Appearing before the UK’s COVID-19 inquiry in November 2023, our ex-General Chief Scientific Adviser (GCSA) Patrick Vallance was invited to comment upon the Johnson administration’s — including the eponymous Johnson himself — scientific and statistical literacy. 

Vallance’s private diary from the lockdown years, submitted to the inquiry as a piece of evidence, reported that Johnson was “clearly bamboozled” by the slurry of data and graphs being presented to him by his scientific advisers to make sense of the pandemic. According to Vallance, Johnson struggled with basic epidemiological concepts (“He finds relative and absolute risk almost impossible to understand”, “[He] struggled with whole concept of doubling [times] … just couldn’t get it”) and that, on the whole, “watching [him] get his head round stats is awful.” 

The Inquiry’s counsel (Mr Connor) was fastidious in underscoring the significance of Johnson’s skills-gap, asking, “Mr Johnson, it hardly needs saying, was the man who was making decisions that had incredibly broad impacts on the whole country, and it was critical, was it not, that he did understand the advice that he was being given?” “Yes.”, Vallance responded. And the UK, it seems, was not alone in being led by a scientific illiterate. Vallance recounts: 

Well, I think I’m right in saying that the Prime Minister at the time gave up science when he was 15, and I think he’d be the first to admit it wasn’t his forte, and that he did struggle with some of the concepts, and we did need to repeat them often. I would also say that a meeting that sticks in my mind was with fellow science advisers from across Europe when one of them, and I won’t say which country, declared that the leader of that country had enormous problems with exponential curves and the entire phone call burst into laughter, because it was true in every country. So I do not think that there was necessarily a unique inability to grasp some of these concepts with the Prime Minister at the time, but it was hard work sometimes to try to make sure that he had understood what a particular graph or piece of data was saying.

Even pre-COVID, Vallance (like Dominic Cummings, Boris Johnson’s erstwhile Svengali and a key player in the UK’s COVID-19 policy) was already on record as arguing in favour of programmes to boost the UK government’s scientific and statistical competence. In the wake of his testimony, he has been joined by other voices including the Royal Statistical Society (RSS) who, in the run up to the recent general election, have published a manifesto and sent letters to most of the main parties’ leaders demanding that parties to ensure that their members and prospective ministers to receive statistical and scientific training. This, the RSS says, means giving ministers an understanding of how to interpret the ‘data-based evidence’ being presented to them by their scientific advisers and (echoing Vallance’s remark that “…it is entirely appropriate for decision-makers to challenge science advice…”) of what sorts of questions to ask.

On its own terms, Vallance and the RSS’ proposals are reasonable given the nature of modern government. Per Michel Foucault and his heirs, complex scientific knowledge and the modern state’s exercises of power weave around one another like the snakes on Hermes’ staff as, in making new parts of the world legible (or old parts legible in new ways), the former makes possible the latter while the latter justifies the former’s continued expansion and funding. It is a matter of practical reality that actually making economic policy requires grasping somewhat rarefied macroeconomic notions like “aggregate demand”, “credit swap lines”, or “asset bubbles”, just as making public health policy during a pandemic requires some understanding of “exponential growth”, “doubling times”, and “absolute/relative risk”. As such, Vallance’s revelations aren’t just embarrassing – they also suggest that our politicians are limited in their capacity to do their jobs, representing and effectively intervening into the world on behalf of our needs and interests.  

Their proposal, however, does not go far enough and its failure to do so stems from the establishment’s (as represented by Vallance, the RSS, and the inquiry more broadly) systematic misunderstanding of COVID-19 decision-making. They (the establishment, that is) still see lockdown as a legitimate policy approach, and so continue to draw the wrong lessons about science’s role in COVID-19 policymaking. To them, the Johnson administration’s central error was not imposing lockdown early or hard enough, and, as Vallance’s testimony epitomises, they identify ministers’ inability to understand the science being presented to them by their epidemiologists and virologists as having played a driving role in this. Once, however, you see lockdown for the breathtakingly predictable catastrophe that it was and examine decision-making processes that made it possible, it is clear that, in COVID-19’s wake, scientific literacy amongst our political class is not sufficient — we must also demand that they become critics, even sceptics, of science. 

I am not, to be clear, suggesting that our politicians be plucked from a pool of astrologists and chemtrail-truthers (not least because it is a stretch to describe either of these as “critical” or “sceptical”…) but instead that those elected gain a shrewder understanding of a scientific claim’s contents. To see why, briefly consider the Janus-faced role that science and scientific ideas actually played in the UK government’s COVID-19 policymaking. 

