https://www.telegraph.co.uk/news/2021/12/21/covid-modelling-industry-biased-towards-doom/
The latest Sicron could kill up to 6,000 people-a-day. This arrives just in tiAGE doom modelling suggests that Omicron could kill up to 6,000 people-a-day. This arrives just in time to propel some members of the Cabinet into pushing for another “circuit breaker” lockdown. As the saying goes, the worst thing about a two week lockdown is entering its third year.
Models have played an outsized role in the Covid era. They have spurred ministers into some of the most authoritarian measures imaginable, causing immense human suffering and economic catastrophe. Now we learn that decisions are being made to take away our most fundamental freedoms on the basis of sexed up evidence.
This came from a chance Twitter exchange between Graham Medley, the chair of the SAGE sub-group responsible for modelling (SPI-M), and Fraser Nelson, the editor of The Spectator.
Medley reveals that modellers are presenting “scenarios” and “not predictions”. Things that could happen, but not the chances that they will. But, as American statistician Nate Silver points out, the “media treats them as predictions and governments use them to rally support for their preferred policies.” All Medley is doing by calling them scenarios is refusing to take responsibility for the policy consequences or accuracy of the models.
Covid models have consistently been based on extremely pessimistic assumptions. Their latest feature is to assume that Omicron will be as deadly as Delta, clearly contradicting the South African data. Medley claims it would not be worthwhile modelling a more optimistic scenario because it “doesn’t inform anything”. Somehow Medley believes that a scenario which indicates that the NHS is not about to be overwhelmed would not be enlightening.
“Decision-makers are generally on (sic.) only interested in situations where decisions have to be made,” he continues. “We generally model what we are asked to model.” The Nuremberg defence returns with vengeance, but only by raising more questions. Who is instructing the scientists to only develop extremely pessimistic absolute worst-case scenarios? Since these telling revelations, and with hospitality venues losing vital pre-Christmas revenue, it has emerged that officials have failed to model the economic impact of further restrictions.
Scientists should not be cheerleaders for interventionist policies by only presenting gloomy scenarios. They should present all scenarios, and their respective probabilities, enabling policymakers to make informed decisions based on costs and benefits of different actions. ‘Red teams’ should scrutinise all the assumptions and conclusions in real-time. Their results should be compared to reality and used to improve later models.
This is how much-maligned ‘finance bros’ operate. They model scenarios, make a prediction weighing up probabilities, and then compare results to real-life and throw out inaccurate models. There are clear feedback mechanisms. An analyst who gets it wrong will be responsible for losing money and would be sacked.
Unfortunately, in the case of SAGE modelling, there is no such accountability. In recent months the models, particularly those after ‘freedom day’ in July, have proven dramatically overly pessimistic. The UK has not even reached half the number of deaths of the most optimistic of the official scenarios. Yet nobody has been fired or faced negative reputational consequences.
The central challenge of modelling is complexity. The epidemiological models are both attempting to predict the behaviour of tens of millions, that is in itself rapidly changing, and how that will interact with a new virus or variant for which we do not have full information.
This isn’t a new problem. We have known for a long time that central planning inevitably fails because we do not know enough and would be foolish to assume otherwise. The Soviet models about how the economy would operate, where and when grain or steel was required, how much should be produced for whom, did not end up being the most accurate. Markets proved much more effective.
Nobel-prize winning economist Friedrich Hayek warned that: “knowledge of the circumstances of which we must make use never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess.” That’s why he preferred the price mechanism to send signals about needs and wants.
Unfortunately we do not have the luxury of price signals when it comes to public health. Accordingly, models aren’t entirely useless. Scenarios of what could happen are an important policy tool. But we must be extremely humble about what we do, and do not know. We cannot allow activists to use doom mongering modelling.
The least we can expect is for decisions to be made based on an impartial assessment of the most likely scenarios, balancing all the costs and benefits of different actions. That fact the system is so biased towards doom, unaccountable regardless of errors, and predisposed towards the destruction of liberty should send a shiver down the spine of anyone who prizes normal life.
Matthew Lesh is the head of research at the Adam Smith Institute
No comments:
Post a Comment