The yardstick you wrote yourself

Posted on: 3 August 2026

British readers have spent the past decade watching institutions explain, in the calm register of technical correctness, why the thing that went wrong did not fall within the relevant category. The formula is familiar enough to have lost its power to reassure. What happened in a Senate committee room in Washington on Wednesday is the same structure in a louder key, and it is worth looking at closely precisely because the noise around it obscures how ordinary the mechanism is.

Anthony Fauci, eighty-five and retired since 2022, invoked the Fifth Amendment more than a hundred times before the Senate Committee on Homeland Security chaired by Rand Paul. Paul has said the committee will vote in the coming days on a contempt of Congress referral. The American commentary has arranged itself into the two available positions, persecution or guilt, and neither position explains anything.

The charge is narrower than the coverage suggests. Paul is not reproaching decisions taken in the chaos of early 2020 using the standards of 2026. He is contesting a specific sentence delivered to the Senate on 11 May 2021, when Fauci stated that the National Institutes of Health had never funded gain of function research at the Wuhan Institute of Virology. Lying to Congress was a criminal offence then as it is now, so this is not a case of rules written after the game. It is something more interesting: a statement that is true under a precise technical definition and false in ordinary usage, where the precise technical definition had been produced by the apparatus the statement was protecting.

The documentary sequence is public and repays being read in order, because the order is where the mechanism becomes visible. In October 2014 the United States government suspended federal funding for research that might confer enhanced pathogenicity or transmissibility on influenza, MERS or SARS viruses. It was a pause rather than a permanent prohibition, opening a deliberative process on risks and benefits. In January 2017 the White House Office of Science and Technology Policy issued guidance known as P3CO, which was not binding law but a description of the attributes a review mechanism should possess before an agency could resume funding such work. In December 2017 the Department of Health and Human Services published its own applying framework, and on 19 December 2017 the National Institutes of Health formally lifted the pause, stating that they were doing so in accordance with the existence of that framework.

That step deserves rereading. The suspension did not lapse because the risk had diminished. It lapsed because the apparatus disbursing the money had equipped itself with a criterion, written by itself, for assessing the money it disbursed. The Congressional Research Service records the detail that establishes the real scale of the thing: the Department of Health and Human Services was the only federal agency to build a review process in response to the 2017 guidance and the only one subsequently to report funding in that category. Not a distributed system of oversight, then, but a system of one, in which the body doing the checking and the body being checked belong to the same organism.

The committee applying that criterion operated out of sight. In April 2022 the House Energy and Commerce Committee wrote to the Health Secretary complaining about the opacity of the process, noting that in the summer of 2021 the chair of the review panel had briefed congressional staff without disclosing the names of its members. A panel deciding which research on pathogens of pandemic potential deserves public funding, whose membership Congress could not persuade it to reveal.

When the question eventually arrived at the EcoHealth Alliance project with the Wuhan institute, the answer was consistent with everything preceding it. According to NIH testimony to the House select subcommittee, the project fell within neither the criteria of the 2014 pause nor those of the 2017 framework. Formally unimpeachable, and the formal unimpeachability is the whole problem, because the criterion the project failed to meet had been drafted by the same structure that funded the project.

One distinction matters before the argument slides where it should not go. The framework belonged to the Department, not to the institute Fauci directed, and the review panel was chaired by an official from the preparedness side rather than by him. He did not write that definition with his own hand. Were this to become a personal indictment it would lose the only thing in it worth having, which is that the mechanism operates at the level of the apparatus and would operate identically with anybody else in the chair. Individuals matter less than structures here, as they usually do.

Stated plainly, the mechanism runs as follows. Whoever holds the power to define the terms by which they will be judged acquires real, immediate, demonstrable protection, since any challenge can be defused by showing that the contested fact falls outside the definition. It is an insurance policy that works extremely well, and it works for as long as the people doing the judging are peers, which is to say people who accept the legitimacy of the technical distinction and admire its precision.

The moment judgement moves out of the community of peers and into a political arena, the same policy inverts. A definition produced by the party it governs cannot be falsified by anybody standing outside it, and a claim that no outsider can test stops being a claim and becomes a position. Popper made the point about scientific theories and it holds identically for institutional defences: what is built so as to be immune from refutation forfeits, along with the risk of being refuted, the capacity to persuade. In a committee room the answer that under the operative definition then in force the project did not fall within the category does not produce acquittal. It produces the settled conviction that something is being concealed. Not because it is false but because it is invulnerable, and people distrust by instinct whatever cannot be struck.

This will read as unfamiliar only to somebody who has not been paying attention at home. The Grenfell Tower inquiry spent years establishing how the testing and certification regime for cladding systems had been shaped by the manufacturers whose products it was supposed to constrain, and the resulting defence, that the assemblies complied with the applicable standards, arrived in public as something closer to a confession than an exoneration. The same logic sits, in a gentler form, at the centre of British corporate governance, where comply or explain invites the company to account for its own departures from the code and the quality of the explanation is assessed largely by people with an interest in its being adequate. In ordinary years this looks like proportionate, intelligent regulation that avoids the deadweight of prescriptive rules. It has never once looked like that during a crisis.

I have watched the architecture work and then turn in settings with no connection to virology, in industry technical committees where specifications were drafted by the manufacturers required to meet them, defended by the unanswerable argument that only they possessed the competence to draft them. The competence was real and the system held for years. It stopped holding when the first serious failure reached a table that was not a table of technical people, and at that point the defence that the standard had been met sounded like an admission, given who had written the standard. Nobody had lied. That was worse than if somebody had, because it removed even the relief of being able to point at a culprit.

The Fifth Amendment invoked more than a hundred times closes the sequence almost too neatly. With the shelter of definition exhausted, what remains is the procedural shelter, a full and legitimate constitutional right that costs a great deal more in public credibility than the one before it. Each successive refuge is more secure in law and more expensive in reputation, which is the real reason a defensive strategy built on controlling categories functions as debt. It is never extinguished. It is refinanced at rising rates, and the maturity date is set by somebody else at whatever moment suits them.

Internal risk committees reporting to the same executive line that approved the transaction. Sustainability self-assessments where the metric, the threshold and the verification originate in one office. Compliance reviews handed to the function that designed the procedures under review. Each of these arrangements looks efficient and competent and faintly virtuous at the point of adoption, and each is the same structure, with the party being measured holding the ruler.

The cost stays invisible through the normal years and presents itself exactly once, when a crisis arrives and somebody from outside asks for an account, at which point the technically correct answer has already been devalued by having been prepared in house. The negative lesson is close to embarrassing in its simplicity. When you are invited to write the rules by which you will be judged, that is not a concession granted to you. It is an unpriced transfer of risk in your direction, and you accept it because it has the shape of a privilege.

Fauci at eighty-five will not pay much, whatever the committee decides next week. The bill falls on the people currently in post who will have to draft the next framework, knowing they have precisely two options, to write it themselves and remain exposed, or to let it be written by somebody without the competence to do so. The dilemma does not resolve, and anybody presenting it as a question of personal integrity has misunderstood where the mechanism sits.