Back on Mar. 21 I complained that I hadn’t seen any cost-benefit analyses of the lockdown policies that had just been applied where I live. Some have been posted since, but I’ve finally bothered to make my own. Here are two.
So far as I can tell, you aren't modeling the costs of uncertainty. You're making arguments about an average IFR, but we don't really know what it is. If your prior includes a small probability that the IFR is much higher, that could significantly affect your expected loss for ending the lockdown. Conversely, a significant benefit of the lockdown is that it gives us more time to determine the exact IFR (and its dependence on things like age).
I could see hope if we were to copy an entire package of policies that had worked elsewhere. But the public seems strongly against that. What we have actually been doing for the last month, it doesn't look like it will work well enough to be worth its cost. http://www.overcomingbias.c...
In countries like S. Korea, Taiwan, Australia, Germany, China, New Zealand, etc. they have either eradicated the disease or are on a path to where they can re-open the economy with test and trace for the stragglers. A short sharp lockdown followed by a full reopening (rather than the only partial re-opening that is the best you can do when the disease is still prevalent and people are free to protect themselves) has very different dynamics.
Wouldn't you want to include largely eliminating the disease as an option if that's what some other nations have done?
human cost to the dead person"dead person's estate and family" - FTFY, dead men don't care.I think the science crowd could put a number on the secondary impacts of a person dying, e.g. productivity loss from grieving, etc.
In some areas, it's already there. It's a local, not national phenomenon, for the most part. Also, it depends on testing capability and public health resources for contact tracing, etc.
A really solid, enforced lockdown would bring it to that level in a month to three, depending on the start point. It has been demonstrated in a number of other countries. The math is really not hard - the number of new infections per interval is Rt times the number of current infections. Rt < 1.0 means the epidemic will fade.
Here's a link to a basic SEIR epidemic model where you can set parameters by dragging icons. It starts with the parameters for this virus (SARS-CoV-2).
So... set the population and initial infected, then play with R0 (which really should be labeled Rt). I'd uncheck the boxes for "Recovered," "Hospitalized," and "Fatalities" to get the best view of it.
How many additional months of lockdown in the US do you expect before the number is brought down to that required for testing and tracing to become viable?
There are two things to keep the infections low. Yes, I do mean that the infections can be brought down to a threshold where the contact tracing can work. But, it requires a whole lot of testing, and good social distancing.
When the effective reproduction number is below 1.0, the number of infections falls and keeps falling as long as that is true. You achieve that by social distancing and sanitation, until you have enough testing capacity, and the infection rate is low enough, that you can switch tactics. That is the federal plan.
Even with switched tactics, you don't just go back to business as usual - you still need social distancing, but nothing as severe as now. Most sectors of the economy can operate as normal. Some will still suffer.
When you do that, as the countries I mentioned are doing (see South Korea as a good example), the total number of infections before a vaccine arrives is a lot lower than any herd immunity approach can do, for basic math reasons.
As for protecting the vulnerable, that is a lot easier if there are fewer infections going around. It is simply not possible to prevent all exposures. As one who is vulnerable, and is trying to do so, I know that. I cannot live in a bubble - I have to eat, I might need medical care, my appliances might fail, etc. We have been trying to protect nursing homes, and look at the catastrophes there.
My claim is that the number of infections will be roughly, not exactly, the same—provided that the lockdown is not maintained until a vaccine is developed, as I noted. If you think this is incorrect, could you please explain your reasoning or provide a link to an explanation? In particular, I'd like to know what mechanism will prevent the number of infections from reaching the herd immunity threshold once the lockdown is lifted. (I assume you don't mean that a country with an estimated +10M infections can rely on lockdowns to bring down infections below the threshold of 10k or so cases for testing & tracing to become viable, not to mention that this approach didn't even work in much smaller and better governed nations like Singapore, or that it is a non-starter for most nations on Earth which lack the testing & tracing infrastructure needed to implement it and the resources to develop it.)
I agree with you about the effects on the medical system; my claim concerned the number of infections, not the number of deaths, which can vary for that reason.
As a side note, deaths can also vary, probably even more so, due to the age profile of the population affected. I'm surprised to see so much emphasis put on hospital effects, relative to age effects, given that the latter can be so significant—mortality increases exponentially with age, doubling every seven years or so. Moreover, since lockdowns protect everyone, they protect the least vulnerable just as much as the most vulnerable. Yet if the number of infections is roughly fixed, this indiscriminate protection can increase deaths substantially: for every additional young person protected, there's roughly one more old person infected.
Even if SSC is down, it lives on in his army of minions
Hanson's response is the sort that only a massive asshole gives.
"assume the recovered are immune"
One of many of your incorrect assumptions.
