I conclude that we all prefer not to know.
Just discovered the downloadable data (from GitHub) behind the Johns Hopkins map.
I am really dissatisfied by the statistical discussions I'm reading. I feel as though there is more information and intelligent synthesis out there than what I am accessing. For instance, NW posted this this lovely graphical thing, but as its authors acknowledge, it's a guesstimate, and I'm not clear that I understand the assumptions behind it.
I think Drum makes a decent stab at it here. Deaths are the right metric -- rather than cases, although hospitalizations are also a good, solid metric -- and the grey dotted line provides an instructive Italian comparison. I'm a little unclear on why the vertical axis counts deaths as a multiple of day 1 deaths, but that shouldn't change the curve from a graph of deaths-per-million population, which makes more sense to me.
I wonder why Americans aren't dying so much.
6 before seeing 4. I'll have to have a look at that.
I've been seeing quite a few articles on the hypothetical seasonality of this thing, and how the biggest outbreaks are all occurring in a thing band. At first I thought it was irresponsible journalism on thin data, but it keeps getting repeated, and I'm wondering if the evidence has been stacking up or if it's just too desirable a story. Opinions?
"occuring in a THIN band" (worldwide, a thin band of temperatures)
The same thin band that happens to be most densely connected by air travel.
6.last: The effective and proactive steps put in place by the President have saved thousands of lives.
It's a little similar to the Guns, Germs, and Steel hypothesis for how so many things spread more easily laterally rather than longitudinally. So is it just an artifact of 10?
9: Buzzfeed has some reporting indicating that US deaths are being undercounted.
It's because of Sesame Street. All American epidemiologists count "One, ahh, ahh, ahh. Two, ahh, ahh, ahh."
I've been obsessively looking at these:
https://www.worldometers.info/coronavirus/
https://covidactnow.org/
https://www.latimes.com/projects/california-coronavirus-cases-tracking-outbreak/
If you wanted to you could spin some geodeterminist story about civilization developing earliest and most richly on the fringes of the Indo-African-Asian collision belt (Central Yangtze, Val Padana, Iranian plateau, Spanish meseta). Except those are totally different sets of environments, sedimentary fans versus high altitude plateux. And I think Korea and Japan are totally different orogenies. And then there's North America. And the fact the pandemic is obly just getting started, see 2 and India comments elsewhere. When all's done there'll b proportionally more corpses South than North. In conclusion fuck determinism and also everything else.
And the "thin" band is actually 10-20° of latitude depending how you count. Out of ~120° in the world where people live.
8: in favour of that: we know quite a lot about coronaviruses, and they tend to be sensitive to temperature and humidity in terms of spread. Hence why you get more colds in winter. It's not unlikely that R might vary quite a lot for SARS-CoV-2 depending on climate.
Against it: 10, and also that thin band happens to have a lot of people living in it. So, really, no surprise that there have not been clusters of thousands of cases in the Sahel or the outback; there aren't clusters of thousands of people there.
What strikes me about Drum's graphics is that they're good but they're starting from the one death per million point - that implies that the starting point is sensitive to a lot of fairly random factors. Why are people in the US not dying as much? Maybe it's because the curve for the US happened to start a few days earlier: that starting-point death in the US came, for whatever reason, a few days earlier in the general spread of the epidemic than it did in other countries.
10: That doesn't account for the relatively low number of CV cases in the Philippines, Malaysia and Indonesia.
Is it possible Italy's death rate is dropping partially due to scaling up hospital resources over time?
Governor">https://www.google.com/amp/s/www.businessinsider.com/coronavirus-mississippi-governor-executive-order-most-businesses-essential-2020-3%3famp">Governor of Mississippi sabotages all the local measures by declaring almost everything "an essential business". What a deadly fuckwit.
Josh Marshall has had a lot of good coverage of this, with a footer on all his posts giving links to reliable sources of data/expertise. Here's his latest, as an example.
24: The linked article itself is really good, and pretty depressing. As a general rule, the grim stuff seems to be better thought through than the happy talk.
Kevin drum's charts also assumes that on each day, each nation is at a single point on the curve. That isn't what's happening. Wuhan got hit before the rest of China, and also started recovering before the rest. In the U.S., Washington State may already be past the worst, New York and Louisiana are clearly in the worst phase, and about 20 states haven't had enough deaths yet to be at day 1.
