The Presumptions
An archive that hides its rules is asking to be trusted on faith. This one states its rules before it applies them, so that you may hold it to them. Each presumption below is rebuttable, and each names what would rebut it. That is the difference between a doctrine and a dogma.
Attribution over diffusion
Where people did the work, the record names them as the authors of the result. Causes may be many; authorship is not thereby dissolved. To spread the origin of a thing across so many factors that the people who made it disappear is not neutrality — it is erasure performed with a scholar's vocabulary. The enslaved built the wealth of the American empire: the cotton that was the nation's leading export, the capital that filled its banks and insurers, the labor that fed its mills and its shipping. The record says so plainly, and does not launder it into “a variety of contributing conditions.”
Presumption of authorship
Where Black or enslaved labor, technique, or invention produced a thing, the archive credits them as its producers. This is a presumption, not a conclusion: it sets the starting point and it places the burden. Whoever would place the credit elsewhere carries that burden, and the archive will hear them.
Testimony is evidence
First-person and community record — the narratives, the oral history, the family archive, the thing the grandmother said and the church minute book wrote down — is admitted on its own standing. It is held to no higher bar than a secondary source written by someone who was not there. A people's account of their own life is evidence, not anecdote awaiting authorization.
Corroboration, not permission
The archive cites scholars to corroborate, never to authorize. Their work strengthens a record; their inattention does not weaken it. Where the academy never looked — and on this subject it often did not — the archive says so, and then it looks.
The instrument of the theft is not a witness
A patent, a trademark, a byline, a copyright, a deed, a company name — these record who was permitted to own. They do not record who made. They are instruments of legal standing, and for most of this country’s history Black people were barred from holding them. An enslaved man could not receive a patent: he was not a citizen and could not swear the oath. His enslaver could not receive it either, not being the inventor. So the invention was used, and no one was credited. The Confederacy then wrote a law expressly permitting masters to patent their slaves’ inventions — a confession, in statute, about who was doing the inventing. The absence of a Black patent is not evidence of the absence of Black invention. It is evidence of the bar. Therefore the archive will not accept official credit sitting in another name as disproof of Black authorship. That credit is not the refutation of the theft; it is the fingerprint of it. Nathan Green taught Jack Daniel to make the whiskey, and the bottle carried the other man’s name for a century and a half. The trademark proved only who was allowed to own a company.
Records begin as machine-assisted research: a subject is brought by a contributor or commissioned by a keeper, researched against the open literature and the archives, and drafted with its sources attached.
Every draft is then put to an adversarial verifier, whose job is to assume the draft is wrong until it proves itself: that each citation is a real, findable work; that the sources actually support the claims; that nothing is overclaimed. A draft that fails is held, and no clock will publish it — it waits for a human, indefinitely. A draft that passes and that no keeper has amended enters the record after seventy-two hours, marked In review: publicly readable, and not yet verified by a person.
The Verified seal is applied by a human being, and only by a human being. A record marked In review has not yet earned it. Every record carries a public history of every change made to it, by whom, and when.
We set this out plainly because the alternative — pretending a crowd wrote what a system drafted, or calling a record verified when no one has read it — is the kind of lie this archive exists to correct. The research is assisted. The seal is human. The sources are real, or the record does not run.