Evidence Preservation Before Authority
Flight recorders and the independent NTSB investigative function, plus blameless near-miss reporting (ASRS), preceded and outlasted the FAA's full enforcement authority — a weak correction system becomes a stronger one primarily by preserving evidence and widening plurality before it hardens any handle.
What decision changes?
Build incident-reporting infrastructure, interaction logging, and near-miss taxonomies well before the enforcement authority to act on them exists — the evidence base is what makes later authority possible to grant credibly.
Across every route by which safety institutions get founded, one component reliably comes first: the mechanism that preserves evidence. Aviation is the cleanest illustration. Flight recorders — the “black boxes” — and the independent crash-investigation function that reads them were built up well before the aviation regulator had its full modern enforcement powers. The investigators could establish what happened long before anyone could compel the industry to act on it. That ordering turns out not to be an accident of history but close to a design principle: the evidence base is what makes later authority possible to grant credibly, because a legislature can see exactly what the new powers would act on.
Two structural choices made the American aviation system work, and both repay attention. First, in 1974 the crash-investigation function (the NTSB) was made fully independent of the regulator (the FAA) — precisely so that the body establishing what went wrong is never the body whose own decisions might be implicated in the wreckage. An investigator who reports to the regulated-and-regulating agency has a reason to look away from certain findings; an independent one does not. Second, the Aviation Safety Reporting System (1976) offered pilots and controllers immunity from enforcement action in exchange for candid reports of near-misses. It deliberately traded punishment for information, on the theory that a near-miss you hear about is worth more than a violation you punish. Aviation’s extraordinary long-run safety record was substantially built from this blameless-reporting data.
Both mechanisms are notable for working while weak. A black box that cannot yet compel any design change is still worth installing, because it preserves the fact pattern for whenever a stronger handle arrives. A reporting system with no enforcement teeth still accumulates the incident taxonomy that later rules are written from.
For AI, the equivalent bet is concrete and available now: build the incident-reporting infrastructure, the interaction logging, the near-miss taxonomies, and the independent investigation capacity well before the authority to enforce on any of it exists — and keep the investigators structurally separate from both the developers and any agency tasked with promoting the technology. History suggests this is not a consolation prize for weak governance; it is the load-bearing first stage of strong governance.
One of eleven historical case studies in Institutional Genesis, Memory, and Decay — see the overview for the full life-cycle map, or read the complete appendix.
What would count as evidence?
The 1974 separation of the NTSB from the FAA (so the corrector is not the entity whose decisions it might need to indict) and the 1976 Aviation Safety Reporting System (immunity from enforcement in exchange for candid disclosure) both worked while still institutionally weak.