Models

I have a feeling that our response to this pandemic may lead to some reflections on the role of scientific models in the decision making process. I would normally err on the side of defending scientific advisors, but I have a sense that they might face some justified criticism. I, of course, don’t know all the details of what information was presented, how it was presented, what pressures the scientific advisors faced, and how the decisions were made. However, it does seem as though many – who should probably have known better – failed to recognise the strengths and weaknesses of the scientific models that were being used.

Scientific models typically allow us to ask “what if” questions: What will happen if we do nothing? What will happen if we encourage social distancing? What will happen if we enforce a partial lockdown? What about a full lockdown? What will happen if we wait a week before doing something, rather than starting now? etc. They’re typically more properly presenting projections, rather than predictions; they’re predictions that are conditional on us actually following the scenario that was modelled. There’s also always some level of uncertainty, so the questions should maybe be more properly phrased as “what could happen if….?”

However, we seem to be treating these models as predictions without always making clear that there is quite a lot of uncertainty involved, both in terms of how they model the infection itself and in terms of how they’re handling the various possible societal scenarios. Of course, it’s important to test models against real world data and if there is a good match, and if there is confidence that the assumptions in the model closely match what happened in reality, then one can be confident that the model is capturing many of the important processes. However, it’s still important to remember that all scientific models are simplified representations of reality that can never really capture all the complexity.

Another important aspect of using scientific models is to sanity check the results; do they make sense? It’s not clear that this has been done particularly well in this current context. James has been highlighing this in a number of his posts. Specifically, some of the leading researchers were still presenting numbers that no longer seemed reasonable. For example, suggesting that the lower limit to the number of deaths might be around 7000, when we were already pretty close to getting there [edit: see update at bottom of post.].

There’s probably a lot more that could be said, and I may return to this topic at a later stage. I think it’s important for people to recognise both the strengths and limitations of scientific models. They can be very powerful tools, but they’re never going to perfectly represent reality. The scientists involved should be willing to acknowledge this and should, in my view, also be checking that their model results make sense. Decision makers should also be aware that scientific models have strengths and limitations; they can certainly guide decision making but can’t really define it. I don’t think this takes anything away from the usefulness of such models, it is simply something that I think is important to recognise.

Update:
As Steve Forden points out, there was a stage where the lower bound for the number of deaths (5000) was presented at the same time as the group was projecting this number of deaths for the following week.

This entry was posted in Philosophy for Bloggers, Research, Scientists, The philosophy of science, The scientific method and tagged , , , , , , . Bookmark the permalink.

461 Responses to Models

  1. Jim Hunt says:

    Afternoon Ken (UTC),

    You probably have better things to do than follow my Twatter feed closely? However I feel compelled to point out at this juncture that I have been following UK Covid-19 modelling efforts very closely, with the emphasis on the sterling work of a certain James Annan:

    http://GreatWhiteCon.info/2020/04/covid-19-in-the-united-kingdom/

    I also have some “Shock News!” to impart, hot off our locked down presses way down here in SilicInny Valley:

    Using the once Great British vernacular, I have a feeling that at long last I have Andrew Neil by the short and curlies. However he may well have other ideas, in which case any and all “likes” and/or “retweets” would be much appreciated.

  2. Jim Hunt says:

    P.S. Dear Mod – Please r/Silc/Silic/ in my prior musing:

    https://twitter.com/hashtag/SilicInnyValley

    TIA

    [Mod: Fixed, I think.]

  3. Dave_Geologist says:

    WRT your update ATTP, an additional point is that not only are they projections not predictions, the same researcher can make apparently contradictory projections at the same time, using the same model but different parameters within an uncertainty range, or with different assumptions about government action or the effectiveness of government action. Another level of “if this, then that; if the other, then another”. And of course in a fast-moving situation the run was probably a few days before the newspaper report, when deaths were lower.

