Or you can watch it and say, "What in this supports my views?"
It depends how you feel about being wrong. If it's perfectly okay to admit screwing up then the first is easy. If you don't like it when somebody comes up and tells you, "You screwed up, and here the proof," then you're likely to prefer the second.
Here are some problems.
He points to a line with negative slope. He points to variations in the negative slope. And says that a variation demonstrates the effectiveness of lockdowns. The question is, if the slope was negative, would it have continued to be negative and just take a day longer to get to the same point? What's causing the curve to flatten and go negative in the first place? He doesn't ask those questions--which boil down to, "Are the lockdowns effective?" He's asking, "What evidence do I have to support my views?"
That's the critical thinking problem.
There's also the data problems.
1. The original Chinese data, as far as I know, are when tests are reported. He says, and it's known, and the Chinese government adapted for, the fact that the data for reported positives always lag the dates the test samples were taken. He includes an offset to account for something like the mean time it takes to report tests.
That's China. In the US, the CDC and some states are really clear--they say that the positive tests reported for a given day will be adjusted in the next couple of days as tests come back. The # of tests done per day is accurate, but they backdate positives and negatives and post them on the sample-taking date. This is slightly in favor of his argument, because it means that the offset to get from reported positives to actual onset of symptoms could be less than he's assuming. Not all states do this because, really, in the end it's not a big deal. And I expect that whatever they do for public consumption they give the good data to epidemiologists. When they were altering (publicly) the reported positives by changing numbers that were 2 and 3 days old, people's paranoia took over. (Really, "paranoia runs deep, into your soul it will creep" is a valid statement.)
Don't know about other countries. Even the US data isn't all that clear to me what's reported. Over the course of a week it all averages out, but his analysis crucially requires a reasonably accurate day-by-day reporting. A couple of days off? Sure. But not knowing if your reported data are delayed or not ... Can't account for what you can't know, not when you're looking at phenomena that requires that level of accuracy.
2. Then there's the whole "double your test rate, double your positive rate." Let's say I set up a testing station the same day you do, on opposite sides of the street. I charge $1. You give each person $5. I test the 100 who thought no wait was worth $1 and I get about 50 infected people; you test 1,000 and you ID 500 infected people. Gee, my side of the street only had 50 sick people, your side had 500, I don't want to be on your side of the street! Or, maybe, I can say that the rate on either side was the same--50%. But is that the real prevalence in the population? No, because if you're not sick you're not likely to want to pay $1 or stand in line.
So the # of positives isn't that meaningful *unless* the number of tests stays perfectly constant AND there's a large number of tests. US data meet neither of the necessary conditions.
Oddly, after watching this I'm less convinced that the statewide lockdowns account for all the data. The more draconian the lockdown. Important? Sure. Very helpful? Absolutely. But the one, saving thing that we look to for survival? No. It cuts down on the religious fervor associated with them. Is social distancing important? Yeah. But that's not a lockdown, and some examples of enforcement are just place insane. A picture of families walking near each other--no, no, you may be cooped up in an apartment for hours each day, but in public you need to be 6' apart. Couples holding hands? How horrible! Yes, you sleep in the same bed, but what's really important is the distance for 30 minutes in the park.
The real tests will be several. First, what happens in some red states--the doom-and-gloom predictions are still being made, but the early ones were wrong. Even "case loads are still soaring in these areas" statements sometime represent the arrival of 20 test kits and 8 new positives. Small number stats. One show pointed out how the rates had risen--for three days, after a weekend slump in reporting and a sharp drop on the day of the report. The rise in rates were an artefact of the reporting mechanisms, nothing real on the ground. The death toll for that day in that state, wracked by disease and under incompetent management was 3 (as opposed to 600 in a well-managed state with only 30x the population ... which would lead us to expect 90 dead, still not 600).
Second, what happens in states that significantly relax lockdowns but keep social distancing? (They're different, but I already see people hedging their bets and merging them.) I mean, spring break didn't move the curve as predicted, and CTers have to claim that the death toll exploded and it's all being covered up.
Third, Sweden. Yes, it's much nastier than in Norway and Finland. Yippity-do. What matters isn't the deaths between now and August but between now and 2022. Meaning that ultimately everything is speculation. It's like a theater critic trying to submit his review of a play before he's finished seeing Act 1, but still criticizes the ending. Or perhaps giving the post-game wrap up and declaring the victor in game 7 of the World Series before the first inning is over and the score's 1-0.
= new reply since forum marked as read
