Nate Silver

Nate Silver @NateSilver538

FWIW, I think the "populist revolt is starting against lockdown!" takes are rather overdone based on where public opinion *currently* stands. But, of course, things could change, and probably in a way where opinion becomes more partisan. — PolitiTweet.org

Nate Silver @NateSilver538

This has been fairly noisy data but there has been a bit of a decline over the past several days in Americans' concern over coronavirus, although it's still very high. https://t.co/mnZkHDvJSs https://t.co/XD6dnLjGJN — PolitiTweet.org

Nate Silver @NateSilver538

@Herring_NBA @JADubin5 Lasagna is delicious too, though. — PolitiTweet.org

Nate Silver @NateSilver538

@conorsen I've found my resolve holding up better than I thought, but I think it's going to be harder for people once they start seeing *other* people do stuff. — PolitiTweet.org

Nate Silver @NateSilver538

@yeselson They don't have a ton of ICU cases so far, but the concern is that it's still getting worse, and will continue to get worse with their current strategy. It isn't getting worse at a super fast rate because they are doing *some* social distancing, but it is getting worse. — PolitiTweet.org

Nate Silver @NateSilver538

@kmedved Yeah, I think that's right. And I'm not sure why we should necessarily have expected deaths to peak last week given that new cases had not really begin to decline yet. The case is stronger now that new cases are declining, so fingers crossed we'll see that in deaths in 7-10 days. — PolitiTweet.org

Nate Silver @NateSilver538

Yeah. The one thing to be slightly mindful of is that states may now be going back and counting deaths that they missed before, and it's not exactly clear how this might show up in the data. But Occam's razor is that we hadn't hit the peak in the number of deaths yet. — PolitiTweet.org

Nate Silver @NateSilver538

On the other hand, the new cases data remains more encouraging, despite a fairly bad day from NY. US-wide, it does look possible that new cases have begun to decrease (at least if you adjust for testing volume), though there will probably be bumps along the way. — PolitiTweet.org

Nate Silver @NateSilver538

Not great. It looks like this week is likely to bring a new peak in deaths, rather than Tuesday being a one-day anomaly. Each day has continued to bring a higher death count *than the same day one week earlier*, which is a good day to avoid day-of-week effects. — PolitiTweet.org

Nate Silver @NateSilver538

US daily numbers via @COVID19Tracking: Newly-reported deaths Today: 2,492 Yesterday: 2,299 One week ago (4/8): 1,901 Newly-reported cases: T: 30K Y: 26K 4/8: 30K Newly-reported tests: T: 161K Y: 147K 4/8: 140K Share of tests positive: T: 19% Y: 18% 4/8: 22% — PolitiTweet.org

Nate Silver @NateSilver538

Related to this: so many parameters about COVID-19 are so uncertain that societies don't really know what trade-offs they're making. An underrated reason to have done/keep doing social distancing is simply to have more time to collect more data and evaluate evidence. — PolitiTweet.org

Nate Silver @NateSilver538

One thing about Sweden is they're not doing that much testing. It's not **awful**, but it's pretty bad; slightly lower testing rates per capita than the US. It seems hard to know how well your unusually lax strategy is working if you aren't really testing people. — PolitiTweet.org

Nate Silver @NateSilver538

Things that have made me more pessimistic, conversely. https://t.co/DxETa4XIxd — PolitiTweet.org

Nate Silver @NateSilver538

Thinking back to how I might have expected things to unfold a few weeks ago, it seems like we're in neither the best-case scenario or worst-case scenario but somewhere in between. FWIW, here are the things I've gotten more optimistic about. Pessimistic stuff in the next tweet. https://t.co/f4lSYc4Oy3 — PolitiTweet.org

Nate Silver @NateSilver538

2) But we also want to know *when* deaths occur to see how the curve is bending. Going back and reporting/inferring a bunch of new deaths that actually occurred weeks ago could confuse this. So it's helpful to have a separate category. Also helpful to track when deaths occurred. — PolitiTweet.org

