Tuesday, April 21, 2020

Nobody will argue your virus math

They will argue your virus assumptions.
To err is human; to really foul things up you need a computer.
- Unknown, from the 1960s
This is the problem with the virus models (as with all models).  The math is the easiest part, and modeling software libraries are pretty thoroughly debugged.  The crank and the gears that turn are doing what they're supposed to do.

But if the input data is wonky, then the output will never be accurate.  Some of the inaccuracies are unescapable, for example delays in reporting deaths.  Different countries have different reporting policies and cadences, and there's really not much that can be done about that.  After all, the data are the data.  But some really important data are poorly known, or not known at all:
  • True infection rate can only be known with mass testing of the population.  You don't need to test everyone, but you do need to test a statistically significant sample.  For the US, we're talking about hundreds of thousands or even millions of people.  Nobody is working on this, so we simply can't know what the infection rate really is.  several small scale studies strongly suggest that it is much higher - perhaps ten times or more higher - than is being reported.
  • The recovery rate can only be known with mass testing of the population.  We are seeing that many people who catch the virus are asymptomatic; since they don't feel (or act) sick, they don't get care and so are not included in the total number of infected.  Again. small scale studies suggest pretty strongly that the recovery rate is much higher than is being reported.
  • The number of deaths includes a wide range of victims, some of which almost certainly were not killed by the virus - rather, they were killed by other ailments ("co-morbidities") but also had the virus along for the ride.  Right now, all of these are included in the virus death rate, overstating its lethality by an unknown amount.
  • Since lockdowns vary by geography and by what is allowed or prohibited, tracking the result of the lockdowns is fiendishly hard, and I haven't seen anything that is even trying.  We're told that "we're all in this together" and "we're making a difference" but there's no data about how much - if at all.
Put all of these together and it's simply impossible to provide good inputs into any model.  This is basically a game of guessing, and the result has been a series of model re-works that dropped the projected death toll from 6+M to 500,000 to 120,000 to now (maybe) 60,000.

Which is about what the death toll from seasonal flu is.  I'm not saying that this new virus is no more lethal than the flu; I'm saying that the people running the models are saying that maybe this is the case.  Those models have been wildly inaccurate repeatedly in the past, so take the latest model results with a huge grain of salt.

But governments are sitting on top of increasingly restive populations.  The populations have very good reason to be restive - the world economy has taken probably a $10T (that's Trillion) hit.  That hit has been very unevenly distributed.  In economic terms this is a very regressive tax targeting the poor and working classes.  Ask Louis XVI or Czar Nicholas II how that turned out.


And so very interesting things are happening to the data.  Very interesting things indeed:
Last week we saw that we went from dying from, to dying with, to just plain dying. Even people who haven’t been tested are now classed as dying from coronavirus. This is juicing the numbers in the direction of the models. Whether this was intentional, to avoid confessing to the most colossal and costly blown forecast of all time, or this is more panicked over-reaction, I’ll let you decide. Either way, they did get a boost in the numbers from the re-definitions, which we’ll see below.  
We saw yesterday that counting who dies of flu or any virus is not so straightforward, that it’s always the result of a statistical model. Every single flu death is not trumpeted from every media organ for months on end, but if they were, then we’d have counts similar to the way we have counts for coronavirus.  
Something else strange in the numbers. Remember how every week I’d cut and paste the CDC’s update flu hospitalization and death estimates? Can’t do it anymore, because why? Because the CDC stopped reporting on them. This could be because of over-burdened government workers, or because flu deaths aren’t as sexy as coronavirus deaths (even though the totals are similar), or because something else is going on.
And it's not just here in the New World:
It isn’t only in the US where the numbers are looking funny. France, too: 
It’s already happening – France has stopped publishing the weekly mortality report in mid March – where (the lack of) excess mortality could readily be seen. Instead they created a new publication focused only on covid deaths.
Is this all political ass covering by the health services?  Beats me.  Certainly it looks like the motivation exists.  Remember, the reason for all this government-imposed misery was to protect the health care services from becoming overwhelmed.  Well, we're seeing hospital ships with no patients and Army field hospitals being demobilized because there are no patients and the Javits Center (which had been turned into a makeshift hospital) empty and there are doctors and nurses being laid off all over the country.  And there are very odd things happening to the data.

