Professor Lockdown, Legover, Pantsdown, call him what you will, may have stepped down as a government adviser (Professor Stepdown), or even been fired (Professor Crackdown) and ordered to keep a low profile (Professor Heads Down), but his pernicious modelling persists (Professor Stayed On), with its tens to hundreds of thousands of deaths (Professor Mow Down) and an overwhelmed NHS (Professor Meltdown), while he himself remains in post on full pay (Professor Drawdown).
Any lingering doubts as to whether Fergie’s ghost lives on to haunt the ‘collaborative scientific effort’ might be dealt with by reading this ‘public version’ of an undated letter signed by a self appointed soviet of twenty six numerologists. The way they see it, the science is imbued with the sweetness and light of collaboration and consensus, and Fergie should never have been put in the insidious position of having to step down. The way Dr No sees it, there is an even chance the letter, linked to by the Guardian, is a partial or complete hoax. Even if it isn’t, the notion that all scientists speak with one voice is arrant nonsense. Only three days ago, the Daily Telegraph reported that Johan Giesecke, the Swedish state epidemiologist, had described the UK’s lockdown policy as ‘futile’.
Nonetheless, modelling can be seductive to some. In the 2001 foot and mouth disease epidemic, Imperial College modelled predictions, to which Ferguson was a party, so seduced the then new Chief Scientific Adviser, Sir David King — now chief pongo at the Independent SAGE group, but then the head of a newly formed government appointed elite scientific group with direct access to Prime Minister Tony Blair — that he promptly and unequivocally adopted the 3km contiguous cull that lead to over six million farm animals being slaughtered. Terrible as that toll is, it was further aggravated by the wider effects of damage to livelihoods, mental health and the economy.
At the time, there was great tension between the tweedy veterinarian old guard, who favoured testing and indeed vaccines, and trendy new modelling vanguard, who said only culling would do. King didn’t bat an eyelid. Seduced by the modelling, he just knew what needed to be done. Rather than confuse poor Tony Blair with a menu of options, he presented the contiguous cull as the only game in town. ‘We had calculated a whole range of scenarios but I simply said that this is the one that will work,’ he later said. ‘So it wasn’t a matter of giving what I thought would be a confusing set of options.’ This wasn’t so much advisers advise, ministers decide as advisers decide, ministers preside.
Now there are at least three difficulties with what happened back then over foot and mouth disease, and what is happening now over covid-19. The first is conflating modelling as an exercise in fitting a model to the data, and modelling as an exercise in forecasting, as if they are one and the same thing. We can see this if we take a simple example. Imagine a sailor planning a passage across the Channel. He knows, from past measurements, what progress his yacht will make good in various wind conditions. Though he wouldn’t be so pompous as to say so, he has a model for his boat’s performance in different winds. He has fitted a model to the data. So far, so good.
He then uses the model to forecast his progress during his planned passage, and so establish an ETA. Unfortunately, his model only incorporates some of the factors that will affect the passage time: it is an over-simplification of the world as it really is. As any sailor knows, an ETA is just that, an estimated, or forecast, time of arrival. Only very rarely does reality bear out the forecast. The model may fit the known historical data, but it gets more and more wildly erratic the further it is used to predict the future. We see this on a daily basis in the weather forecasts, generated as they are by some of the most complex models know to man on computers of mind bending capabilities.
The sailing example can also be used to show the second defect in modelling, which we might call blinkering, or a tunnel vision effect. This is a variant of if the only tool you have is a hammer, then everything else becomes a nail. Because the model focuses on the yacht’s performance, other options get excluded. For example, if the sailor really needs to be on the other side of the channel by a certain time, then do the obvious – take the ferry. In the same way, modelling culling for foot and mouth disease, or lockdown for covid-19, has the effect of squeezing out consideration of other options, of which there are many. But they don’t get a look in.
The third problem is the power of models and their conflated forecasts to seduce the scientists so completely that the scientists, who at the point they started forecasting stopped being scientists, becomes advisors on a mission. It happened in 2001 over foot and mouth disease, and all the evidence today suggests the same thing is happening all over again for covid–19.
Happening all over again, only more so. We have seen how the Ferguson forecasts, with their compelling case for hard lockdown, became the only game in town, in the same way that the contiguous cull became the only game in town in 2001. But something else has changed. The advisor-minister axis has been tilted. Back in 2001, King fixed Blair by using the Model T Ford car sales trick. The advisor might have decided, but the minister could still preside, at least in appearance. This time round, the government has forfeited any capacity for independent thought by adopting its slogan ‘led by the science’.
At first glance led by the science seems fair enough, in a what else would you expect sort of way. But look closer at the construction, and what we see is a verb being used in the passive voice. But if we use an active voice, we can see the truth: the science leads the government, just as the master leads the pupil. That would be tolerable if the master was benevolent, but what if the master is a delinquent, in thrall to wild forecasts? Is it not the responsibility of government to take an active role, by taking in the whole picture, to see the modelling forecasts as just one wave in a sea of advice, and then remember the old way: advisors advise, ministers decide.
Footnote: Dr No can’t help adding that even the CSA now agrees with this. Perhaps belatedly realising ministers planned all along to use scientific and medical advisors as human shields, and that he, Sir Patrick Vallance, was first in line to ‘take one for the team’, the CSA told the Health Select Committee last week that ministers are ‘informed by science, not led by science’ — see 1m44s in this ITV video. Boris take note.