26: Yeah, I've been wishing he'd break it out by state. Although NYT has been doing that with doubling time for deaths by state.
Also at the moment, their chart for countries is mirrored right-left which is weird and makes it look like the pandemic is tidily coming to an end.
This seems like a nice data visualization. Mentioned in my Pandemics class today.
18: Among people who, uh, have exited the disease-having state, the difference between the US and Italy is much smaller than the death/cases ratio: 62% recovered for the US, 56% for Italy. And most of those US recoveries have been in the last day or two. I think it reflects more of the US victims being earlier in their course of disease. It's complicated by unimaginative's point in 26, though.
28: Brings to mind Rimmer's speech about the experience of WWII on the mirror world with a backwards timeline. The Wehrmacht (or the Soviet Army, whichever) marching backwards across Europe, raising the dead as they go, taking the bullets that killed them and tidily storing them in guns.
I've heard that joke about Rodney King: If you run the tape backward, it looks like the cops are helping him up and sending him on his way. It just depends on your perspective.
It's Slaughterhouse-Five that has the first instance I know of:
American planes, full of holes and wounded men and corpses took off backwards from an airfield in England. Over France a few German fighter plans flew at them backwards, sucked bullets and shell fragments from some of the planes and crewmen. They did the same for wrecked American bombers on the ground, and those planes flew up backwards to join the formation.
The formation flew backwards over a German city that was in flames. The bombers opened their bomb bay doors, exerted a miraculous magnetism which shrunk the fires, gathered them into cylindrical steel containers, and lifted the containers into the bellies of the planes. The containers were stored neatly in racks. The Germans below had miraculous devices of their own, which were long steel tubes. They used them to suck more fragments from the crewmen and planes. But there were still a few wounded Americans, though, and some of the bombers were in bad repair. Over France, though, German fighters came up again, made everything and everybody good as new.
When the bombers got back to their base, the steel cylinders were taken from the racks and shipped back to the United States of America, where factories were operating night and day, dismantling the cylinders, separating the dangerous contents into minerals. Touchingly, it was mainly women who did this work. The minerals were then shipped to specialists in remote areas. It was their business to put them into the ground, to hide them cleverly so they would never hurt anybody ever again.
The American fliers turned in their uniforms, became high school kids. And Hitler turned into a baby, Billy Pilgrim supposed. That wasn't in the movie. Billy was extrapolating. Everybody turned into a baby, and all humanity, without exception, conspired biologically to produce two perfect people named Adam and Eve, he supposed.
They were actually named Adam and Steve, but Steve was a woman's name in the original human language.
We're #1, we're #1, we're #1 USA USA
http://91-divoc.com/pages/covid-visualization/
There has to be some under counting of death in the US because of the fact that so many people will refuse to see a doctor here. Also it is still early for us in terms of transmission. People have to have this for 3 or 4 weeks before they start dying, and relatively few are that far along here.
My favorite recent item of news: https://twitter.com/i/status/1242994018725908486
Also many of the active measures Trump took to increase the spread of the virus are very recent so they mostly haven't yet been included in the statistics. I'm thinking in particular of him having homeland security hold huge numbers of people close together at airports.
I'm thinking about mailing this one to some contacts but maybe everyone already knows more than they want
https://www.reddit.com/r/Residency/comments/fp1vt1/3_days_of_inpatient_care_in_new_york/
I'm on the City of Hayward's email list, so I got this report from the first day of operation of the city's new walk-up/drive-up test center. 54 of 207 tests administered Monday came back positive. Now that's from a group of people who met the screening criteria and were extra likely to be symptomatic, but it's still clear evidence that the disease is prevalent in the community. https://mailchi.mp/hayward-ca.gov/first-day-results-from-hayward-covid-19-testing54-of-207-positive-for-infection?e=39c42ae2eb
You can find the screening criteria in the article, but the key is that this center doesn't require a doctor's order for the test, and can test up to 370 people per day. It sounds like they will be doing some mobile outreach to vulnerable groups in the community, in addition to testing on site. They plan to report each day's results to the public with around a 48-hour delay, so you can check back with the city's web page later to get future results.
Hayward is a city of around 150,000 people in the SF Bay Area.
