About that modelling….

Readers may recall that I said the Burnet Institute’s COVID modelling from September 2021 should be revisited in late October to check the actual outcomes against the modelling predictions. It is now late October. The following actual data is sourced from covidlive.com.au

Predicted deaths: 964

Actual deaths: 280

Predicted hospitalisations: 1,666

Actual: 738

Predicted intensive care cases: 360

Actual: 130

What does this comparison of actual against expected prove? The modelling was wrong, of course. It was wrong by a factor of around 3 in these measures. Modelling is just a guess based on certain assumptions. Burnet’s modelling was clearly poor quality, but many modellers around the world have been proven to be just as bad or worse. The real lesson is that modelling is not gospel truth and should not be held up as justification for a government coercion on the populace. Nor should the modelling be used to prove how effective those coercive measures were: ‘See how much worse it would have been had we not imposed the lockdown?’ Modelling may have some uses, but should only ever be one input among many. Modellers that have a track record of dud predictions should be treated with scepticism or, preferably, ignored.