Relying on computer models to forecast events is a precarious business, as we’re now being reminded.

By Peter Murphy~

Computer models compiled by scientists, statisticians and public health experts to predict the number of deaths resulting from COVID-19 have been drastically scaled back this past week. This is hopeful news, but has wider implications. There also should be a serious look-back, given the wildly inflated early predictions of numbers of deaths in the United States.

Computer models are only as good as the assumptions built into them. If the inputs are faulty, the predictions will have shown to be flawed based on real life outcomes. This is playing out with the coronavirus models, and wreacking economic havoc worldwide. This modeling problem has ample precedent.

Flawed computer models have long been rampant in predicting planetary global warming for at least the last 30 years, even as they continue to influence public policy. Perhaps the most famous falsehood was the “hockey stick” prediction of rapid…