Collective fail on following WHO guidelines:
Seek information updates at specific times during the day, once or twice. The sudden and near-constant stream of news reports about an outbreak can cause anyone to feel worried.
4: Re; The Hopkins datasets; is it just me or are the US time series not there . (It is in the daily report files, but those are not convenient for looking at the trends without a lot of work.) The visualization in 35.1 use the JH data and show the US as well as world; and provides a link but from what I can see it is the World minus the US data. I must be missing something.
I don't go for visualizations, I just read the API straight-up.
I don't know if this has been posted but Bing's Covid tracker is very good, especially for checking the curves of various countries. The US doesn't look good at all.
39: I really need to start following those WHO guidelines. I have several sites that I check regularly (the CBC for day-to-day updates; WaPo for an American angle); and while those sources are more-or-less trustworthy, I check them way too often throughout the day, and I need to cut back to once or twice every 24 hours.
The Canadian federal govt's Coronavirus disease site is reassuringly boring, which helps calm me down.
Really interesting concept for visualizing growth in new cases over time, which makes it really distinctive when a country has got growth under control.
45: That is quite interesting.Thanks.
Here's a study that attempts to analyze how well different states are complying with social distancing, according to cell phone data.
After watching a minute and a half of 45 I felt a compulsion to research stellar evolution.
47 is quite similar to the electoral map, even down to the county level.
45 is neat, thanks. I learned my dad is generating his own plots out of curiosity; he'll like that idea.
I think I am going to scrub Keynes off my list of approved prophets after reading this, from 1930:
I see us free, therefore, to return to some of the most sure and certain principles of religion and traditional virtue--that avarice is a vice, that the exaction of usury is a misdemeanour, and the love of money is detestable, that those walk most truly in the paths of virtue and sane wisdom who take least thought for the morrow. We shall once more value ends above means and prefer the good to the useful. We shall honour those who can teach us how to pluck the hour and the day virtuously and well, the delightful people who are capable of taking direct enjoyment in things, the lilies of the field who toil not, neither do they spin.
With the NYTimes county dataset available for download I have made my own spreadsheet to look at per capita rates. Have not started on time series stuff yet, but here is the cataegorization for most recent data for counties with the highest rates:
Over 1000 cases/million population
Ski counties: 4 in Colorado, 1 in Idaho, 1 in Utah.
NYC and environs (in addition to NYC, Rockland, Westchester, Bergen (NJ), Nassau, Orange and Suffolk).
New Orleans and 2 suburban counties.
A rural Arkansas county where there was a spread in a church.
Two counties in SW Georgia.
Many caveats apply of course, including local testing protocols and availability, and things like whether they are reporting hospital/where administered versus residence.
Looking at death rates, the data is much noisier, but was does pop out is New Orleans and some rural counties in Mississippi and Georgia.
53: It appears that the Mississippi case was a date problem. The Georgia ones are not( county with Albany and a neighboring county.)
Why would you even put dates in soda bread?
So, we're at 147 now. 57 are in Gallatin County -- a ski county -- up 11 from this mornings count. Still 9 here.
Toole County, up at the border, has 5 cases. All women. Feel free to distract yourself with this: http://www.boxing.com/dempsey_vs._gibbons_the_fight_that_ruined_a_town.html
And I hear that one of our cases is a friend of a friend, an ER doc. Who'd tested negative and then tested positive.
We're at 590 now, with a first reported fatality.
Someone pointed out to me that Drum's compilations of data that was linked to earlier is per capita and that seems wildly inappropriate, you're going to have a bunch of local outbreaks and the initial low numbers won't be affected by the size of the national population, in the earliest stages not the local population size either. Right?
But the earliest raw numbers seem to small to draw super strong inferences from, and you can't extrapolate from them when there are very different geographical-demographic factors in different countries and other factors, and you don't see the effect of lockdowns in deaths until some time has passed.
Infection rates are comparing apples and oranges to some extent because of very different testing regimes.
So I figures hospitalization rates is what you really want, but I don't see anyone compiling that? At least here in Sweden, there is data.
I don't know. The advantage of comparing per capita to me is that you can better compare if hospitals are likely to be overwhelmed. I was mostly pushing back against absolute number comparisons.
