One price, inspected — a synthetic explainer

One price, inspected.

When software sets a price and a hard question arrives later, what can anyone actually show?

A single local file. No login or analytics. Once opened locally, the walkthrough sends nothing unless you follow the Proofborne link.

This is what one decision-level record can look like.

Four findings, each on its own axis, each stating what it was checked against and — where it rests on someone else’s word — whose word that is. Read the small print: it is doing the real work.

Notice what is missing: there is no aggregate result, no score, no pass. Each badge is one axis’s status; they do not add up to a verdict.

Six weeks later, one question puts a label in dispute.

A reviewer asks the question nobody had asked. The Market Demand Index is forward-looking — which on its own tells you nothing about provenance, since a forecast can be built entirely from public or historical inputs. So what is this one actually built from?

Read the mechanics carefully. The record stays exactly as it was, and remains true about what was presented at decision time. What failed was the record’s ability to rely on the vendor’s classification once that classification was contested — not proof that the vendor’s account was false, or that the inquiry’s account was correct. The record had marked that classification uncorroborated from the moment it was written. Anyone who read CONFORMANT as “clean of competitor signal” was reading more than the record said — though the badge invited the mistake. Uncorroborated is not a warning about content, and not a claim that nobody had looked: it records that this record carries no independent corroboration of the vendor’s classification. That declared dependence is what made the affected claims scopable — identified, dated and corrected — once the classification was contested.

Nothing is deleted, nothing rewritten. Corrections link; they never erase. That is what makes the record worth anything at all.

Five weeks before Hotel D’s price, a gateway saw nothing at all.

At Hotel B — another of the vendor’s twelve clients in this market — the pricing system was unavailable for four hours on 9 February 2027. A regional manager changed forty room-nights through nineteen distinct rate decisions, by spreadsheet upload, directly into the property-management system. The pricing gateway recorded nothing during the outage; the nineteen gaps surfaced only later, when reconciliation was run against that system.

The register restores who and when. It does not restore what information entered those nineteen decisions — and that is the question the record exists to answer. The uncomfortable rule: “all prices flowed through the controlled path” is itself a claim that needs evidence. A February record claiming 100% coverage would simply have been false. The record that honestly reports its gaps — with the manager named and every timestamp attached — is the one whose scope can be checked.

A chain of custody for a price.

Like a chain of custody, it captures and preserves where the evidence came from and how it was handled — material that can support authentication. And, equally, it never decides the case. The analogy has one more limit, and the record wears it openly: a custody chain tracks identified hands on an item whose identity can be checked independently, while two of this record’s four inputs are issuer assertions this chain does not independently verify. Custody preserved is not content verified.