    I don’t see anything wrong with that, but it’s hard enough to explain to a knowledgeable audience, let alone comprehend after its been filtered through journalists. I would expect a constantly updated lower bound to be routinely falsified a few days after it was made, because reality is probably somewhere near the middle and we’d expect the uncertainty range to narrow as the lower bound in particular is shifted upwards by new data. And where we’re dealing with growth having doubling times in the order of a week, it’s entirely reasonable for predicted deaths in a week to be comparable to deaths to date, which is by a data definition rather than a model definition the lower bound of total deaths.

  4. Dave,

    the same researcher can make apparently contradictory projections at the same time, using the same model but different parameters within an uncertainty range, or with different assumptions about government action or the effectiveness of government action.

    Indeed, and – as you say – there is nothing necessarily wrong with this. They could represent different possible outcomes. The problems come in when they’re both presented as near-term predictions, rather than as two outcomes that will depend on what we end up actually doing.

  5. Dave,

    And where we’re dealing with growth having doubling times in the order of a week, it’s entirely reasonable for predicted deaths in a week to be comparable to deaths to date, which is by a data definition rather than a model definition the lower bound of total deaths.

    Except this isn’t – as I understand it – what happened. It seems that the same group presented a projection for total deaths (not simply deaths to date) that was essentially the same as their projection for the number of deaths for the coming week.

  6. Dave_Geologist says:

    Yes ATTP, I realised I was confounding matters somewhat with implying that the lower bound is what has happened, whereas the model lower bound is probably a reasonable lower bound projection at the time the model run was set up.

    There isn’t enough information in the thread to understand whether he was referring to two different models*, mean lower bound in a literal sense (distancing worked a treat and deaths fell off a cliff the next day), whether they were estimates a few days apart in a fast-moving epidemic, or even if the journalist had misunderstood and confused lower bound for deaths next week with lower bound for deaths in total.

    * He did say in the Nature piece that they ran the agent-based model as their gold standard, but that to quickly assess multiple interventions they used a simpler differential-equation-based model, benchmarked against the agent based model. I’d have thought a factor of two failed the benchmark test, but maybe not if you assume the question is “by what %age will action X reduce deaths relative to action Y”, not “how many deaths do you expect from action X”. Box’s adage applies.

  7. Joshua says:

    > I think it’s important for people to recognise both the strengths and limitations of scientific models.

    I think this is the most important point.nit to assume that here weren’t important modeling errors, but as we’ve sernd with climate change, most people don’t really get it when models create projections they aren’t predictions, that understanding CIs and probability ranges is
    hugely important, and that modeling in situations s like COVID-19 are extremely sensitive to particular parameter settings.

    IMO, the main thing at issue here isn’t the science of modeling, but the science of how people process risk and conditional probability (particularly when there is an overlay of ideological biases).

  8. Joshua says:

    .nit to assume =, not to assume

  9. Joshua says:

    Oops. Sorry, Dave. I didn’t mean to post something so similar to what you posted, as if you hadn’t posted it.

  10. Joshua,

    IMO, the main thing at issue here isn’t the science of modeling, but the science of how people process risk and conditional probability (particularly when there is an overlay of ideological biases).

    Indeed, and I worry that we’re going to get choruses of “we can’t trust models” rather people concluding that models need to be used in ways that take into account their strengths and limitations.

  11. Joshua says:

    Anders –

    > Indeed, and I worry that we’re going to get choruses of “we can’t trust models” rather people concluding that models need to be used in ways that take into account their strengths and limitations.

    Is already happening in full throat in the US. I find it very distressing. I’m in the odd position of being worried that the deaths will be less than some of the projections.

    Of course I’m glad that fewer people will likely die. Of course I’m glad that the chances are lower that I will die, or my family, or friends.

    But I am worried that people who don’t really j derstand modeling will leverage their lack of understanding to use modeling (and modeling error) to harden their ideological biases.

    The same type of problem exists in public health more generally. Those with a partisan agenda will ALWAYS have the voice to say either that (1) public health officials are alarmist or (2) that they acted too slowly.

  12. Joshua,

    Those with a partisan agenda will ALWAYS have the voice to say either that (1) public health officials are alarmist or (2) that they acted too slowly.