Epidemiological concepts and modelling made the world legible to ministers and, drawing on work previously done by a number of the government scientists and their colleagues across the Atlantic, provided them with a rationale for imposing sweeping lockdowns. However, in doing so, epidemiology’s concerns came to dominate the government’s decision-making process, effectively side-lining concerns that were difficult, even impossible, to express in epidemiological terms. Lockdown policy was packaged and delivered in slogans like “lowering the R-number”, “flattening the sombrero”, or “saving our NHS”, each of which reflect a variable on an epidemiological model. Lost from view were things like “learning losses”, “loneliness”, or “the importance of normality” (the latter of which, by the way, earlier pandemic influenza plans had acknowledged). For a while, we were made to live, breathe, and die by the epidemiological model and everything else was treated as secondary or irrelevant.   

We systematically ignore how scientific claims often contain a normative judgement about what matters to human life

Now, there is a confluence of reasons explaining why epidemiological modelling was accorded this level of importance, including institutional (e.g., many of the government’s most prominent advisors, including the Chief Medical Officer Chris Whitty, were epidemiologists) and political (e.g., once presented with graphs showing the imminent “collapse” of the NHS, it was all that ministers could focus on) ones. However, underlying these was also a reason that we might describe as “ideological” and pertains to the particular significance that we, in the modern, post-Christian West, accord to scientific claims. In his final, posthumous collection of essays, the physicist, philosopher of science, and professional gadfly, Paul Feyerabend observed that (1) we systematically ignore how scientific claims often contain a normative judgement about what matters to human life and practice and that (2) we tend to treat scientific claims as unique or privileged sources of information. Let’s briefly consider each in turn.

Feyerabend argues that when science makes claims about what is real “out there” in the world it is, in effect, often also making claims about what human life and practice should be organised around. To take an almost farcical example of this, consider scientists like Richard Dawkins who challenge religious claims about miracles and angelic visitations with scientific claims about what is, in fact, real in an attempt to break people from their religious practice (it is not clear, however, that Feyerabend’s observation holds for all scientific claims — quantum physicists who say that macro-objects like tables or chairs “aren’t real” don’t often expect us to start living accordingly…). In the context of policymaking, scientists make claims akin to Dawkins’ — in giving a description of what is “out there’’, they present ministers with an account of what particular things government should be focussing on and organising its policymaking efforts around. In providing ministers with models that described the world in terms of “reproduction numbers”, “mortality rates”, and “ICU-bed demand”, epidemiologists provided ministers with the focus of their policy. 

This alone, however, does not explain why epidemiological modelling became COVID-19 policy’s sole basis. To understand this, we need to turn to the second of Feyerabend’s observations where he argues that we tend to think of scientific knowledge in quasi-monadic terms, i.e., that there is only one science and even only one type of genuine knowledge. As such, when science makes claims about a particular domain, we tend to behave (like the UK’s ministers did in 2020) as though they are the most, and even only, relevant sources of information about that domain. Of course, this in turn does not explain why epidemiology out-competed other sciences like economics (which, for its part, was already warning about the possibility of a recession) and to do so we need to extend Feyerabend’s analysis further than he does in his book. Looking back at the early months of 2020, it is clear that, just as there exists a hierarchy between scientific and non-scientific knowledge, there also exists a hierarchy between the highly quantitative, computer-modelling-based pro-lockdown arguments and more qualitative or historical claims. As the historian Toby Green wrote in 2021, 

… it is worth noting that the scientists who have spearheaded the lockdown polices have been computer modellers. […] these models were attractive to governments which […] were already strongly drawn to data-driven models of policy development. The imposition of the initial lockdown policies thus emerged from the privileging of these computer-simulated models over the experience of medical history in the treating of new pandemics.

Epidemiological concepts and models captured pandemic policy by, inter alia, (1) presenting ministers with an account of what they should focus on and (2) engulfing the minds of men already prone to hard science-worship and data-addling.  

The implications of COVID-19 policy for ministers’ scientific education, then, go far beyond the ones drawn out in the RSS’ letters. Alongside greater scientific and statistical literacy (which, as I say, is necessary for politicians to do their jobs), there needs to be an ideological shift of sorts, a grand demystification of the scientific enterprise. Ministers have to be better at reading and interpreting graphs certainly, but they must also learn to see them as partial and leading and in need of systematic balancing against other, competing accounts of what matters. Pursuing the former without the latter risks simply making ministers ever more uncritically receptive of scientists’ narrow worldviews and disciplinary priorities. To continue reaping the benefits of science-based policy, while avoiding its terrible harms, science needs to be seen for what it is — useful and (often) fascinating but reductive and (equally often) value-laden. 

For what it’s worth, I would have voted for any party promising something akin to this.