I need to post more frequently.
Hope you’re sitting down for this: enshrined in English and American tradition and laws is the idea that it’s not always a good thing to do what the experts tell you to do. This wisdom was hard-earned over centuries, as experts were discovered to be – still sitting down, right? – alas, human beings, subject to all the temptations, all the weaknesses, pride, vanity, greed, and fear that all other people are subject to. Therefore, when really important, life and death decisions come up, we don’t defer to experts.
We call this wisdom ‘the right to trial by a jury of your peers’. We do this, even though there is always a judge and a couple other lawyers, at least, right there in the room, with years of legal training and experience, who know the law far, far better than any of the jurors.
The commoners, the untrained laymen, are made to listen to these experts, certainly, but the decision is ultimately theirs.
Being people, the jurors will sometimes screw up. Being people, so will the judges. But random jurors assembled for a specific job and then dismissed once the job is done are a lot less likely than professional judges to think expertise gives them the right to boss people around.
Perhaps we should generalize this principle a little bit? We could call it representative democracy.
I was sent for comment an article to which I will not link, published 10 days ago, where we little people are told, among other things,
- Our healthcare system is already collapsing.
- Countries have two options: either they fight it hard now, or they will suffer a massive epidemic.
- If they choose the epidemic, hundreds of thousands will die. In some countries, millions.
- And that might not even eliminate further waves of infections.
- If we fight hard now, we will curb the deaths.
piece of blatant propaganda article ends with a call to action:
If you agree with this article and want the US Government to take action, please sign the White House petition to implement a Hammer-and-Dance Suppression strategy.
It begins with an appeal to our herd instincts: all the cool kids are descending into terror, we should, too:
This article follow… with over 40 million views and 30 translations. If you agree with this article, consider signing the corresponding White House petition. Over 30 translations available at the bottom. Running list of endorsements here. Over 10 million views so far.
This masterpiece is suppose to be the work of “normal citizens”:
This article has been the result of a herculean effort by a group of normal citizens working around the clock to find all the relevant research available to structure it into one piece, in case it can help others process all the information that is out there about the coronavirus.
The normal citizen under whose name this work appears has an MBA from Stanford, an engineering degree, and heads up a billion dollar company. Doesn’t everybody? Just your average Joe.
We are presented with a bunch of pretty charts backing up the author’s terrifying claims. These charts are the product of a model. This model. Using this model, the author predicts that around 200 million cases of COVID 19 in the US alone, with around 11M deaths, UNLESS WE TAKE DRASTIC ACTION NOW! While I’m not an epidemiologist nor even a doctor, I am a numbers guy and do know models. So I looked it over, played with the assumptions. Here’s what I found: the model assumes
– that 100% of the US population is susceptible to the virus. Hmmm – I’ve never heard of an infection to which nobody is immune, even in theory. In practice, we all know people – my wife is one – who get exposed to every bug under the sun (e.g., as a school teacher) and never get sick. So COVID 19 is assumed to be worse, in some sense, than all the other stuff floating around at all times.
– that the CASE fatality rate – 4.5% – is the mortality rate. This weird, if not dishonest, assumption is also made here, where I go for my only-lightly-cooked data: “*Death Rate = (number of deaths / number of cases) = probability of dying if infected by the virus (%).” This is of course ridiculous, and a freshman-level logic error: your chances of dying if infected by the virus is (number of deaths/number of infected persons). Only in the strictly theoretical case where every single person in the world is tested such that ALL ‘cases’ are known, would the case rate equal the true mortality rate. In the real world, the case rate will be higher, most often much, much higher, than the true mortality rate, as infected people with minor symptoms or no symptoms at all are unlikely to become cases. Such people might vastly outnumber the people who do become cases. There are other problems with this approach as well.
– that each infected person will in turn infect another 2.5 people on average. This is known as the R0 of a disease. Now, if you knew how many people were infected at some point in time, and knew when the first infection took place, you could back into this number more or less. Lacking that basic data, which we do lack, this number is basically pulled out of thin air.
I stopped after doing some what-iffing with just these three assumptions. The percentage of people susceptible and the R0 seem to me to be functions of each other to a large extent. The first acts as a cap to the total number of infections – you’ll never get more than everybody infected – while the second is more of a speed thing when coupled with another variable, the numbers of days an infected person is contagious. Together, these variables determine how many infections will occur.
