Submission

I welcome the opportunity to make a submission to the Covid Response enquiry. I am a private citizen, and was 69 years old at the outset of the pandemic. In 1977, aged 25, I suffered an attack of poliomyelitis, which nearly killed me, left me extensively paralysed, and importantly for present purposes, gave me a ‘victim’s understanding’ of viral infection, and the threat it poses to individuals and to society. I have no other relevant qualifications.

So many errors were made in responding to the Covid outbreak that it is impossible for a private citizen to cover them all – that would require the writing of a book. For that reason I have decided to confine myself to one area of fallacy which seems to me to bedevil policy-making in the wider sense – the unquestioning faith reposed in computerised numerical modelling.

From the start of the Covid outbreak, claims were made for its lethality which turned out to be exaggerated by orders of magnitude. As the panic developed, policy-makers, claiming to be ‘following the science’ were influenced by computer-modelled predictions, in particular the work of a group of epidemiologists at Imperial College, London. This group, led by Prof Neil Ferguson, forecast, horrendous mortality unless draconian, and unprecedented non-pharmaceutical interventions (hereafter referred to as “Lockdown”) were imposed on the entire population. Computerised modelling, and notably that of the Imperial College team, significantly influenced Australia’s Covid response.

Helpfully, Ferguson et al applied the same logic to the case of Sweden, forecasting mortality of a similar order should that country, as it appeared inclined, refuse to impose the coercive lockdown measures recommended by Ferguson et al.

Sweden has experienced Covid mortality no worse than its locked-down neighbours, and appears to have avoided the surge in excess, all-cause mortality which has afflicted the ‘locked-down world’. It has achieved this admirable outcome without paying the grievous social and economic cost of lockdown.

Had policy-makers in the UK, and here in Australia enquired into the history of Prof Ferguson’s modelled predictions, they would have discovered a string of vast over-predictions of contagion, going back to the early days of widely-available personal computers, and including Swine Flu, Bird Flu, Bovine Spongiform Encephalitis and Foot and Mouth Disease. In each case, his predictions exaggerated the threat by orders of magnitude.

Computer-modelling can provide useful tools for policy-makers, but it must be treated with great caution, and is never likely to perform the comparative evaluation of social, economic and health risks which must remain the responsibility of politicians. Modellers who over-predict the severity of a risk rarely, if ever, suffer damaging consequences, whereas underestimating a risk is likely to damage their careers. Furthermore, those careers are sustained by their producing alarming predictions, to which they can purport to propose a remedy. There is thus a huge imbalance of moral hazard built into their field. It beggars belief that any responsible policy-maker, in the UK or here in Australia, aware of this litany of failure, would have given any serious credence to Ferguson’s Covid modelling.

Above all, policy-makers, when urged to a costly course of action by scientists using models must, in future,

• Require them to show evidence that their models have predictive skill.
• To avoid ‘cherry-picking’, require them to show the outcomes of ALL their past work which has had policy-forming outcomes.
• Require them to reveal their modelling assumptions, and share the code/algorithms used to model them.
• Set up a ‘red team’ of competent numerical modellers to scrutinise and interrogate the work of the policy-proposing modelling team.
• Remember that computers cannot model disparate values such as schooling lost or social isolation. These are, and must remain, the province of politicians.

Tom Forrester-Paton,

December, 2023