Pages

Thursday, April 23, 2020

The Virus Models Are Irrelevant

... because the lock down is irrelevant. In the end, we'll have to reopen the economy with the vast majority of the population still without antibodies.

It's like holding your breath underwater. Pretend there's a sniper above the surface. He's going to shoot you when you come up. He's willing to wait an hour for you. You're doomed.

I read a headline, but couldn't find the story and don't feel like looking for it now, where Governor Cuomo was bloviating about how New York needed more funding from the Feds in order to open back up for business. If I were Trump, I'd just shrug and say, "Then don't open up. There isn't any money here for you or anyone else."

The states get their money from taxes. Taxes are paid by people and businesses who work and produce things. If you shut down your economy, you don't get any tax revenue.

How long can you hold your breath?

Even with the latest data showing the infection rate is about 50x higher than we thought, it still means under 10% of the population has been exposed. All of those models and worries that the hospitals would be overrun? Oh well. We're out of time to keep things closed. Not only that, keeping things closed is simply delaying the inevitable while at the same time it's ruining our budgets and destroying our economy.

We bought all the time we could buy. We now have enough information to overcome the misunderstandings we had because the Chinese lied and lied and lied. There are a couple of decent treatments available. That's as good as it's going to get.

Bonus Rantlet: I saw an article in the WSJ describing the plight of African nations. They're going to get whacked with this thing. They can't care for all of their sick under the best of circumstances. This is overwhelming. The article talked about their debt loads and mentioned that some in the West were considering forgiving those debts. With what?

15 comments:

  1. The lockdown wasn't supposed to stop people from getting sick.

    It was supposed to make it so people got sick slowly, withotu overwhelming the medical system.

    We are not even close to overwhelming the medical system, although we're going to get there just from the people who are told their, oh, heart surgery or such is an "elective" surgery.

    ReplyDelete
  2. Agree with basically all of points in the post and Foxfier's comments.

    My minor disagreements are that I don't see the models as irrelevant, per se , merely as (worst case) predictions of what might happen, but did not because we changed our behavior, thus intentionally we made them unrealistic. However, they remain relevant as we start reopening the country, even if only as warnings of what might happen if we reopen foolishly.

    You said:
    "We bought all the time we could buy. [Indeed! We also bought all time we needed to buy.] We now have enough information to overcome the misunderstandings we had because the Chinese lied and lied and lied. There are a couple of decent treatments available. That's as good as it's going to get."
    Here I disagree strongly. We can also use the data we've earned with this lockdown, to reopen smartly. And while I agree that testing and tracing are useful strategies to help keep the spread down, trying to turn it off is as hopeless as... as... well, jumping into a pool and holding your breath to avoid a sniper.

    So what is the smart way to reopen? It is to keep the oldest folks isolated (and I hate to admit it, but in this context "old" is 55+) and to isolate those with the known risk factors (a.k.a. the co-morbidities listed here. If you really want to see why, I covered it in reasonable detail on my blog.

    ReplyDelete
  3. "None of this matters. We're going to have to reopen our economy long before the population is virally prepared. We might as well open it now."

    Yeppers.

    ReplyDelete
  4. My minor disagreements are that I don't see the models as irrelevant, per se , merely as (worst case) predictions of what might happen, but did not because we changed our behavior, thus intentionally we made them unrealistic

    *shakes head* I keep seeing that said, but it's not so. The Imperial model had a worst-case and an "if we take immediate action" scenario, with a bunch of different actions. So did the IHME.

    We were significantly below their lowest numbers before they were revised.

    In the case of the IHME models, they re-did them with new numbers, and by the time it was up on the website the model for "we keep doing exactly what we are" had a lower estimate that was significantly higher than reality.

    If you go to the website, now, you'll notice they got enough brains to model the lowest projected numbers to be at most very, very slightly above the current numbers.
    https://covid19.healthdata.org/united-states-of-america


    So what is the smart way to reopen? It is to keep the oldest folks isolated (and I hate to admit it, but in this context "old" is 55+) and to isolate those with the known risk factors (a.k.a. the co-morbidities listed here. If you really want to see why, I covered it in reasonable detail on my blog.

    I don't know if you heard, but New York is REQUIRING nursing homes to take folks who are known to be infected. To make room for that surge that will be here, any day now.... they did just recently remove the ordered DNR for 911 services, at least, so maybe they'll remove this, too, before they kill off more people from sheer insane stupidity.

    ReplyDelete
  5. Ohioan, you're right.

    We can also use the data we've earned with this lockdown, to reopen smartly. And while I agree that testing and tracing are useful strategies to help keep the spread down, trying to turn it off is as hopeless as... as... well, jumping into a pool and holding your breath to avoid a sniper.

    Thanks to the models and the time we bought, we now know that a majority of the country can go back to work. We also have the ability to warn people in the at-risk category of the risks and let them make their own decisions.

    As an aside, I love the comments that disagree with me because they take my half-baked rants and bake them the rest of the way.

    Foxie, I think I'm agreeing with you. As I understand what you're saying, it's that blind obedience to these models is a bad idea. I would generalize it farther and say that there are worse things than death. National currency collapse, for example ...

    ReplyDelete
  6. Ilion, nice to see you!

    ReplyDelete
  7. The models were *always* shit.

    ReplyDelete
  8. A computer model is exactly the same thing as a back-of-the-envelope calculation, except that is has been mechanized, so it returns the results of massaging the initial assumptions -- whether those results are useful information of garbage -- quicker and more consistently (*).


    (*) The human doing the back-of-the-envelope calculation may get distracted, making a mistake. A computer model will always return the exact same result; changing an assumption, or a data input, yields a *different* model, which yields a different result.