The model is highly sensitive to the fatality rate and the R0. As R0 approaches 1, the epidemic collapses – the spread is so slow as to be invisible, as each person infected infects only one person before recovering or dying. Similarly, reducing the death rate from the ridiculous 4.5% to a more believable rate of 0.5% (correction – I mistyped .05 instead of 0.5 when first posted. Meant 0.5%) or less, and the number of dead falls rapidly. Do both these things, and the projected US deaths fall under from 11 million to a few hundred thousand. Assume that only a relatively small percentage of people are actually susceptible to the virus – on the Diamond Princess, only 18% of the passengers and crew got the disease, even though they were crammed together on a cruise ship for days – and the ceiling for infection falls. (I initially thought it might be so simple as to reduce the total number of infections by 82%, but that doesn’t sound right on reflection.)
In other words, take extreme numbers as inputs, and the model will dutifully spit bad extreme numbers as output. Use more realistic inputs, get much less panicky outputs.
Yet the article treats the extreme case as gospel, and so makes exactly the sort of dramatic claims using emotional language that is the hallmark of anti-science:
- Our healthcare system is already collapsing. No, it’s not. “Collapsing” is a scare word. What is happening: hospitals in New York City and a few places in Northern Italy have more COVID 19 patients requiring hospitalization than capacity. Resource allocation problems, while real and a potential tragedy for some patients, is not ‘collapsing’. In a few weeks, the issue will pass, based on what happened in Wuhan (which has your basic totalitarian communist socialized medicine situation. Healthcare is not fungible.)
- Countries have two options: either they fight it hard now, or they will suffer a massive epidemic. Classic Trolly Car Problem forced binary. First, assume outcomes are known – either A or B (and not C through Z) will, not might, happen; then, install a switch: You get to choose! Between A or B! In reality, Japan and Sweden chose a slightly modified A – and nothing much happened; South Korea chose B-Lite, and nothing much happened. Italy and New York City chose A then B by way of some other letters, and it got bad, although not nearly as bad as the article implies, and it’s already getting better – the trends are down in both those places; peak infections are, if not already past, are anticipated to be so within a week or two.
- If they choose the epidemic, hundreds of thousands will die. In some countries, millions. Not really. See above.
- And that might not even eliminate further waves of infections. We have the god-like ability to stop future infections? I understand the logic here, but disagree with how much agency we grant to people versus nature doing what nature does. The article give us much more agency here than prudence would dictate.
- If we fight hard now, we will curb the deaths. Well, some places fighting hard, after the fashion laid out in the article, such as where I live, and there’s nothing like an epidemic here. Some places, such as Japan, hardly did anything at all, have been infected longer than the US – and have 54 total deaths, in a population of 125M that skews old. Magic? Other places seem to have different experiences all over the board. Maybe it depends on other variables? Maybe this model assumes a homogeneity that doesn’t exist in reality?
Ultimately, models are only as useful as their ability to produce useful predictions. You test models against reality, not the other way around, and – this is the hard part – do not give models much standing until and unless they’ve proven useful. The question with this model, as with any model: has reality backed it up? Are things turning out as predicted?
Pretty much ‘no’. As it is now, it looks likely that total COVID 19 deaths in America may peak not at 11M, not 100K, but more like 10K. Unless all hell breaks loose somewhere here outside the hotspots of NYC/Newark, Washington state, Luisiana and three spots in California, total deaths may not hit even half that.
Nothing like the author’s B scenario took place; his predictions even if we did the full fascist lockdown on the day his article was published was for far more deaths than that.
All this ignores the elephant in the room: COVID 19 doesn’t seem to kill anybody – the death total so far is of all deaths where the deceased has or is even suspected to have the infection, regardless of any other conditions. There are only a handful of cases worldwide where some seemingly healthy person in the prime of life got infected and died without any other underlying ailments. The prudent thing would be to wonder if there wasn’t something else going on even in those unusual cases, rather than assigning all deaths where COVID 19 was involved or suspected as epidemic deaths.
This epidemic is 100% propaganda. All around the world, those in love with centralized control have been suffering setbacks; here was an opportunity to reassert themselves that could not be passed up. Plus, the incoherent babbling of Biden and the criminal activities of his son, among other unpleasantries, have been driven from the ‘news’. Just the way the chips fall, I suppose.
