Disease models

Disease models remain worthless, on par with astrology with the honesty of alchemy.

Daily Cases Forecast

Color bands represent 95% confidence interval for each model used. Many of the models were not even in the same planetary system as earth.

Their ensemble (arbitrary averaging of the models) did not have the sign correct showing an increase as the real world went down even sharper.

And here’s the ensemble projection in mid January

One model had forecast over 50 million new weekly cases – causing the skewed Y-Axis here

Deaths Forecast

Official disease model output from the CDC for forecast deaths with 95% confidence intervals shown. The white dots are the “ensemble” (they average all the models together). Again, the sign of the ensemble was in the wrong direction.

The only purpose for disease models is to generate scary news headlines. The disease models otherwise have provided no useful, actionable output.

In the spring of 2020, the Neil Ferguson’s ICL model projected 2.2 million dead in the U.S. by August of 2020. That was the basis for locking up the entire “non-essential” civilian population for months.

Simultaneously, the UW IHME’s model projected a massive shortage of ventilators in New York, a shortage that never materialized. In response, the taxpayers spent billions to manufacture more ventilators by summer time – of which none were ever used. The ICL and IHME models are the definition of useless.

Disease models have the accuracy of astrology and the honesty of alchemy.

In my state, from late October to end of January, they used short 3-week model projections. Even then, in 11 out of 13 model runs, they had the sign wrong. Tossing a coin would have been more accurate.

Disease models are worthless.