It seems your friend wants a metric that distinguishes between one swamped city vs 100 brand new outbreaks scattered all over a country? A map seems well-suited for that.
62 was to 60.
To 61, why are hospitalization rates more compelling than death rates?
oh, because of the shortened lagtime to see the effect of social distancing?
This is frightening
https://twitter.com/brett_mcgurk/status/1244031461486784512
Anyone have an opinion (Mossy?) on Radio Free Asia?
Seven funeral homes currently serve Wuhan -- a huge conurbation of three cities: Hankou, Wuchang and Hanyang.
Social media users have been doing some basic math to figure out their daily capacity, while the news website Caixin.com reported that 5,000 urns had been delivered by a supplier to the Hankou Funeral Home in one day alone -- double the official number of deaths.
Some social media posts have estimated that all seven funeral homes in Wuhan are handing out 3,500 urns every day in total.
Funeral homes have informed families that they will try to complete cremations before the traditional grave-tending festival of Qing Ming on April 5, which would indicate a 12-day process beginning on March 23.
Such an estimate would mean that 42,000 urns would be given out during that time.
If this isn't made up from the whole cloth it is plausible. As has so often been said, we need to trust what the Chinese government does, not what it says.
How many people usually die during that time?
Maybe they're just stocking up on urns like everyone else is stocking up on toilet paper.
I started plotting MI's data in the style of 45. It dropped today! I am hoping so hard it's real, but if it holds, the stay-at-home orders are working. Most of the Detroit hospitals are near capacity, and there's statewide coordination to maybe take overflow in hospitals with rooms, so I'm not sure how bad it will be at peak for hospitalizations, but I'm starting to feel a glimmer of hope about not having thousands of preventable deaths here. I'm so worried about everywhere else, still, though.
67: AFAIK RFA does solid reporting. The underlying investigation for all reports evidently is being done by Caixin, which has a good reputation. As I linked elseblog:
There were 56,007 cremations in Wuhan in the fourth quarter of 2019, according to data from the city's civil affairs agency. The number of cremations was 1,583 higher than those in the fourth quarter of 2018 and 2,231 higher than the fourth quarter of 2017.From the official stats starting here:
Q4 2019 - 56,007
Q3 2019 - 42,151
Q2 2019 - 29,625
Q1 2019 - 14,700
Q4 2018 - 54,424
Q3 2018 - 38,149
Q2 2018 - 28,717
Q1 2018 - 14,833
If we take 42,000 urns to approximate Q1 2020 cremations, that's 27,300 excess deaths, ~185% of Q1 2019; implying death rate ~5X the official number, which IIRC is consistent with with rough OSINT based analysis (also centered on crematoria) I saw and linked a week or two ago. I'll try to find that and relink. Obviously a bunch of heroic assumptions built into all this.
Also, science people: do those Wuhan numbers look right? Does one normally see so much variation across the year?
I don't know. America mostly uses ground burial and sky burial instead of cremation.
Speaking of data questions, California has 64,400 pending test results as of a couple of days ago.
Someone posted something about a federally mandated limit on the number of tests allowed to be done each day in Dallas? Is that a real thing or misinformation or a misunderstanding? Are they being told to save materials or are they just being stopped so there aren't so many positives that it looks bad? And is there such a federal limit applied anywhere else?
71: Those numbers do look a little funny. It's not super easy to find data from other countries, but either it shows higher cremation rates from Q1-Q4 or higher mortality rates. I assume there may be some variability on what % of mortalities get cremated, but I find it really hard to believe either 4X as many people die between Oct-Dec as Jan-Mar or cremations become super popular at Christmas.
It would be a weird way to fudge data, though. Some quick back-of-envelope: China's all-cause mortality rate, estimated, in 2017 was 7.8 per 1,000 people, est'd. 2017 population was 8.09 million. That would give you roughly 63,000 deaths per year, 15,800 per quarter. I'm not sure what to think, but now I'm very curious!
As the eldest of four daughters and no sons, I hold the title to our family's burial plot at a large RC cemetery in the city (because my dad was very insistent, in his final days, that I do the paperwork to transfer that title...). There are six people buried there that I know of (my parents; my paternal grandparents; my paternal great-great-grandparents). There is room for two more cremated remains in that plot; but no room for another coffin burial.
