The Domesday Book

Twenty years after Hastings, William the Conqueror wanted to know exactly what he had conquered. At his Christmas court in Gloucester in 1085 he ordered a survey of England, and through 1086 his commissioners rode circuit through the shires, convening local juries and recording, manor by manor, who held each piece of land, who had held it before the Conquest, how many ploughs and how much woodland and meadow it contained, and what it was worth. The returns were compiled into what we now call the Domesday Book — two volumes, the whole of England south of the northern counties reduced to a single ordered account of holdings and values. The Anglo-Saxon Chronicle, written close to the event by men who resented it, records the thoroughness of the inquest — so exacting, the chronicler complains, that not one hide nor virgate of land was left out. The complaint is itself a form of testimony. The survey was total, and everyone knew it.

The name came later, and it is the more revealing for that. Within a century the survey was being called Domesday — doomsday, the day of judgment — and a royal treasurer, writing around 1179, explained why: just as no man can escape the sentence of the Last Judgment, so no man could evade or overturn the verdict of this book on who held what. Its judgment was final because there was no higher record to appeal to. This is a striking thing to say about an administrative document. People did not call it Domesday because it was long, or beautifully written, or commissioned by a powerful king. They called it Domesday because its account of holdings was treated as beyond appeal — the settled reference a dispute was measured against rather than a claim that reopened the dispute.

The record with no appeal

It is worth being precise about what earned the book that standing, because it was not one thing but a combination, and the combination is the whole point. The survey was systematic: it did not sample or estimate, it aimed to cover the country, shire by shire, hundred by hundred, manor by manor, on a fixed schedule of questions asked the same way everywhere. It was sworn: the facts were not the commissioners' impressions but the testimony of local juries, men of the hundred put under oath, whose answers could be cross-checked against the accounts of the lord and, where they conflicted, tested. And it was fixed: the returns were written down, compiled, and kept, so that what had been established could be consulted afterward rather than re-argued from memory by whoever spoke last and loudest.

Take any one of those away and the authority collapses. A survey that covered only the counties a king happened to care about would settle nothing about the rest. A survey that recorded a scribe's guesses rather than sworn testimony would be an opinion, however tidy. A survey that was compiled and then lost, or quietly amended, would be a rumor with a good provenance. What made Domesday consequential was that it was complete and produced by a defined procedure and fixed in a form everyone could point to. Completeness alone would have been a heap; procedure alone, a method with nothing to show; fixity alone would have preserved whatever happened to be written, true or not. Together they made a record that could stand as the evidence.

The book was not believed because it existed. It was believed because of how it was made — and the two are not the same, though we constantly mistake the first for the second.

Authority earned by method

This is the principle the Domesday Book quietly established and then embodied for nine centuries: a single, comprehensive, systematically gathered, written record, produced by a disclosed process and treated as authoritative, can settle disputes about rights and holdings by becoming the reference everyone consults instead of a claim each party re-litigates. The power is not in the ink. It is in the fact that the way the record was made gives it standing. When two men disagreed about who held a manor, the question was not whose story was more persuasive but what the book said — and the book could be trusted to say something because it had been assembled by sworn inquest, systematically, and kept unaltered. Authority, on this model, is not asserted by the party who holds the record. It is earned by the method that produced it, and it is earned in advance of any particular dispute, which is precisely why it can resolve disputes rather than become one.

We tend to get this backward. We treat authority as a property of the source — the crown, the institution, the system — and the record as merely the source speaking. Domesday inverts that. The record outlasted the dynasty that made it and kept its authority under kings who had nothing to do with William, because its standing was lodged in its method, not in its patron. A record whose authority depends on who is currently vouching for it is only as durable as their tenure. A record whose authority is earned by how it was made can be handed forward, consulted by people who never trusted its author, and still hold. That transferability is the mark of the real thing.

An authoritative record for the machine's claims

Now consider what machine systems produce. They generate vast records of who holds what and what is the case — balances, entitlements, scores, classifications, determinations of eligibility and risk — and in sheer volume they dwarf anything a medieval clerk could compile. But they almost never have the method that earned Domesday its authority. They are not systematically gathered against a fixed, disclosed set of questions; they are assembled from whatever inputs happened to be at hand. They are not produced by a defined and inspectable procedure a skeptic could examine; the process is undisclosed, or nondeterministic, or has since changed. And they are not fixed and appealed-to as a common reference; they are regenerated on demand, each query a fresh assertion that may not match the last. So they settle nothing. Either they are endlessly re-litigable — every party disputing the number because there is no established record to point to — or, worse, they are trusted on faith, which is the pre-Domesday condition the survey was built to end.

The Domesday Book is the model for the thing we are missing: the authoritative record earned by method. Not a record that decisions rest on merely because it exists and is official, but one they can rest on because the way it was made — systematic, verified, fixed, and disclosed — gives it standing that survives the parties who made it. This is what a Decision Receipt is for, and it is why provenance and replay and contestability are not ornaments on such a record but the sources of its authority. A machine determination of who holds what earns the right to be the reference, rather than a claim to be re-fought, in exactly the degree that it was produced by a method a doubter can examine and re-run.

But Domesday carries a warning inseparable from its lesson, and honesty requires naming it. The survey was an instrument of a conquering power. William ordered it to tax and control a country he had taken by force, and its finality served him and his heirs as much as it served any dispossessed Englishman seeking to prove a claim. An authoritative record serves whoever commands it. A record from which there is no appeal is a magnificent settlement of honest disputes and a terrifying instrument of an unaccountable one — the same fixity that stops the powerful from rewriting the past also stops the wronged from correcting it. So the lesson has two halves that must be held together. Authority earned by method is what a record needs to be worth resting a decision on; authority without contestability is how a record stops serving accountability and starts entrenching whoever holds the pen. The book must be beyond casual dispute, and the person it judges must still be able to make it answer. Domesday got the first half right and, being what it was, was never asked for the second. We are building records at a scale it could not imagine, and we do not have that excuse.

— Dispatches · Summit Cognitive

---