In 1775, Percivall Pott found a link between soot and scrotal cancer. The surgeon’s insight led to the 1788 Chimney Sweepers Act, probably the first legislation that could claim to be, in modern Whitehall lingo, evidence-based. In the two centuries since, politicians have learned to pick and choose the advice they follow, and use scientists as cover for awkward decisions—making the furore over government drugs adviser David Nutt’s sacking all the more curious.

Nutt has framed his October dismissal by the home secretary Alan Johnson as a conflict between politics and science. Referring to the 2008 decision to upgrade cannabis from a C to a B class drug he said on BBC Radio 4’s Today programme that “this is the first government that has ever in the history of the Misuse of Drugs Act gone against the advice of its scientific panel.” He also accused the government of a similar sin in continuing to label ecstasy as a class A drug, and of “making scientific decisions before they’ve even consulted with their experts.” He later wrote that “my sacking has cast a huge shadow over the relationship of science to policy.”

There have been two different assumptions underlying the way that Nutt has framed his own demise. The first is that the Advisory Council on the Misuse of Drugs (ACMD), from which he was dismissed, is a scientific body. The second is that drug classification is itself a scientific question. Both are open to debate.

On the first point, the ACMD is a “scientific panel” (to use Nutt’s phrase) only insofar as it is governed by a Whitehall code also used to run other scientific committees. But it had, until recently, 31 members—probably 20 more than you need to cover the bases on the medical effects of drugs, and fewer than half of them had a PhD. (Several members have resigned in protest at Nutt’s sacking.) Its broad remit, covering drugs rehab and education, reveals its true role: bringing together voices from the “drugs community.”

The second is more serious. Even the ACMD’s own advice, issued in May 2008, against restoring cannabis to class B doesn’t attempt to argue that science provides a definitive verdict on the drug’s relative harmfulness. But Nutt sees things differently, judging by a paper he published in the Lancet in 2007, and which he cited in a Guardian article written in the aftermath of his dismissal. This paper scores drugs on nine criteria of harm, ranging from “intensity of pleasure” (which affects how addictive a substance can be) to “healthcare costs.” The scores are added up, with the highest corresponding to the most harmful drug, with heroin being the most harmful, and cannabis rated 11th. The paper claims broad agreement between experts on the scores in each category, but this still leaves open the question of what the categories themselves should be, and how much weight should be given to each judgement where there can be no absolute answer. For instance, there is no compelling reason in Nutt’s ratings why the catch-all of “other social harm” is given the same weighting as “acute physical harm.” Such problems of methodology were raised by Nottingham University’s John Britton and colleagues, in a letter to the Lancet in June 2007. Britton pointed out that smoking kills more people than the other 20 drugs in the study combined, questioning “the validity, value, and indeed rationality of a system that ranks tobacco ninth in terms of potential for harm.”

The ACMD’s conclusions, which reference Nutt’s paper, can be fairly characterised as expert advice based on the evidence. But they are still in a different category to genuinely scientific advice—that the planet is getting hotter, for instance. Yet even the term “expert advice” overstates the strength of Nutt’s argument. The main effect of changing drug classification is to alter sentencing policy. This is clearly not a simple medical issue, and has wider consequences for policing and the justice system. So whatever the ACMD’s assessment of relative drug harm, there are other good reasons for changing classifications too.

Nutt does deserve credit for fulfilling his role as an independent adviser, and not allowing the politicians to gag him. But he has at times glossed over the differences between what science can tell us, and what his own views are on the policy. And while Nutt has seemed to have the backing of other scientists, this has largely been thanks to the media presence of Colin Blakemore, former head of the Medical Research Council. Blakemore is not entirely independent in this argument, being a co-author of Nutt’s Lancet paper.

It is easy to trot out Margaret Thatcher’s line that “advisers advise, ministers decide.” But it is nearly four decades since the 1971 Misuse of Drugs Act was passed and its regime has clearly failed. Drug use is ever more widespread, while ministers spend more time hiding than deciding. Within this, the Nutt affair is not really a case of politics vs science, but an attempt to use science to shape policy. Nutt wanted policy to put more emphasis on public health and less on policing, and science was his tool. When given a quasi-political role, such as chair of the ACMD, it turns out that a scientist can be almost as political as the politicians themselves.

Whitehall boasts a science minister and a chief scientist overseeing a network of government scientists, who in turn are scrutinised by two science select committees. But all this scientific advice too often dominates decisions in areas like drugs policy, that should properly be the domain of the social sciences, where the partial character of research findings is widely understood. So we need to rethink the language and institutions used to connect evidence to policy. In short, it’s time to move beyond Percivall Pott.