Writ large, I don't even want to think about this might mean...
Um, is there any particular reason you think somebody you don't know of is there?
Dorothy Sayers cos-play getting out of hand.
Um, is there any particular reason you think somebody you don't know of is there?
There is! Headstones were expensive; and therefore many (especially pre-20th century) were buried without a headstone, or any sort of marker at all.
Isaac Chotiner is out with another one of his expert dissections of a charlatan. This time it's Richard Epstein, a lawyer who thinks his horrifyingly wrong COVID-19 theories are better than the epidemiologists', and somehow managed to influence the Trump administration.
||
The HS tight-linked to my workplace has been released from quarantine all clean-tested. Yay.
|>
||
He subscribed a solemn treaty, with the hope of deriving some advantage from the term of its continuance, and the moment of its violation.
|>
The link in 81 is wonderful. Chotiner's ability to get blowhards to write their own epitaphs deserves a Pulitzer:
My view about it is what you're asking me to do is, when I think everybody is wrong, to remain silent, and the stakes are too high. So my view is there's all these experts on the other side.
71: those figures really, really look like cumulative numbers rather than per quarter. I refuse to believe that almost exactly twice as many Chinese people die in Wuhan in Q2 as in Q1, and almost exactly three times as many in Q3 as in Q1, and so on.
Another thing that might be skewing the numbers is that ISTR reading in January that the province had banned burials - so that all dead, whatever the cause, had to be cremated. That would spike the cremation numbers a bit - though I can't imagine many people in a massive city were burying their dead anyway.
I think you must be right about those being cumulative totals.
81 is indeed great. Epstein sounds borderline deranged, though he landed one good point: Chotiner is just a journalist and he doesn't actually have any better grounding in this stuff than Epstein does. He has a politics degree.
And so all he can say is this rather weak sniping about "he says X but I talked to someone else later on in square brackets and they said not-X". Why didn't they get someone to write the article, or at least conduct the interview for Chotiner to write up, who actually did know more about it than Epstein?
71 How does that track with Italy? .About 4 times the excess deaths over the reported COVID-19 fatalities.
86: that is really, really embarrassing for Bloomberg if so.
Not to mention that, what, are we all just trusting data coming out of the Wuhan civil affairs agency now? Because, what, they have this fucking stellar reputation for being unflinchingly honest when it comes to reporting the spread of this exact epidemic? We're all just going "sure they literally sent the secret police round to shut up a doctor who was warning about the true state of the COVID-19 epidemic, but we're pretty sure they'd never do it again?"
It is by far the safest approach to simply ignore all public health and epidemiology numbers coming out of China right now. Just say "maybe they're right, maybe they're wrong, but we cannot use them as the basis for any sort of deduction about the true state of the epidemic in China." The medical stuff, treatments and case histories and so on in the Lancet, I would be fairly confident about. The rest? Come on.
87: That seems like an odd criticism. Chotiner does have a better grounding in this stuff than Epstein because he's been reading and believing actual experts, in the same way that both you and I, neither one of us epidemiologists, have a better grounding in this stuff than Epstein. It's possible to learn and understand things as an adult without having a degree in them. The references to credentialed experts in brackets are for people who don't themselves have that grounding, so who can't use their own judgment and knowledge to tell that Epstein is not making sense.
Do you apply the rule that no journalist should work in any area where they don't have superior academic credentials to any interview subject who might be in error generally, because you can't correct someone on a matter of fact without superior academic credentials?
88: that article is good but it makes one logical leap that isn't well founded: that all excess deaths during the pandemic are the result of infection. They may well all be the result of the pandemic, but that's not the same thing. Italy's healthcare system has been overwhelmed. If you turn up at a clinic in Nembro with, I dunno, peritonitis right now, safe to assume that your chances of a happy outcome are lower than they would be at normal times: there are fewer doctors because some are sick, there are fewer hospital beds (especially ICU) because they're full of COVID patients, and the medical care you get will be poorer quality because staff will be tired and overworked, because working in PPE makes you less efficient and less competent, and because supplies will be short. If you die, you won't have died of COVID-19 or even have died with COVID-19, but there's a very real sense in which you died because of COVID-19.
