As I'm writing this, the number of deaths is already higher than that in the second lowest flu season in the last 7 years for which there is complete data. And yeah, that's "fairly standard" in my book. (Would replacing "fairly standard" with "not unusual" make you happier?) As I'm sure you know, an outlier would be defined as something like 1 in 20 or less. But whatever: if 2 in 7 makes it an outlier for you, give it another week or two.
Look Rob, you're just wrong on this. We are talking about *deaths* but the site you're linking to is about *cases*. (The numbers on that page also don't match those you mentioned, but that's probably a typo or something.)
I don't know what to tell you, other than to read the paper. Tables 4 and 5 show a huge range of scenarios and their impact on things like the number of deaths. The actual number in the table are for the UK, but as the Table 4 caption explicitly states "Results are qualitatively similar for the US." The table shows that, depending on which exact strategy is adopted, the number of deaths will be reduced by factors between about 5 and 50 (for their reference case of R0 = 2.4 -- they even show numbers for other values of R0). In the US, where the detailed strategy is decided at state level, there is obviously no single "correction factor". (Maybe this is what you mean by granularity?) But you can still tell pretty easily that the numbers they predict are perfectly reasonable. As a country, the US is going to be somewhere in the middle of the severity of those restrictions. So that would be about a factor of 15 or so. So 2.2M for an unchecked outbreak then becomes 150,000 (with obviously significant uncertainty). And, as they explicitly state, the IFR they adopted at the time was 0.9%, so if we think it's 0.4% now, that number becomes 65,000.
I know that everybody here is now going to jump up and down about how useless models are that are that uncertain, etc etc. But that's just a misunderstanding of how models work or what they are used for. Everything you need to vet what they've done is provided for you. If you don't like the IFR -- or better data becomes available -- you update it. That's the point of models. It's not a flaw. And the fact that different suppression measures produce different degrees of suppression is just that -- a fact. It's not a flaw of the model -- it is what it is.
OF COURSE, all models are wrong in detail -- but some are useful. At the level at which this one is intended, it's pretty good -- it allows you to make better informed decisions.
You may not have noticed this, but it's a little unrealistic to expect me to respond to every poster on here who thinks I'm wrong. But fuck it, here are a couple of questions you apparently wanted answered:
- [W]hen schools need to close because of a snowstorm, should that decision be made by the principal, the superintendent, the school board, the board of freeholders/county legislature, the state governor, Congress, the President, or the United Nations?
- This may shock you, but snowstorms aren't infectious. I am totally comfortable with localism where it's appropriate. It's appropriate for a snowstorm. It isn't for a pandemic. Ask Nassim Taleb (who I would have thought you'd like better than I do, even though we can probably both agree he's not stupid).
- [Do you think the] only answer [to the epidemic] is martial law, ankle bracelets, movement tracking, etc.?
- No, because that would be stupid. What I said was that, if and when a suppression based on social intervention has worked, we will need a strategy to avoid major second (and third, and fourth...) waves. And that strategy will probably have to include mass testing and contact tracing. But there should obviously to strict rules: (a) all of it needs to be temporary, ceasing at the very latest when a vaccine becomes available; (b) testing and contact tracing should be done in an anonymized/encrypted way, with anonymization only relaxed to allow actual positive tests to be contacted and traced; (c) if the numbers are kept small enough, it might even be possible to ONLY contact trace new infections. And no, I'm not actually comfortable with any of this at all. But I'm still waiting to hear better alternatives.
- Are you ok with the government telling Walmart what they can and can not sell?
- Yes, in an actual emergency, why the hell not? If it stops a bunch of idiots going out to buy a nicer pair of shoes during a pandemic, great. It's not about those idiots, it's about everybody else those idiots may put in harm's way. In fact, I think it was *workers* in those stores who complained they were being put at risk by bored idiots browsing the stores.
I'll probably take another break from posting here for at least a few weeks, though I would still like to hear your (and Rip's!) response to three questions:
- (1) What's your best estimate of the number of deaths you'd if the outbreak wasn't suppressed? If it's substantially less than, say, 1 million, what was the assumed IFR and fraction of the population who was infected? And where did those numbers come from?
- Regardless of what your estimate for COVID-19 is, if *some* epidemic could be expected to produce, say, 1 million deaths without strict social interventions:
- (2) Would such interventions be justified?
- (3) Do you think the economy could continue to function more or less normally while such an unchecked infection goes on?