Editorial Practice

Editorial Standards

The Library is cited. Figures and claims published here appear in court filings, in legislative testimony, in letters to trustees, and in the work of scholars and journalists. That is the standard we write to. Everything in the catalog must be defensible by a reader who shows up with the underlying source in hand. These Editorial Standards describe how we get there: what counts as a citable claim, how we tell facts apart from interpretation, how we mark uncertainty, and what happens when we get something wrong.

Every claim cites a primary source

If a catalog record states a number, a date, a holding, or a fact, that statement is linked to a primary source. For the Library, primary sources are:

• statutes and constitutional provisions, cited to the published code;
• court filings and judicial opinions, cited to the docket and where possible to a Bates-stamped page;
• agency records produced in discovery or under public-records law;
• published scholarship in books, peer-reviewed journals, and reports of record;
• on-the-record statements by named public officials in their official capacity; and
• machine-verifiable records (a county assessor's parcel map, a state auditor's spreadsheet).

If a school trust acreage figure appears on a state page, the source — statute, audit, study, or production — is linked from the figure itself. The reader should never have to ask where a number came from.

When a primary source is not yet in hand, we say so. The claim is flagged "citation pending" if we expect to land one shortly, or it is held out entirely. We do not publish numbers we cannot defend.

Catalog records carry facts. Reading Room essays carry argument.

The Library has two kinds of surface, and they follow different rules.

Catalog records and figure tables carry facts. A state page tells you what the enabling act granted, how many acres remain in trust, where the corpus is held, what the most recent audit found. Each statement carries a citation. Where a number is contested or partial, a confidence badge says so. These pages do not argue; they describe.

Reading Room essays carry interpretation. When an author connects facts into a thesis, draws an inference, frames a pattern across states, or recommends a course of action, that work lives in a Reading Room essay. Essays are bylined. The reader can see who wrote it and on what authority.

We hold the line between these two surfaces on purpose. A reader scanning a state page should be able to trust that what they see is fact. A reader opening an essay should know they are reading an argument that one person stands behind.

No original argument in the catalog

We adapt a rule that the open encyclopedias arrived at the hard way: catalog records contain no novel argument. If a claim has not appeared in a primary source or in a published treatment by a recognized authority, it does not belong in the catalog. The catalog is a finding aid, not a venue for new analysis.

This is not a bar on new thinking. New analysis is welcome — in the Reading Room, under a byline, with editorial review. Several of the project's most important contributions began as Reading Room essays, including the Sacred Compact sequence and the Eighth Anchor analysis. When an essay's argument matures into something that scholars, courts, or auditors begin to treat as settled, the underlying facts may move into catalog records and be cited accordingly. The argument itself stays in the essay where it was made.

Confidence badges: when a number is uncertain, the page says so

Some figures are clean: an enabling act granted a specific number of sections, and the statute is in the books. Other figures are not. A state may have stonewalled discovery on a question. A published source may round. Two reputable sources may disagree. An acreage may be current as of a 1979 audit and unverified since.

When a figure carries uncertainty, the page where it appears says so on the surface, through a confidence badge. The badge names the reason: partial production, source disagreement, date of last verification, citation pending. A blank cell in a table is a signal, not an omission — it tells the reader we have looked and have not yet found a defensible number.

Confidence badges are now in use on the Counting House figure tables and roll out to state pages as those pages are restructured. The rule applies before the badge appears on every surface: where a number is uncertain, we say so where the number sits.

How we correct ourselves

The Library publishes for litigation-grade audiences. Correction discipline matters as much as publication discipline. We work in three tiers.

Typographical and presentational fixes. Spelling, a broken link, a misformatted date. We fix the page and move on. No correction note.

Minor factual corrections. A date off by a year, a misattribution of a quote within a single source, a rounding that overstated a known figure. We fix the page, and a short correction note is recorded on the page's Corrections subsurface so the change is visible to anyone who wants to see what moved and when.

Substantive corrections. A claim was wrong. A source has been retracted. A holding was misread. In these cases the original wording is preserved in the correction record, the corrected wording is in place on the page, and a note explains what changed and why. Substantive corrections are logged publicly. We expect them to be rare; we expect to be candid when they happen.

Readers who think we have something wrong can use the Submit-Correction control on any page. A librarian reads every submission and responds. Where a correction is accepted, the protocol above applies. Where it is not, we say why.

The Corrections page is being built alongside this one. Until it is live, substantive corrections are logged at the bottom of the affected page.

AI authorship is disclosed

Some work in the Library is drafted primarily by a human, with AI assistance for research, structuring, or copy editing. Some work is drafted primarily by an AI, with a human reviewing and accepting. The two cases are not the same, and we say which is which.

When work is AI-primary, the page or memo carries an AI byline naming the model. When work is human-primary, there is no AI byline; conventional authorship credit applies. The methodology — how the project works with AI tools, what the human-AI division of labor looks like, what gets reviewed and how — is described on the How This Works page. Readers who want to weigh authorship can find the rules there.

Contributors disclose interests

The school trust lands are a working policy domain. Many of the people best positioned to contribute to the Library hold paid roles in state trust land agencies, in school districts that draw from trust distributions, or in organizations whose work bears on the corpus. We do not exclude contributors who hold those roles. We require them to be declared.

A contributor with a material interest in a claim discloses it on the page where the contribution appears. A contributor who holds a relevant paid role discloses that role on their public profile. The reader can then weigh the contribution with the interest in view. The Library's posture is that declared interest is information, not disqualification.

How to flag a concern

If you think a claim on the Library is wrong, or a citation is missing, or a confidence badge is too lenient: use the Submit-Correction control at the bottom of the page in question. A librarian will respond. For higher-stakes disputes — disagreement over editorial scope, authorship, or institutional posture — see the Governance page for the standing escalation path.

These standards are not static. They will be revised as the Library matures, as audiences raise questions we did not anticipate, and as the practice of working at this scale teaches us things we cannot know yet. When the standards change, the changes themselves will be logged.