19 thoughts on “Monday Thoughts (yes, the d*mn virus)”
I was waiting for you to comment on this and you didn’t disappoint!
Also, it isn’t even really fair to compare…anywhere in America with Italy – sorry, Lombardy – since *even now* we are nowhere near as bad.
But write this down. Never forget. *Best case scenario*, 100,000 *American* deaths.
Don’t let that number slip from your mind.
I’m not sure if you heard, but they may not be that bad, either. Even if you ignore the whole refusing-to-treat-those-with-any-existing-conditions-including-being-retirement-age thing that will make their numbers much worse.
Their system reports anybody who died while infected as a kung flu death. One of their officials pointed out that, on review, only about 11-12% of the deaths were probably contributed to by having it. (And there’s a really big problem with it spreading in the hospitals, too.)
When i was in graduate school, one of my professors told the class that if a cancer patient died of anything, the cause of death was considered to be cancer. (Except for being hit by a truck. I asked.)
Now it’s looking like they’re doing the same thing with coronavirus. (Maybe without the “hit by a truck” exception. I haven’t found anyone to ask.)
Right. At the rate this epidemic is raging, we might hit 10,000. (That a huge % of those people sadly would have died this year anyway from their preexisting conditions is never mentioned…)
.We are already at 7000+ with > 1000 per day for the last 4 days. Hard to imagine the total will not greatly exceed 10,000.
You’re right. I was optimistically counting a couple spiky days in deaths to be a sign of the coming peak; but, at least in NYC, that was premature. See, for contrast, California, where our Governor went farther faster, yes, but also where it is sunny and warm. IF – this is the one big if I’m hanging this analysis on – COVID 19 behaves like a typical air- and surface-bourne virus, sunny, warm weather should put a stop to it; New York is hitting it’s spring weather about now, so transmissions should greatly slow.
I always say: I could be wrong, I’m just a numbers guy, not an expert. But I like using history as a template.
About 10 days ago on Twitter, I picked up a couple panicked replies to my failure to panic myself. I offered an over/under bet, but got no takers. At the time, I picked 330K dead, which is the population adjusted American death toll for the 1957-58 Asian flu. Wonder if anyone would have taken me up if I’d gone 100K? Looking pretty good now.
I’m actually not a gambling man, just wanted to see if anyone would put their money where their mouths were.
New York City chose A then B by way of some other letters
I’d disagree on this, actually.
How would you describe “took B and then completely inverted it so that they actively used social and legal pressures to get people to do things that are moronic in even a basic flu season”? Isn’t there some kind of notation for “inverse-B”?
Maybe “Bizzaro B”?
(Yes, you can guess that I am steamed about the “#WuhanStrong”, avoiding folks who’ve traveled internationally recently is raaaaaaacist, no steps to improve safety in the subway, ordering sick teachers to go to school, etc.)
You have a point. I’m giving them (undue) credit here. But putting it your way is almost better: the correlation between taking extreme steps and overall severity is even murkier in NYC and Newark, given the Keystone Kops level implementation.
I will say I dislike Murphy (our governor), but I must admit, I actually think he has done a decsnt job: He actually loosened some restrictions on the “shutdown” (bike shops are “essential businesses”) and gave a pretty reasonable answer of “May” for when we might start seeing the economy running a little bit closer to properly again.
Not everyone is an opportunistic piece of human garbage a la Cuomo.
I think the mistake Trump made is going back to assuming all of America is equal to the specific hotspots around the country, but he has done muuuuuuuuuuch better than the majority of world leaders.
Well now, it turns out these restrictions are truly incredible!
Ignore the guesswork in the article and take note of the important point:
“As of noon Wednesday, the company’s live map showed fevers holding steady or dropping almost universally across the country, with two prominent exceptions.”
Amazing! And here I was thinking we’d have to wait until June to be well on the downswing.
I wouldn’t bet money on a total death number, because too many of the people in charge have a vested interest in making that number as large as possible in order to justify their dictatorial orders.
Dying *with* the virus will be conflated to dying *of* the virus. Italy has already set the example at 8 to 1, so it’s not going to be difficult to double, triple, or even quadruple the actual Covid-19 deaths.
*looks at New York City and their press release only numbers*
*Considers places that are diagnosing by phone*
Reblogged this on It's Karl.