But juicing the numbers can only go so far:
The temptation to juice coronavirus deaths must be overwhelming! The models promised unimaginably huge numbers. We haven’t come anywhere close to them. Millions and millions and millions of lives the world over have been ruined, with more ruin on the way, as the result of trusting expert models. They have to find a way to bring actual numbers in line with models. 
They’re running out of options, though. Dying with from dying from was a good move, and we saw it immediately pop up in the death counts. Dying with suspicion from dying with was also clever enough, and we saw that, too. 
What else is left, though? Only one thing. 
If “dying from” is defined as dying with presence of COVID-19 antibodies, then once we reach herd immunity, which it seems is close in many places, then about 80% of all deaths can be classified as coronavirus deaths.
So where is the crisis?  It's entirely fair to ask this question when literally everything we were told about this "pandemic" has turned out to be somewhere between overblown and flat out wrong.  And quite frankly, not unusual in recent historical terms:

As with the Global Warming hysteria, if the science were as settled as we're being told there would be data falling off of trees confirming everything.  Instead, everywhere we look we see data that calls the projections into question.  If we want science-based public policy then its reasonable to ask for, well, science-based public policy.

Two days ago I posted that we need to re-open the economy.  We're seeing a million people a day lose their jobs, so that's 2 Million more people without a paycheck, just since Sunday.  It may be for nothing: there's actual data that suggests that the lockdown doesn't do much, if anything: Sweden hasn't implemented a lockdown at all and their death curve looks basically identical to the USA's:

The caveat about the problems with the data is a good one, and echoes what I wrote here.  But the data are what we have, and if the lockdown - and the 25M newly unemployed - were actually effective you'd think you'd see something.  You don't.

Enough, all ready.  There is simply no rational, science-based justification to keep the lockdowns in place anymore.  We see this recognized by Governors (who are starting to end the lockdown) and by the population in general (who are starting to willfully violate the lockdown).  Everybody but the "experts" is starting to recognize this, and the "experts" may be refusing to recognize it so that they don't get blamed

15 comments:

  1. There's so much bad science and general sloppy thinking in this mess that it boggles the mind. If you're in a hospital intensive care unit, every patient you see is in bad shape. If your hospital is busy the natural extrapolation is that it's horrible out in the world. But a sample of one tells you very little.

    In the same way if you only measure one thing (deaths from/with/by) of one disease, that's the only thing you know anything about. If you're not counting suicides, depression, domestic violence or any other symptom of misery caused by the shutdown, it didn't happen.

    The argument "how many people do you want to see die" without a counter question is rhetorical self-congratulations.

    How do we know that shutting down the economy had any effect at all? A statistical study in Israel published last week concluded that regardless of whether the country quarantined like Israel, or went about business as usual like Sweden, coronavirus peaked and subsided in the exact, same, way. His graphs show that all countries experienced seemingly identical coronavirus infection patterns, with the number of infected peaking in the sixth week and rapidly subsiding by the eighth week. Being a scientist, when asked why that is, he said, "I don't know." He's a statistician looking at data and that pattern is all he can know.

    https://townhall.com/columnists/marinamedvin/2020/04/15/israeli-professor-shows-virus-follows-fixed-pattern-n2566915

    ReplyDelete
  2. Yours was an excellent treatise on the matter.
    But I must say, I rushed through it and got the drift.
    The numbers are meant to confuse and panic.
    We were supposed to be flattening the curve, not the economy.
    Or were we?
    This could not have played better into the hands of those who would destroy us.
    Not letting a good crisis got to waste.

    ReplyDelete
  3. Agree with all. It's time to open up and get on with life!

    ReplyDelete
  4. I would agree that data is wanting to get a reasonable conclusion or projection.

    After that I also noted a lot of bias in the conclusion. The worst being we are less than 4months (at the extreme) into the epidemic and we have no idea if the bug is wide spread or not and the death rate can only be extrapolated against known infections that we have little data on.

    Right now 800,000 tests are positive out of 4.3 million and roughly 44,000 died with another 75,000 sickend. To extrapolate against
    US population in say 1958 where the US had less than half the people of now is also a statistical gaming. That should ahve been corrected for known cases and deaths per million.

    You left out the more meaningful pandemic the 1918 Spanish flu.
    That had no media.

