Traceability

A correction is a crossing. Evidence enters somewhere, reaches a decision, and consequence lands, and the crossing spans unlike things: readings, records, meetings, authorities, actions. While everything works, nobody needs the path. The need arrives afterward, when the same failure returns and someone has to establish what the last repair actually reached, or when a decision must answer for the basis it was made on. By then the path either survives as something reconstructible, or it is gone, and no amount of good intention rebuilds it.

Memory continuity is about survival: the relevant prior state persists, with the connective tissue that keeps it intelligible. Traceability is about reconstruction: the crossing itself, who decided what, on which basis, with what consequence, can be rebuilt from what survived. The two come apart in both directions. An archive can be complete, ordered, and perfectly intact while the crossing stays unreconstructible, because the decision was never recorded against its basis, or the consequence was logged in a system the evidence never touched. And a crossing can be reconstructible from fragments, painfully, when the surviving pieces happen to bind. Intact records make reconstruction possible. They do not, by themselves, make the crossing reconstructible.

Archival science built the deepest machinery: the principle that records travel with their provenance and original order, the bond that ties a record to the records around it, and the custodial history that lets a later reader establish what a document was to the people who made it. Metrology built the chain model: a result is traceable when an unbroken chain of comparisons, each link accountable, connects it to its reference. Data provenance work made derivation machine-readable, so that what produced what can be recorded as data in its own right, and its own specification is candid that representing a derivation is not the same as being able to reconstruct it. Configuration management built the change record, the practical crossing for controlled artifacts: what changed, when, by whose authority, from which version. And legal practice built chain of custody, the oldest consequence of the idea: evidence whose path cannot be established loses its standing in the room where standing is decided. Different regimes, different kinds of trace, and one shared requirement: where a result must answer for where it came from, the path is not disposable.

Traceability, in this corpus, preserves the reconstructibility of the corrective history where later correction depends on it: on knowing what evidence reached which decision, what consequence followed, and whether a prior correction reached the structure that generated the failure. That last dependency is the sharp one. Recurrence is one of the strongest tests a correction faces, and it needs no trace to be evidence: the return of a failure class under comparable conditions already counts heavily against the prior repair. What reconstruction adds is diagnosis. Across years or decades, the earlier repair has to be reconstructible if a later investigator is to determine where it stopped, what it changed, and whether it ever reached the generating structure. Recurrence can show that a correction failed without showing why. Reconstructing why is traceability's work. A system that cannot rebuild its own corrective history can still be corrected today. What it may lose is the ability to determine whether earlier corrections reached structure, and where they failed when they did not.

Data provenance supplies a useful fact about the machinery: the provenance of a path can itself be represented as a record. Once it is, everything the corpus establishes about records can apply to that record too. Its survival is memory continuity's problem. Its entry into a later decision is a standing question. And under adversarial load, it is attackable by the same taxonomy that attacks any record: it can be thinned, recontextualized, or stripped of consequence while every underlying document sits untouched. An attack on the trace does not disable correction wholesale. What it impairs are exactly the forms of correction that depend on reconstructing the corrective history. That is an obvious attack surface: a party need not destroy the underlying evidence if it can sever the path that makes the evidence usable later. The path may be the cheaper target.

Traceability can preserve the reconstructibility of the crossing. It cannot make the reconstructed account true; checking claims against a basis beyond the system is verification's work. It cannot admit anything into a decision; that is standing. It cannot make the records needed for reconstruction survive in the first place; memory continuity sits beneath it. A reconstructible path is not a corrected system, and a perfect trace of a failure that reached no authority is a well-documented failure. The condition earns its place in the aircraft not by doing those jobs but by making one specific thing possible: a corrective history that can still be interrogated after everyone who lived it is gone.