side constituency at all; the schoolchildren are scattered across 197 districts, the parents are not organized for fiduciary advocacy, and the State Board of Education has not historically treated itself as the beneficiaries’ representative in the trust relationship. For the AI-era trusts, the lesson is structural and uncomfortable. A forever-trust will only behave as a forever-trust if some organized party, generation after generation, takes responsibility for behaving as the beneficiaries’ agent. The trust instrument cannot itself produce that party. It must be cultivated and renewed.
Why this matters for the institutions arriving now
The forever-fiduciary question is becoming acute in the AI era for reasons specific to the era. The institutions that are now being designed to manage long-horizon resources — sovereign wealth funds, climate-adaptation funds, AI-safety endowments, intergenerational treaty trusts, the various “perpetual” foundations whose founders have specified two-century or longer horizons — face a beneficiary class that is even more unable to assert itself than the schoolchildren of any given Western state. The schoolchildren at least exist; they can be pointed to in any given classroom, and an organized constituency on their behalf can be assembled out of parents, teachers, and the institutions that serve them. The beneficiaries of an AI-era forever-trust — future generations not yet born, populations that may not yet exist as political entities, the long-run flourishing of humanity in relation to systems whose own development cannot be predicted — are abstractions even at the level of identification. The trust instrument will have to do more of the work of beneficiary representation than the school trust ever did.
The lessons the school trust experience offers the architects of those new institutions are correspondingly more important. Institutional separation, investment professionalism, transparency, doctrinal anchoring, and beneficiary-aligned constituency are not merely best practices for the school-lands context. They are the minimum design elements without which a forever-trust will, on the school-lands evidence, slip its purpose within a generation or two of its creation. The Eighth Anchor frame holds that getting these elements right — explicitly, in the trust instrument and the surrounding institutional architecture — is the central design problem of the coming century’s forever-institutions. The school trust lands are the working laboratory in which the problem has been studied longest. They are not the answer. They are the body of evidence from which the answer, if there is to be one, will be assembled.
The American school trust lands were the most ambitious multi-generational fiduciary commitment in the American constitutional tradition. They were designed by men who had read the chancery reports of the seventeenth century and who knew what the word “forever” meant in a legal instrument. They were rolled out across thirty-five states over 174 years. They have, in most of those states, failed in some measure to deliver on the promise they were designed to deliver. They have, in a handful of states, succeeded — and the difference between the two outcomes is design.
The institutions arriving now to manage the AI-era forever-trusts will be making the same design choices, under harder conditions, with less margin for error. The Library exists, in part, because the school-trust evidence is the closest thing the human record offers to a treatment effect, and because the architects of the new trusts ought to read the evidence before they finalize the instruments. The forever-fiduciary question is not academic. It is the question on which the credibility of every multi-generational institution arriving in this century will, in the end, be judged.
Citations
Cases
- National Parks and Conservation Association v. Board of State Lands, 869 P.2d 909 (Utah 1993).
- County of Skamania v. State, 102 Wash. 2d 127, 685 P.2d 576 (1984).
- Department of State Lands v. Pettibone, 702 P.2d 948 (Mont. 1985).
- Lassen v. Arizona ex rel. Arizona Highway Department, 385 U.S. 458 (1967).
- Andrus v. Utah, 446 U.S. 500 (1980).
- Trustees of Vincennes University v. Indiana, 55 U.S. (14 How.) 268 (1852).
- Idaho Watersheds Project v. State Board of Land Commissioners, 133 Idaho 64, 982 P.2d 358 (1999).
- State ex rel. Ebke v. Board of Educational Lands and Funds, 154 Neb. 244, 47 N.W.2d 520 (1951).
Statutes
- Land Ordinance of 1785, Journals of the Continental Congress, 28:375–381.
- Northwest Ordinance of 1787, 1 Stat. 51 (re-enacted 1789).
- Utah Enabling Act, 28 Stat. 107 (1894).
- Oregon Admissions Act, 11 Stat. 383 (1859).
- Utah Code Ann. tit. 53C (SITLA, as enacted 1994) [CITE PENDING — chapter and session-law citation].
- Utah Code Ann. tit. 53D (SITFO) [CITE PENDING — enactment year and bill number].
[CITE PENDING] items
- Current SITFO permanent fund balance (approximately $4 billion) and current annual distribution figures — verify against the most recent SITFO Annual Report.
- Oregon DSL biennial distribution and portfolio-return figures — verify against the most recent DSL biennial report.
- Comparative permanent-fund rankings across the ~20 trust-lands states — verify against NASTL summary tables and individual state reports.
- Elliott State Research Forest transaction structure, appraisal methodology, and proceeds — described from the public record; specific dollar figures and dates warrant verification against DSL and ESRF Advisory Committee materials.
Reading Wing, Court Room. America’s School Trust Library at schooltrusts.net.