    Is there panic and hype, hard yes. For the economy its bad but
    then again in the past people just died. However if we go back
    to work and normal interactions and the curve continues to be
    continuing upward will there be enough people alive at work to
    get anything accomplished? If a restaurant has three deaths
    that seem to come from there will that be a death knell to
    continued operation. That's assuming the customers return and
    are not scared off. That may be extreme or on the mark. It
    could also extend to many other parts of the economy and the psychology may not be as predictable.

    We need to go to work but carefully and without adequate testing
    or a know immunity roulette might be about the same or better.

    Eck!

    ReplyDelete
  5. There's a limit to how much they can screw with the numbers. The number of overall deaths from all causes is down substantially because no one is out driving drunk, Traffic accidents are down, Drug OD's are down, etc etc . Here in Mass they are now saying we can't reopen because the number of cases are up. Sure …. as testing increases they will find more cases. the total cases will go up because they are not removing cases from the totals. That's not the right metric to use. But it' the only number the media is focused on ( Today -- Tomorrow it will be some other metric ) If the data doesn't show what you want -- different data.

    ReplyDelete
  6. Well yes and no, BP.

    Given the way this one was handled, and the morons involved... yeah, we need to check the math too! HAR HAR HAR!!!

    I knew this was bogus right off the blocks. The first tell was the run out in results. When they differ by orders of magnitude you KNOW something ain't right. When the clowns started talking about flattening the curve with mickey mouse lockdowns - game over. The math and the assumptions were hopelessly botched. Sorry fags. It's the flu. But they kept digging: all the results are peer reviewed!!! Science!!!!! Welp…. no. Only certain scientists were getting air time and they all happened to be liberal fart catchers more than anything else. Dissenting scientific opinions were being suppressed or censored. Then they started trying to pin all the problems on Trump.

    Far as I am concerned, the quarantines are over, my rights and freedoms apply, and you will disrespect them at your peril. The medicos can either prepare a rational case for my evaluation or FOAD.

    ReplyDelete
  7. I’ve been reading a lot about this in the last week from everywhere, main stream to tin foil.
    My opinion is this virus, while being a serious strain of flu, is being overblown to create a situation where the banks and big corporations can get a massive bail out without pissing off the hoi polloi.
    Follow the money. Our global financial situation is a disaster and about to disintegrate at any time. They need to wipe the slate clean.
    Follow the money. They're pushing Favilavir in China, Remdesivir here, both expensive drugs the drug companies would love to sell shitloads of. Hydroxychloroquine isn't being pushed in the same way. Why?

    You're right about the data. There is none, which makes my brain go WTF. And that which is out there is suspect or outright bogus, so is isn't actually data.

    Thank you for seeking the truth.

    ReplyDelete
  8. One wonders...

    Is it so "not bad" because of the steps taken to prevent it?

    We will never know what would have happened if we hadn't taken the steps we did.

    I fear we may be learning the wrong lesson from this.

    What if the lesson isn't so much "see? We overeacted, we shouldn't do that next time" so much as it is "we did exactly the right thing, and it's a good thing to remember next time it happens?"

    ReplyDelete
  9. First, getting reliable, accurate and high quality data on anything that's complex is not a simple task; second, getting moderately accurate data across multiple sources in a fluid environment is not just excruciatingly difficult, it's usually almost impossible.

    That said, skilled and honest data people know all that, and they offer such data with boatloads of caveats about the data's accuracy and reliability and repeated cautions about basing decisions on questionable data.

    We have none of that here.

    We're being fed data from either incompetent data people or outright fraudsters. If it's incompetence, they have no authority - or right - to be offering such data to be used for decision making.

    If it's fraudulent, that's a whole different problem, one calling for an entirely different solution. And, if one still insists on using severely questionable - or fraudulent - data for decision making, an entirely different course of action is called for.

    This isn't some weekend hobby, and it's well beyond any temporary damage caused to job lots of peoples' lives, it's toying with the destruction of a substantial part of modern human civilization.

    "Ask Louis XVI or Czar Nicholas II how that turned out." I would not rule out Louie and Nicky having a whole lot of company in the near future. I pray we can avoid that.

    ReplyDelete
  10. I would argue that even the models might be suspect. We know almost nothing about the models themselves. For example how is the transmissivity handled? Is it some sort of diffusion model or a hot spot model with growth? What is the nature of the growth terms, is there a term to cut off the growth? I spent decades populating complex models with various sorts of experimental results and even long established well understood models have holes in them. Until I know something about the model itself I'm skeptical.

