A few updates:
1)
I imagine that y’all have all seen Trump’s plan to “let” states loosen restrictions. It’s actually a lot better than I expected from him, since it contains metrics that the states should base it on:

States that attempt to restart their economies will be responsible for setting up testing sites; tracing the contacts of individuals who test positive; and conducting “sentinel surveillance” aimed at identifying individuals infected with the coronavirus but not displaying symptoms, and then tracing their recent contacts, Birx said.

I forget where I saw them, but there are specific metrics for each of those conditions that Trump expects states to meet in order to loosen up.

 

2)
Important info: I’ve been mentioning caveats to the IHME (U Washington) model for a while, but it turns out it’s much worse than that; they’re not doing any epidemiological modeling at all, just superimposing the Wuhan curves onto US data. From StatNews (yesterday):

 

HME uses neither a SEIR nor an agent-based approach. It doesn’t even try to model the transmission of disease, or the incubation period, or other features of Covid-19, as SEIR and agent-based models at Imperial College London and others do. It doesn’t try to account for how many infected people interact with how many others, how many additional cases each earlier case causes, or other facts of disease transmission that have been the foundation of epidemiology models for decades.

Instead, IHME starts with data from cities where Covid-19 struck before it hit the U.S., first Wuhan and now 19 cities in Italy and Spain. It then produces a graph showing the number of deaths rising and falling as the epidemic exploded and then dissipated in those cities, resulting in a bell curve. Then (to oversimplify somewhat) it finds where U.S. data fits on that curve. The death curves in cities outside the U.S. are assumed to describe the U.S., too, with no attempt to judge whether countermeasures —lockdowns and other social-distancing strategies — in the U.S. are and will be as effective as elsewhere, especially Wuhan.

3)

So where can we find better models? There are a bunch out there, but I’m not sure which ones are considered the best. Does anyone else have a good sense of that? The StatNews points to a couple, calling out this machine-learning-based one as especially accurate so far, and also mentioning this one.

InsightMaker is an interesting site where you can look (in a somewhat intuitive/accessible flowchart-ish way) at models created by other people, and tweak them yourself, as deeply as you like (or even build your own from scratch). They probably won’t be as good as the most accurate ones out there, but they can be pretty damn good.

 

4)

At least for as long as we maintain the current level of social distancing it seems like we’re past the first-derivative peak nationally (ie the number of new cases per day is clearly going down), which is an enormous relief. I don’t necessarily think we’re at the peak of cases yet, although I think hopefully that’s coming soon as well, as the number of new daily cases descends toward zero. Does anyone disagree? I’m sure it’ll continue to get worse in some specific places, but it overall we’re headed in the right direction as far as I can tell (see for example the “new cases” and “daily growth rate” sections here).

Of course that may all change the minute we start relaxing our current social isolation, if we do that badly. But it’s very good to know that we’re able to clamp down enough to keep it under control.