We must make decisions now, or very soon. We aren't likely to learn that much more on IFR than we have soon.
So far as I can tell, you aren't modeling the costs of uncertainty. You're making arguments about an average IFR, but we don't really know what it is. If your prior includes a small probability that the IFR is much higher, that could significantly affect your expected loss for ending the lockdown. Conversely, a significant benefit of the lockdown is that it gives us more time to determine the exact IFR (and its dependence on things like age).
I could see hope if we were to copy an entire package of policies that had worked elsewhere. But the public seems strongly against that. What we have actually been doing for the last month, it doesn't look like it will work well enough to be worth its cost. http://www.overcomingbias.c...
Did you read the Hammer and Dance article I referred you to earlier? https://medium.com/@tomaspu...
In countries like S. Korea, Taiwan, Australia, Germany, China, New Zealand, etc. they have either eradicated the disease or are on a path to where they can re-open the economy with test and trace for the stragglers. A short sharp lockdown followed by a full reopening (rather than the only partial re-opening that is the best you can do when the disease is still prevalent and people are free to protect themselves) has very different dynamics.
Wouldn't you want to include largely eliminating the disease as an option if that's what some other nations have done?
human cost to the dead person"dead person's estate and family" - FTFY, dead men don't care.I think the science crowd could put a number on the secondary impacts of a person dying, e.g. productivity loss from grieving, etc.
In some areas, it's already there. It's a local, not national phenomenon, for the most part. Also, it depends on testing capability and public health resources for contact tracing, etc.
A really solid, enforced lockdown would bring it to that level in a month to three, depending on the start point. It has been demonstrated in a number of other countries. The math is really not hard - the number of new infections per interval is Rt times the number of current infections. Rt < 1.0 means the epidemic will fade.
Here's a link to a basic SEIR epidemic model where you can set parameters by dragging icons. It starts with the parameters for this virus (SARS-CoV-2).
So... set the population and initial infected, then play with R0 (which really should be labeled Rt). I'd uncheck the boxes for "Recovered," "Hospitalized," and "Fatalities" to get the best view of it.
How many additional months of lockdown in the US do you expect before the number is brought down to that required for testing and tracing to become viable?
How could I quantify it at this point? I'd say better than 50% but that's a SWAG.
There are two things to keep the infections low. Yes, I do mean that the infections can be brought down to a threshold where the contact tracing can work. But, it requires a whole lot of testing, and good social distancing.
When the effective reproduction number is below 1.0, the number of infections falls and keeps falling as long as that is true. You achieve that by social distancing and sanitation, until you have enough testing capacity, and the infection rate is low enough, that you can switch tactics. That is the federal plan.
Even with switched tactics, you don't just go back to business as usual - you still need social distancing, but nothing as severe as now. Most sectors of the economy can operate as normal. Some will still suffer.
When you do that, as the countries I mentioned are doing (see South Korea as a good example), the total number of infections before a vaccine arrives is a lot lower than any herd immunity approach can do, for basic math reasons.
As for protecting the vulnerable, that is a lot easier if there are fewer infections going around. It is simply not possible to prevent all exposures. As one who is vulnerable, and is trying to do so, I know that. I cannot live in a bubble - I have to eat, I might need medical care, my appliances might fail, etc. We have been trying to protect nursing homes, and look at the catastrophes there.
What do you mean by "quite good"? 50%?
900 million life-years is given for that (with a simple calculation) here: https://slatestarcodex.com/...
My claim is that the number of infections will be roughly, not exactly, the same—provided that the lockdown is not maintained until a vaccine is developed, as I noted. If you think this is incorrect, could you please explain your reasoning or provide a link to an explanation? In particular, I'd like to know what mechanism will prevent the number of infections from reaching the herd immunity threshold once the lockdown is lifted. (I assume you don't mean that a country with an estimated +10M infections can rely on lockdowns to bring down infections below the threshold of 10k or so cases for testing & tracing to become viable, not to mention that this approach didn't even work in much smaller and better governed nations like Singapore, or that it is a non-starter for most nations on Earth which lack the testing & tracing infrastructure needed to implement it and the resources to develop it.)
I agree with you about the effects on the medical system; my claim concerned the number of infections, not the number of deaths, which can vary for that reason.
As a side note, deaths can also vary, probably even more so, due to the age profile of the population affected. I'm surprised to see so much emphasis put on hospital effects, relative to age effects, given that the latter can be so significant—mortality increases exponentially with age, doubling every seven years or so. Moreover, since lockdowns protect everyone, they protect the least vulnerable just as much as the most vulnerable. Yet if the number of infections is roughly fixed, this indiscriminate protection can increase deaths substantially: for every additional young person protected, there's roughly one more old person infected.
I guess there are a lot of people you don't know.