    ReplyDelete
  11. Anybody who is old enough to have seen different flu epidemics come and go knows that this was all a profoundly stupid overreaction by government nannies. You cannot out wait a respiratory virus. We will all have our chance to catch the Kung Flu over the next three years or so. After that, it will mostly have mutated into just another obnoxious wintertime bug.

    This was a test of the emergency power grab system. This was only a test.
    We failed.

    ReplyDelete
  12. Divemedic's son, at Confessions of a Street Pharmacist, has reported that he and the staff at a Harlem emergency room and ICU that they "just know" when someone is positive from seeing the symptoms.

    Perhaps they are right, but without confirmation from testing it's just their informed opinion.

    I think we're seeing way too much informed opinion from medical professionals lately and not a lot of science.

    ReplyDelete
  13. 1) I absolutely want widespread, nigh universal testing.
    I want it with tests that can differentiate this exact virus, from the common cold, with 99% accuracy. Otherwise it's a talking drug dog with a head cold.

    2) Most of the country went ugly early.
    if you think that had no effect on the nationwide experience of this pandemic, I have a bridge to sell you, cheap.

    3) The one glaring counter-example was a state that derided all suggestion this was going to be bad, and instead told one and all to carry on with life as usual, and still has its subway open for business every day.
    They now have 15,000 dead in their largest city, more dead than all but five other countries.

    4) The "Let's open up" argument is essentially the suggestion that we now adopt the m.o. described in #3, above.

    5) I appreciate the economic chaos this has caused to date. I also abhor the jackholian examples of bureaucratic overreach seen all over the place. The latter does not undo the reality of #2.

    6) Frequently, bomb disposal experts, knowing full well they are exposing themselves to potential death, may attempt to defuse explosive devices. They choose to assume that risk. How anyone proposes to expose the entire population to an unknown risk, with unknown consequences, and to a virus that kills not just the aged or infirm, even if only rarely, is calculus I cannot do.
    If you can't undertake to explain what article of the Constitution gives anyone (and if so, whom) the deciding authority to expose others to a deadly disease, with neither their consent, nor even asking after their informed opinion on the matter, then AFAIK, no one possesses that authority. 45,000 deaths into this, no one can reasonably claim they didn't know this was so serious.

    That being the case, you're going to have to test everyone out of this predicament, however long it takes. Proceeding forward with lifting lockdowns unilaterally, without that much of a good faith effort, falls somewhere on the spectrum between criminal manslaughter and premeditated murder, and pleading sheer magnitude of the crime, or blaming it on "act of God" isn't going to cut it.

    It's one thing when a massive storm comes up, overwhelms your ship, and people drown.
    It's quite another when you walk along the rail throwing people into the sea.

    ReplyDelete
  14. The most definitive statistics we have are related to the age of death. Using the most recent cdc statistics people under age 55 are 3 % of the total deaths. If we use Aesop's WAG of 3 % Corona death rate overall we get a death rate just under the death rate of seasonal influenza.

    So let's make social distancing mandatory for people over 55 and for people with known comorbitdities. Give special store hours. Work from home deals. Stimulis checks, you name it.

    It is understandable if people think this unfair but I think it could be the best approach. We have to face the fact that we won't stay closed until it is over but this approach would give us the best chance not to turn into a third world craphole and protect us from being overcome by Covid-19.

    Now these statistics aren't perfect and this plan needs a lot more work but we need to be working towards a plan that is based in scince and math and not in fear. Too much Dunning-Krueger effect going on in this country. This reply could be part of that but we need to start somewhere.

    ReplyDelete
  15. How about testing that out, as well as doing near-universal testing, starting in the states with the lowest populations?

    That way, if you fail, you fail small.
    And it's much easier to do the testing of everyone when you start with small populations.

    I also don't think it's out of line to declare, for instance, the Five Boroughs of NYFC a COVID "Hot Zone", requiring far stricter precautions, and a much more cautious approach. When one city has provided 1/3 of all deaths in the country, they've clearly been doing a lot of things wrong.

    ReplyDelete

Remember your manners when you post. Anonymous comments are not allowed because of the plague of spam comments.