We haven’t heard much from Sir Patrick Vallance recently. A few weeks ago he wrote an article extolling the reliability of Sage modellers. They speak “scientific truth to power,” he said. He angrily rejected rumours of any negativity bias. Dozens of scenarios had been calculated for omicron deaths, he said, with a huge range of variables. What I’d love to ask him is why, if there was no bias, did every single one of these scenarios end up overstating the threat? Why were they wrong, so wildly wrong? Again?

This isn’t about parading Sir Patrick around Trafalgar Square with a placard of incorrect Sage graphs around his neck. The important question is whether he even thinks there was a problem, whether Sage is capable of error correction – and what he thinks about the fact that the country was very nearly locked down on what turned out to be seriously duff advice. Given that it could happen again, and at any time, are lessons being learned? Or are our scientific advisers still in collective denial?

In an era where all our lives are decided by the quality of epidemic modelling, the ability to scrutinise advice is vital. But Sage models are compiled within a wall of secrecy, protected from scrutiny. Their full figures are never published, nor is the code for them released. This makes error correction far less likely and constitutes a massive flaw in our democracy. “The scary thing is that Vallance has had more power than any of us,” says one Cabinet member.

But what if the error goes far deeper? What if lockdown itself was also built on a premise that turns out to be false? At the time, it was a huge, untested experiment. And two years on, the results are in – from countries, states and regions all over the world. One of the leading universities in the United States, Johns Hopkins, in Maryland, has just collated the data and its conclusion is startling: “We find no evidence that lockdowns, school closures, border closures, and limiting gatherings have had a noticeable effect on Covid-19 mortality.” In other words: an abject failure.

If true, this would be devastating. It would suggest that much of the pain lockdowns caused – the children denied education, ruined businesses, mental health issues, the undiagnosed cancers – was avoidable. And for what? The Johns Hopkins study finds that overall, lockdowns reduced Covid mortality by just 0.2 per cent. “Lockdown policies are ill-founded,” it concluded, “and should be rejected as a pandemic policy instrument.”

This isn’t to say that people should (or would) have carried on as normal without lockdowns. The great flaw in the theory was the failure to realise that, even without stay-at-home orders, behaviour adjusts. People hunker down. Mobile phone data now shows that Brits were doing this, even more than Swedes, before the lockdown order was given. The big post-Covid question is whether, in a high-information democracy, lockdowns are needless because people can be trusted to judge the risk, see the news unfold and act independently.

Might this new study find its way to Sir Patrick’s inbox? And might there be a committee somewhere in Whitehall carefully looking at this evidence to see if it is right? You can bet not. This is about politics, not science, and that has been the case for some time. A government that imposed three lockdowns – with huge financial and human cost – will have no interest in studies saying it made a calamitous error. Nor will Labour be saying so, given that Sir Keir Starmer was even more keen on lockdowns than the Tories.

At some point, politicians become so wedded to policies that they can never allow themselves to believe they were errors. Tony Blair will never accept that the Iraq war was a mistake, just as Margaret Thatcher never disowned the poll tax. But both had strong opponents, providing robust democratic challenge. This time, lockdown – in spite of its lack of scientific evidence – was backed by Left and Right, Holyrood, Westminster, Cardiff Bay. It is precisely in such consensus that the biggest mistakes in politics are most likely because there is no challenge, no inquiry, no one to identify mistakes.

This matters because there will, soon, be a new Covid variant. Genomic sequencing means we’ll start to detect new pathogens that might have gone unnoticed even a decade ago. If so, we’ll face the same questions: what to do? Can the healthcare system cope? The risk is that, having cried wolf so many times, Sage would not be believed even if its models were right. Track record matters. A recent Swedish book about the country’s refusal to lock down uncovered emails from health officials saying – in effect – that since Imperial’s Professor Neil Ferguson and his team got swine flu so badly wrong, their figures for Sweden’s Covid deaths would probably be incorrect too. (So it was to prove.)

Given Ferguson’s record, it was never clear why so much store was placed on his original suggestion that lockdown could potentially reduce Covid deaths by up to 98 per cent. At the time, even Sir Patrick and Sir Chris Whitty didn’t buy it. Both rejected lockdown – then realised, to their horror, that they risked being accused of causing an extra 20,000 deaths by failing to do so a week earlier. Even this figure came from Ferguson, and has since been debunked.

It’s not much of an exaggeration to say that our liberty, collectively, depends on Sir Patrick (or his successor) putting together a team capable of providing better epidemiological modelling.

A new group should be created, senior to Sage, that would check everything given then add in economic and social effects, to judge the overall effect of lockdowns. And ask basic questions: do these models assume people would not change their behaviour anyway? If not, why miss out such a basic point?

Denmark’s models got omicron right because they did adjust for behaviour. One bank, JP Morgan, got Britain’s omicron forecasts right because it used South African data. Britain has plenty of scientists who could have done the same – but they don’t seem to be in the right place. Next time, they should be.

The lessons on how to handle the next pandemic are all there: we just need a government capable of learning them.