    ReplyDelete
  9. In one of my previous lives, I was a mathematician and did some computer modeling. The environment I was modeling was considerably simpler than what these poor devils are trying to do, so my models were pretty close to reality.

    It's not that the models aren't any good or that they're not valuable, it's just that their fidelity drops as the complexity of what they're trying to model increases. Their fidelity is also proportional to our understanding of the situation. That's why they gradually got better, but still aren't totally reliable.

    That's also why I believe that the climate models are of only modest value.

    ReplyDelete
  10. Foxfire, I won’t speak to the other models, I ignored them from the start (because you really couldn’t know what assumptions they made). I can speak for my modeling. My first run could only be described as terrifying. But it was due to using the ‘data’ from China. Once there was better data (from Italy of all the unreliable places), they were much less terrifying. I never even tried to model Stay-at-home + Social Distancing. There was no way to predict (at least I couldn’t) what they would do to the numbers. But since I was concerned less with prediction than understanding, I was happy to just use a reduction factor on the spread to match the data. It has (as expected) changed only very slowly, and the data and the model still agree to a reasonable degree.

    As to what NYC is doing, it would be a comedy of errors except that it has had a very human toll. Apparently they don’t even look at their own data. The cases have been in near free fall for something like ten days (if you bother to find the data, be advised that any data they report for the last 5 days will still change a lot, so don’t trust the last 5 or 6 points). That DeBlasio and his advisers are still making stupid decisions is, as Forrest Gump said, “Stupid is as Stupid does.”

    ReplyDelete
  11. Definitely dark horror.

    I was extremely worried back in January-- we've got military buddies in Japan and Korea, so it was on our radar-- and about Chinese New Year I was arguing strongly against folks who were laughing about it being "just the flu" and calling folks racist for not wanting to go to places they KNEW had people that visit China regularly.

    I slowly got less worried, adn by the time we got the cruise ship information back, I was calmed down....

    Then states started acting like the up to date information was what we had when China was the only numbers! ARGH!

    ReplyDelete
  12. "It's not that the models aren't any good or that they're not valuable, it's just that their fidelity drops as the complexity of what they're trying to model increases. Their fidelity is also proportional to our understanding of the situation. That's why they gradually got better, but still aren't totally reliable."

    But that's *exactly* to say that these models are not any good, are not not valuable -- they are, in fact, anti-knowledge, even now.

    In the early days, for lack of reliable data to fuel their assumptions, it was *impossible* for the models to yield useful information. And, lacking that reliable data, they decided to treat it as though it were the Andromeda Strain.

    And now, we still don't have reliable data, because the bureaucrats, both in and out of government, are cooking the books. In China, Russia, Iran, N. Korea, and such, they are under-reporting. In Italy, the USA, and such, they are inflating the count of deaths.

    ReplyDelete
  13. I was a computer programmer my entire adult life. I've retired from that. Since last year, I've been working for a non-emergency medical transportation company (*), driving a wheel-chair van. Most of the people I transport(ed) are on dialysis, quite a few (whether or not on dialysis) are in nursing homes.

    I was beginning to worry back in early March, both for my own safely, but even moreso for our clients. Back in those days, it was being reported that one might be infected and spreading spreading the virus for as much as two weeks before showing any symptoms oneself. My great fear wasn't that I'd catch it, but that I'd catch it and spread it without knowing I had it.

    I had scheduled to be sure I wasn't working on Saturday, March 14; planning to go visit my family in Indiana. My sister called the afternoon of the 13th to suggest I delay the trip, in case I couldn't get gas for the return trip -- panic-buying hadn't yet hit my area in Ohio, and the various governments in the US had not yet announced that they were going to destroy the economy, but she suspected that that was on the way.

    Then, that night, I woke up sick. So, I didn't go to work for the next week and a half. I was scheduled to work on Wednesday, March 25 -- one trip only (but the guy cancelled as I was on my wat to get him): in that week and a half that I'd been home mildly sick, everything had crashed, and just about the only trips my company was still making were for dialysis.

    Since there wasn't enough work for all the drivers, and since I can get by without the income from that job, I told them to give preference to drivers who need the income. So, for the next couple of weeks, they called me every day with the message that I wasn't scheduled to work the next day. Then, a couple of weeks ago, they shut down operations entirely ... until July 1 (or even later).

    So, I haven't worked, or earned any money, since March 13. And, while I can get by without that income -- at any rate, until the economy collapses and destroys my IRAs -- I *had* budgeted for it.

    I don't *know* whether I've already had the Fung Flu (if I did, for me it presented as a very mild flu) -- for, despite that they seemingly have a levy request at every (**) election (**'), the local Board of Health (as also the State Board) was useless. The bureaucrats didn't even extend their hours ... which were already shorter than people with real jobs work.



    (*) I could easily be earning 2 or even 3 times what they pay me were I to un-retire.

    (**) At aly rate, seemingly every off-election, preferably the primary. The only time a levy request is on the ballot during a general election is if it has already failed to pass before.

    (**') Either: "This is not a new levy, so renew it!" or, "This new levy will cost the average homeowner only $.35 a day, so pass it!"

    ReplyDelete
  14. Great input, all.

    Ilion, these models are not any good, are not not valuable -- they are, in fact, anti-knowledge, even now is not something with which I can agree.

    Models, at least done properly, exist to help a human answer a specific question. They only need to be good enough to do that. I would bet that the original models were requested so that leaders could know just how bad things could get. Despite their inaccuracies, I think they did that.

    When you think about it, we all model reality in our heads. We estimate the future and make decisions based on those mental models. The computer models are simply a codification of our understanding, up to the limits of digital circuitry.

    That's the story I'm telling the jury, at least. ;-)

    ReplyDelete
  15. I said nothing about models in general. I said that *these* models were shit from the start.

    ReplyDelete