There's also something called "reversion to mean"- if you act on outliers, you're basically guaranteed that the numbers will make it look like your actions are effective, because the outliers pretty much always revert to the average just because that's what outliers do.

The usual example is putting a new stoplight on the intersection in the city with the most accidents in the last year. Then next year, the number of accidents goes down. Must be that stoplight, right? Except it was probably going to go down anyway, because the intersection with the most accidents in the city last year was likely a statistical fluke to begin with, and it's average number of accidents is lower.

And we see it both ways with the wuhan virus. A state has a record high number of cases last month? Better make up a bunch of bullshit restrictions, things are getting dangerous! Then next month, when it's lower, it must be because those restrictions worked, not that it was just a coincidence that it was that high to begin with, and it would have come down anyway. And the other way, where there's a sudden dip- well, I GUESS we can remove a few restrictions... oh no, next month the numbers went up, better put all the lockdowns back in place!

Record lows are usually followed by an increase, record highs are usually followed by a decrease, and both those factors encourage idiots to have more lockdowns and convince themselves that they work.