[DataDive]

But if you look at the number-crunching process that gave you those percentages, it turns out to be woefully fragile and sensitive to a whole range of other factors

Countless prior models were completely off, but this one is suddenly correct and needs to be treated as a reliable predictor.

The problem is that if you dare to question the accuracy, you are immediately labeled a "denier." The only acceptable answer is that no matter what the prediction is, if it is alarmist, it is correct.