The title register

For most of the history of property, owning a piece of land did not mean your name appeared in some authoritative book. It meant you held a bundle of papers. Deeds, mortgages, releases, wills, marriage settlements — a physical accumulation, kept in a chest or, later and less romantically, a shoebox, each document recording one moment when the land changed hands or was encumbered. To own was to possess the evidence of ownership, and the evidence was only ever as good as the pile was complete. Nobody guaranteed it. The papers asserted; you hoped.

What made this arrangement fragile was that it forced every buyer to become a historian. Under the old system of private deeds, title was not a fact you could look up; it was a chain you had to reconstruct. Before you could safely buy, a lawyer would trace that chain backward — this seller took from that grantor, who took under this will, which followed that settlement — link by link, as far back as prudence and the available parchment allowed. The good root of title had to be established afresh for each transaction, by hand, at cost, every time. And the chain was exactly as strong as its weakest link: a lost deed, a forged signature, an undisclosed mortgage, a defective conveyance three owners back, and the whole edifice of your ownership rested on a flaw you had no reliable way to see.

This is the condition worth naming precisely, because we are going to meet it again in a modern guise. Ownership was a private reconstruction. The truth about who held what lived nowhere in particular — it was distributed across a scatter of papers held by interested parties, and it had to be assembled, at the point of need, by someone tracing a fragile lineage they could never fully trust. Provenance was a labor, and a risk, imposed on whoever needed to rely on it.

The shoebox of deeds

Consider what the buyer of a field was actually being asked to do. Not to consult a record and read off an answer, but to satisfy himself, by investigation, that a long sequence of past events had in fact occurred as the seller's papers claimed. Every gap was his problem. Every forgery was his loss. If a document had been suppressed — a prior sale, a lien, an heir with a competing claim — the deeds in his hands would look perfectly complete and be perfectly wrong, and he would discover the defect only when the competing claimant appeared. The system placed the entire burden of verification on the party least able to bear it and offered, in exchange, no assurance at all. What you bought was not certainty; it was the best story the surviving paper could support.

The costs of this were not merely private. A society in which every transfer of land requires a bespoke reconstruction of history is a society in which land moves slowly, expensively, and under a permanent cloud of latent dispute. The friction was structural. It came from the absence of any single place a stranger could go to learn, authoritatively, who owned what.

Making ownership a public fact

The reform that answered this was title registration, and its most influential version bears the name of Robert Torrens, who introduced it in the colony of South Australia in 1858. Torrens — a customs administrator by background, familiar with how the registration of ships settled questions of ownership cleanly — proposed to do for land what shipping registers did for vessels: replace the private chain of deeds with a single public register in which the state records, and guarantees, who holds title to each parcel. Under the Torrens system, the register is not evidence of ownership. It is ownership. The person named on the register is the owner, and the guarantee stands behind that fact.

The change this worked is easy to state and hard to overstate. It replaced "prove your chain" with "consult the record." A buyer no longer had to reconstruct the history of a parcel back through a fragile sequence of documents; he consulted the register, saw who held the land and what encumbered it, and could rely on what he found — because the state, not the seller's shoebox, now stood behind it. The register was authoritative: it settled the question rather than merely asserting an answer. It was public: a stranger could consult it without the owner's permission or cooperation. And it was guaranteed: where the register was wrong, the loss fell on the system that maintained it, not on the innocent party who trusted it. England, characteristically, moved more slowly toward registered title over the following decades, but in the same direction and for the same reason.

The register did not make the chain of title stronger. It made the chain unnecessary — replacing a history each buyer had to prove with a fact anyone could look up.

What Torrens understood, and what the reform encodes, is that provenance can be infrastructure. The origin and custody of a thing need not be a private matter that each interested party reconstructs at cost and at risk. It can be a public good: a single, authoritative, openly checkable record of standing, maintained so that a stranger may rely on it without tracing the whole history himself. Note carefully what kind of instrument this is. It is not a bill of lading, which travels with goods and records their custody as they pass from hand to hand. It is not a hallmark, which certifies the quality of an object stamped into its surface. It is a public authoritative record of title — of who holds standing — that you consult in place of reconstructing it. That distinction is the whole point.

A register for the decision

Now turn the lens onto the machines. When a consequential decision is made by an automated system — a claim denied, a candidate scored, an account frozen, an alert raised — where did that decision come from, and where did its inputs come from? Today the honest answer is that we are back in the age of the shoebox. Provenance is scattered and private. Each system keeps its own partial chain: this model consumed that feature, which was derived from this table, which was populated by that upstream service, whose source nobody in the room can fully name. Establishing where a decision or a datum actually came from means tracing a fragile lineage across parties, most of whom kept only the fragment they touched. The burden of reconstruction falls, as it always did, on whoever needs to rely on the thing — and it falls at the worst possible moment, when the decision is already being contested and the chain is already cold.

The title register is the model for what is missing. Not a better shoebox — not each system keeping a slightly fuller private log of its own inputs — but authoritative, public, consultable provenance: a record that makes the origin and custody of a decision, and of the key inputs that drove it, a fact one can look up rather than a chain each party must laboriously prove. The standard the register set is consult, don't reconstruct. A stranger to the transaction should be able to establish where a decision came from by consulting an authoritative account of it, not by assembling scattered fragments held by interested parties and hoping none is missing or forged.

This is precisely the discipline a Decision Receipt is meant to carry into the machine age: provenance and chain of custody made into a record that stands on its own, so that the standing of a decision is something you consult rather than something the decider asks you to take on faith. The old world learned, at great expense, that ownership left as a private story is ownership perpetually in doubt — and that the cure was to make it a public, guaranteed fact. We are relearning the lesson the expensive way, one contested machine decision at a time. The register was invented in 1858. The problem it solved is sitting, unsolved, inside every system that cannot tell you where its own answer came from.

— Dispatches · Summit Cognitive

---