THREAD: Why I am skeptical of the government response to COVID-19.

(1) In most states, we have been locked down to combat COVID-19. Most of these lockdowns (not quarantines!) are justified based on experts’ models. But let’s look at one state’s models to see how flawed they are.

(2) My state of Ohio is using a model created by the Ohio State University. The model from 3/28 projected 10,000 cases per day right now *with mitigation*. This was five days after mitigation had started, so at this point there was limited, but some, data.

(3) What is the actual number of new cases per day in Ohio right now? It was 376 yesterday. That’s only 3.76% of what was projected. To use a term very popular in early March, that’s an *exponentially* bad model projection.

(4) Now, if the mitigation model is so far off, can we trust the unmitigated numbers? According to the model, Ohio was supposed to have 62,000+ cases *per day* on 3/22 without mitigation. (Note that Ohio’s lockdown began a day later on 3/23.)

(5) Yet the actual number of new cases on 3/22 was 351 cases—only 0.57% of the projected total. Somehow Ohio was magically saved from all these cases, even though we hadn’t started the lockdown. (See below where I address the possibility of many unreported cases.)

(6) On April 5th, two weeks into the lockdown, the model was updated. Now we were told it would be 2,000 cases per day by this past weekend *with mitigation*. Our actual 376 cases is still only 18.8% of that updated projection.

(7) Note that the 4/5 model still insisted Ohio would have had 62,000+ cases per day on 3/22 without mitigation. Although it radically (but not nearly radically enough!) decreased the mitigation projections based on reality, it didn’t change the original unmitigated projection.

(8) Some might object that either (a) there are way more actual cases than reported; or (b) Ohio citizens did mitigation efforts so well it destroyed the curve better than the models thought it would. Why are these arguments faulty?

(9) Regarding objection (a): If there are many unreported cases, that is a *good* thing, since it means the death rate is way lower than we thought.

(10) Right now there have only been 728 COVID-19 deaths reported in Ohio (population 11+ million), and only 17 deaths reported yesterday, with the peak day of deaths occurring on 4/22 with only 53 reported deaths.

(11) If there are many unreported COVID-19 cases, but still only 728 total deaths, then we have a very, very low death rate. That’s great!

Or perhaps there aren’t as many cases as projected. Either way, the model didn’t accurately project the actual impact of COVID-19.

(12) Regarding objection (b): If the models underestimated how well we would mitigate things, that just proves it’s impossible to truly model human behavior. You can’t project human behavior, and you especially can’t project the behavior of one group of people based on another.

(13) I’d say that a model that misses by so much is worse than worthless; it’s dangerous. Because of this model, over a million Ohioans are unemployed, schools are closed, religious services cancelled, people aren’t receiving medical care, and depression is likely on the rise.

(14) I can be understanding with how our governor (and most governors) initially reacted to COVID-19. If you are told tens of thousands of people will be getting COVID-19 every day, and 3.4% of them will die (the initial, but now disproven, estimate), you’ll take radical action.

(15) But it’s inexcusable how slow governors have responded to new data (not projections, but data!). In most places in America, the impact of COVID-19 isn’t anywhere near as bad as the projections. Great! Yet why are governors acting like the projections are still accurate?

(16) I suspect it has to do with a love of power, and a lot of face-saving. Governors fear that when people discover that they wrecked our lives with fundamentally-flawed models, the people will want to kick those governors out of office. And I hope they do.

/fin