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.

This Vox article on testing is really good. The NYT has an update on testing today as well. And you can see state-by-state testing updates at covidtracking.com.

I wonder what is the percentage of testing is, that constitutes adequate surveillance?  anyone know?

I’ve seen numbers ranging from 150k/day to 22 million/day (in that Vox piece, from Paul Romer — by far the highest I’ve seen).My thought would be that it depends dramatically on the situation — if we’ve managed to get fully past this first outbreak and we’re just stamping out small local outbreaks with testing & tracing, the number could be pretty low. If it’s still everywhere and we’re trying to open things back up, we would need an enormous number.

Other stuff I’ve read in the past few days that may or may not be of interest (none of them are urgent):

• James Heathers describes why we can’t rush into treatments like hydroxychloroquine without thorough testing first, by looking at the history of bone marrow transplants as an example of how that can go badly wrong.
• UnHerd argues that things are actually going fairly well in Sweden (with much less lockdown).
• UnHerd tries to consider how long lockdown can last politically (somewhat UK-centric but applies well to the US too).
• Rolling Stone thinks that the main reason the current situation is so exhausting is that our day-to-day decisions carry way more moral weight than usual.
• SlateStarCodex investigates who did a good or bad job predicting this situation, and why.

Coronavirus patients in areas that had high levels of air pollution before the pandemic are far more likely to die from the infection than patients in cleaner parts of the country, according to a new nationwide study that offers the first clear link between long-term exposure to pollution and Covid-19 death rates.In an analysis of 3,080 counties in the United States, researchers at the Harvard University T.H. Chan School of Public Health found that higher levels of the tiny, dangerous particles in air kno

Source: New Research Links Air Pollution to Higher Coronavirus Death Rates – The New York Times

Study:

https://projects.iq.harvard.edu/covid-pm

(some of this is linked in the previous “summaries of research” post)

https://fivethirtyeight.com/features/why-its-so-freaking-hard-to-make-a-good-covid-19-model/
Testing tracking: https://covidtracking.com/data/
Economic tradeoffs: https://fivethirtyeight.com/features/what-should-the-government-spend-to-save-a-life/

https://www.nytimes.com/interactive/2020/03/26/us/coronavirus-testing-states.html

In the morning of 4/1:

– The FiveThirtyEight article Lib posted on why it’s so hard to model C-19 is really fantastic (although probably not of practical interest). They’ve been doing some good journalism in their science and health section, including weekly surveys of infectious disease experts to try to get a sense of the C-19 outlook.

– NYT data journalism is continuing to kick ass, with a good piece from 3/26 on how each state is doing on per-capita testing (covidtracking is also trying to track tests per state), along with their various updating charts like their daily tracker of C-19 deaths per state and country. I wish they had a centralized table of contents of their C-19 datavis work.

– A few updates and better organization for my bookmarks, and I’m starting to stick stuff I read that I think is good on my blog; that’ll probably be a lot more stuff than my bookmarks.

In the afternoon:

– I’ve been trying to get some sense of what the tradeoff in human wellbeing is for different C-19 strategies. Here’s a starting place, at least: if we treat a human life as worth about 10 million dollars (which is the standard figure researchers arrive at in terms of eg what we’re willing to pay for a safer car, or the premium that dangerous jobs command) we in the US should be willing to spend about 20 trillion dollars on this (if that gets us from a bad outcome of 2 million US deaths to a “good” outcome of about 100k deaths). It’s not clear what the cost of our planned response is on top of the $2 trillion we’ve officially spent (you have to take into account the uncertain effects of economic slowdown, failed businesses, lost jobs etc), but at least one analysis (which I haven’t yet read) suggests that we still come out comfortably ahead using the strategy we’re using. Hoping to try to read that paper at some point, but fivethirtyeight does a good job taking at least a basic look at it. Obviously it’s uncomfortable to think about tradeoffs between human life and money, but in practice we’re all doing it all the time even if we don’t talk about it…