Nate Silver @NateSilver538

This would be good if true, in two senses: 1) We know there are many unreported deaths. Reasonable priors from all-cause mortality numbers seem to be that we're missing on the very rough order of 1/4 to 1/2 of the deaths (likely varies by location). https://t.co/NOF9LPNxE1 — PolitiTweet.org

Nate Silver @NateSilver538

One thing to watch for from Cuomo's presser today: New York, perhaps along with other states, will begin reporting *probable* COVID-19 deaths in addition to *confirmed* deaths. https://t.co/uL9lrtBO9E — PolitiTweet.org

Nate Silver @NateSilver538

RT @Avik: @NateSilver538 Plans? We have one of those @FREOPP via myself, @lanheechen, and @bobkocher: https://t.co/7fF4p2w5RY — PolitiTweet.org

Nate Silver @NateSilver538

Because I suppose this point can't be emphasized enough: "It's a really bad idea to open up WITHOUT A PLAN (so we need a plan)" is a very different argument than "we need to stay in lockdown for 18 months". — PolitiTweet.org

Nate Silver @NateSilver538

But I think there is often a default toward "must disprove the null hypothesis to X degree of significance" among people who are generally very smart (e.g. scientists) when that sometimes is a good paradigm but also sometimes isn't in the midst of a real-time, real-world event. — PolitiTweet.org

Nate Silver @NateSilver538

These are complicated questions. It doesn't mean "you can always work off probabilities". Sometimes, you probably do want something closer to "proof", both in terms of how you communicate about something and in terms of implementing policy recommendations. It's case-specific. — PolitiTweet.org

Nate Silver @NateSilver538

The discussion about weather is sort of like the discussion about masks. People are looking for incontrovertible proof when a reasonably high probability (that masks help; that warmer weather reduces spread) is probably enough to help guide our actions. — PolitiTweet.org

Nate Silver @NateSilver538

The consensus has seemed to be that around 1 in 10 infections are detected in the US, in which case the true hospitalization rate would be around 1.7%. But there is much disagreement about how much undercounting there is. — PolitiTweet.org

Nate Silver @NateSilver538

How many COVID-19 cases result in hospitalizations? In states that have tracked cumulative hospitalizations, there's a wide range, but an average of 17% of *known cases* have resulted in hospitalizations so far. However, many cases are unknown. https://t.co/WvxFibwkvH — PolitiTweet.org

Nate Silver @NateSilver538

In case you didn't read the earlier thread, this *doesn't* apply to deaths, and we shall see, but there is some reason for concern that the model might *under-predict* deaths instead. https://t.co/ji4cAIYNVl — PolitiTweet.org

Nate Silver @NateSilver538

And about 1/2 to 1/3 as many ICU beds and ventilators as they originally projected. https://t.co/Y93lP4vhNa — PolitiTweet.org

Nate Silver @NateSilver538

Something else I discovered about the IHME projections at the site listed below. They very sharply reduced their estimates of the number of hospitalizations. The current projections have only about 1/5 as many hospital beds required as their original ones. https://t.co/fT04dhj6xD https://t.co/fJnB9R68f6 — PolitiTweet.org

Nate Silver @NateSilver538

There are other factors that could cut down on transmission, too. (Weather, some degree of herd immunity in areas where a lot of people have it; better technology, etc.) But if you wind up with symmetry, it will be mostly by coincidence, not some sort of inevitable process. — PolitiTweet.org

Nate Silver @NateSilver538

In fact, you might expect regulations to gradually be let up (say, NY permits small gatherings in one's home and a few more services like barbershops) which could make things more asymmetric, not less. — PolitiTweet.org

Nate Silver @NateSilver538

Yeah, how steep the downslope is reflects choices societies are making. And the equilibrium, at least in Western countries, seems to be that they're closing enough stuff down to produce a decline (good!) but not a particularly fast one. — PolitiTweet.org