I see that Josh Marshall makes the same point: https://talkingpointsmemo.com/edblog/this-is-very-important-from-italy-please-read
89: See "heroic assumptions".
Cumulative numbers makes sense. The period in question though is Q1.
The numbers when published had no doubt been cooked for whatever reasons CCP city cadres cook cremation figures. The question is whether these figures have been retroactively cooked on account of Covid. I can't evaluate that.
The test will be checking the Q1 2020 figures against independent evidence. Like IDK urn tenders and satellite pictures of crematoria.
91 I'd like to see figures for both excess deaths total as well as possible undetected COVID deaths.
90: no, I think that someone with superior knowledge - whether formally taught or not - would have made a better job of the interview in this specific case than Chotiner did. Chotiner didn't confront him there and then with things like "but evolution towards avirulence, if it happens at all (and the idea that it inevitably happens hasn't been taken seriously since the early 1990s), takes years; it doesn't happen over weeks during a single pandemic, to any appreciable degree, and it didn't happen at all for HIV as far as we can tell".
Chotiner even gets his rebuttals wrong. Epstein says that syphilis evolved towards avirulence. And Chotiner sits there and nods and then later on gets on the squarebracketophone and says "[Syphilis is a bacterial infection, not a viral infection. "One doesn't have anything to do with the other," Kuritzkes said"...]
But that's irrelevant. Evolution towards avirulence happens, or not, regardless of the type of pathogen. It's about the epidemiological characteristics of the pathogen, not its cladistics!
Do you apply the rule that no journalist should work in any area where they don't have superior academic credentials to any interview subject who might be in error generally, because you can't correct someone on a matter of fact without superior academic credentials?
Well, do you apply the rule that there is no reason for journalists to build up any experience on any particular topic, because every journalist, regardless of background, education, or previous experience, is equally capable of writing about any highly specialised topic? Get some guy off the sports desk, put him in to cover the Supreme Court for a day, he'll do fine?
Nope -- obviously Chotiner does know more about this stuff than Epstein, as he should. But the fact that he has a politics degree doesn't really enter into it.
fwiw, I think Chotiner did a magnificent job: the skill of getting someone to condemn themselves from their own mouth is rare, important, and to some extent works against possessing academic expertise. If I were an epidemiologist interviewing this guy I would have had a very hard time not bursting out with "But that's raving bullshit!" very early on and then it would have been shut down at once and we'd have learned a lot less.
That would have made it entertaining as a twitter spat, in the genre of "our side crushes the other bastards" but a proper interview should be more revealing than that.
It is a core journalistic skill to be able to get a rough working competence in a subject about which you know nothing reasonably quickly. I had a colleague on the Indie would would say quite literally to someone down the phone "Explain like I'm a child of eight" -- and it worked. If Chotiner had a week to prepare for this interview (not impossible, given the New Yorker) he'd have prepped up thoroughly on what the experts think of all this, and internalised it to an extent sufficient to ask the right questions. Which is all you need.
With all that said, I completely agree that the case of syphilis is irrelevant. I can think of one quasi-mechanical reason why it might be harder for an RNA virus than a bacterium to evolve towards non-lethality -- that its genome is so small that there is less room for variations that arise from sequence doubling. But that is just a guess. I know something about journalism and rather less about the biology of pathogens.
If your issue is that you'd prefer Epstein to have been confronted by someone who could convince him he was wrong by speaking from a place of greater authority, that would have been a very different interview, but it also wouldn't have had Epstein exposing his ignorance as completely, probably. Chotiner's technique, and I think fits a good one, is to give his interviewees enough rope to hang themselves with, and maintaining his position of "we're just two laypeople talking about this stuff, I only know what people tell me" was a good way of getting Epstein to show the weakness of his position.
99 was crossed with NW who makes the same point better. On the syphilis thing -- Chotiner quotes experts saying evolution towards avirulence hasn't happened with HIV, doesn't happen generally on any short timescale, and there's no evidence it's happening with Covid-19, which covers evolution toward avirulence generally. I think the response to Epstein's invocation of syphilis that it's not even a virus and has nothing to do with the current situation, in the broader context of what's said in the whole article, seems to me to work fine.
85/86: That's better! I went to the source data but don't read Chinese so gave up.
The sidebar doesn't qualify for a bailout and went to live with its parents.