– I found a source for better info on how we’re doing on tests (StatNews). tl;dr: we’re theoretically now in a position to do 100k tests per day (we’ve only done a total of 160k to date, so I’m a bit skeptical that we’re at that capacity in practice). That may be enough? “A recent report from the American Enterprise Institute co-authored by former FDA Commissioner Scott Gottlieb puts the number of tests needed at 750,000 tests per week.” But there must be a ton of assumptions built into that number, and StatNews doesn’t seem to feel too clear on whether we’ll have enough. I’d still like to find comparable info for ventilators and masks/PPE.

https://www.sciencemag.org/news/2020/03/mathematics-life-and-death-how-disease-models-shape-national-shutdowns-and-other#

Over the past two decades, as conservatives and liberals have drifted ever farther from each other, an influential body of literature has attempted to psychologize the partisan divide — to identify conservative and liberal personality types, right-wing or left-wing minds or brains, and to vindicate the claim of the noted political scientists Gilbert and Sullivan, That every boy and every gal / That’s born into the world alive. / Is either a little Liberal / Or else a little Conservative.

In its crudest form this literature just amounts to liberal self-congratulation, with survey questions and regression analyses deployed to “prove” with “science” that liberals are broad-minded freethinkers and conservatives are cramped authoritarians. But there have been more sophisticated and sympathetic efforts, too, like the influential work of New York University’s Jonathan Haidt on the “moral foundations” of politics: Haidt argues that conservatives actually have more diverse moral intuitions than liberals, encompassing categories like purity and loyalty as well as care and fairness, and that the right-wing mind therefore sometimes understands the left-wing mind better than vice versa.

Both the crude and sophisticated efforts tended to agree, though, that the supposed conservative mind is more attuned to external threat and internal contamination, more inclined to support authority and hierarchy, and fear subversion and dissent. And so the political responses to the pandemic have put these psychological theories to a very interesting test.

Source: Opinion | The Coronavirus and the Conservative Mind – The New York Times

I’m planning to start using this as a repository of interesting stuff I read re: the COVID-19 crisis, basically just so I can keep track of it (I’ve also got some curated bookmarks, but that’s only really important stuff; this is more general. I may go back and delete it all once this is all over; we’ll see.

A floating parade dedicated to painter Hieronymus Bosch (previously) honors the artist’s fascination with the fantastical and absurd in an annual event that embodies his philosophy and aesthetic. The 2019 occurrence of the Bosch Parade included a musical performance played on a partially submerged piano…

A Parade of Earthly Delights: Floating Bosch Parade Celebrates Painter Hieronymus Bosch in Spectacular Aquatic Event

Sadly, though, [a particular article by another author] still falls short of truly bridging the conceptual gap between nerds and “weird nerds.” Speaking as a lifelong member of the weird-nerd contingent, it’s truly surreal that this distinction exists at all. I’m slightly older than Nate Silver and about a decade younger than Paul Graham, so it wouldn’t surprise me if either or both find it just as puzzling. There was no cultural concept of cool nerds, or even not-cool-but-not-that-weird nerds, when we were growing up, or even when we were entering the workforce.

That’s no longer true. My younger colleague @puellavulnerata observes that for a long time, there were only weird nerds, but when our traditional pursuits (programming, electrical engineering, computer games, &c) became a route to career stability, nerdiness and its surface-level signifiers got culturally co-opted by trend-chasers who jumped on the style but never picked up on the underlying substance that differentiates weird nerds from the culture that still shuns them. That doesn’t make them “fake geeks,” boy, girl, or otherwise — you can adopt geek interests without taking on the entire weird-nerd package — but it’s still an important distinction. Indeed, the notion of “cool nerds” serves to erase the very existence of weird nerds, to the extent that many people who aren’t weird nerds themselves only seem to remember we exist when we commit some faux pas by their standards.

[…]

Don’t get me wrong, I’m thrilled to bits that every day the power to translate pure thought into actions that ripple across the world merely by the virtue of being phrased correctly draws nearer and nearer to the hands of every person alive. I’m even more delighted that every day more and more people, some very similar to me and others very different, join the chorus of Those Who Speak With Machines. But I fear for my people, the “weird nerds,” and I think I have good reason to. Brain-computer interfaces are coming, and what will happen to the weird nerds when we can no longer disguise our weirdness with silence?

When Nerds Collide