    Yes, it is sadly rare to see a partisan commentor highlighting that maybe we could have done better than we did, but that given what we knew in advance, it was always going to be difficult to know what action would be best. Hindsight being 20-20……

  13. Joshua says:

    > Yes, it is sadly rare to see a partisan commentor highlighting that maybe we could have done better than we did, but that given what we knew in advance, it was always going to be difficult to know what action would be best. Hindsight being 20-20……

    It runs both ways. I find myself doing it. That’s part of what makes me sensitive to it in others. It really disturbs me that so many are convinced they’re immune to the pattern themselves, even as they’re very focused on pointing it out in others.

    Looking at the warring factions at WUWT with respect to the wisdom of mandated social distancing is at least a kind of clown show example that makes observing the pattern a bit less exasperating.

  14. Joshua,

    It runs both ways. I find myself doing it.

    Yes, I do too. It’s one reason I’m a little reluctant to bash the scientists involved too much. This isn’t easy and it’s not that much of a surprise that it hasn’t gone all that smoothly at times.

  15. Joshua says:

    Anders –

    I listened recently to this pod interview with Daniel Kahneman (starts at about 19:30)

    https://castbox.fm/vb/240741784?_t=19%3A35

    Nothing terribly profound – but it has me thinking more about “noise” and how bad we are in understanding the role of “noise” when we evaluate risk.

  16. Joshua,
    Thanks, I’ll have a listen.

  17. anoilman says:

    I wouldn’t look too closely at the models for the COVID-19 pandemic.

    Right now, modelers are being expected to produce a lot with very little information, and they are being expected to produce all this very fast. All they can generally say is ‘bad’, but its hard to say exactly how bad. Even how we react to it affects how bad it will be. Frankly the health care system really just wants to know what kind of supply issues it will be facing, like number of ventilators. (How is the creeping lack of health care modeled?)

    Many models for pandemics contain numeric guesstimates that you just can’t generalize. For instance what measurement do they use for social connected-ness? I mean if we’re all meant to look at Iceland’s super high % testing as an example, what does it really mean? Are they prone to the exact same size and quantities of gatherings that we are in Rural Canada, or New York City? How tight packed in are they? How do you measure social isolation and the benefits it will have?

    Any aggregate numbers must be different for cultural reasons. I’m reminded of how Ebola spreads. In Africa its common to see the body of a person that has died and pay your respects. However that’s precisely when the body is most infectious. So clearly Ebola can spread more easily in cultures where seeing the dead is a cultural expectation.
    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4412251/

    China’s ‘cure’ wasn’t medical, it was military. Is there a model that includes that? Is there a model with a sliding scale for military intervention?
    https://www.visualcapitalist.com/infection-trajectory-flattening-the-covid19-curve/

  18. Jim Hunt says:

    “Modelers are being expected to produce a lot with very little information, and they are being expected to produce all this very fast. All they can generally say is ‘bad’, but its hard to say exactly how bad.”

    Did I neglect to mention that “climate modeller though probably doing more epidemiology in the last couple of weeks” James Annan’s “projections” have been beating the learned epidemiological modellers that advise assorted governments hands down?

  19. NN Taleb continues to deal knockout punches to anyone misinterpreting probabilities

  20. Russell Seitz says:

    The Bayesians have thus far been rather quiet.

    I expect we’ll hear from them soon enough, because globally, the actuarial priors transcend the spectrum of coronavirus mortality projections– mortality from all causes in the year preceding the Wuhan outbreak was just over 60,000,000, and there remains to be discovered what fraction of that death toll came from diseases that are unrelated to Covid19, but also rendered less infectious by social distancing, masks and increased disinfection.

    The modelers are , quite rightly, back-burnering social distancing phenomena tthat will unfold in retrospect to focus on the problem at hand.

    Unfortunately, collateral damage may also arise from panic driven self medication. Some climate denialists seem equally susceptible to the medical mis- and dis information being touted by their colleagues :

    https://vvattsupwiththat.blogspot.com/2020/04/of-quinine-and-chloroquine-willis.html

  21. Steven Mosher says:

    I dunno it seems pretty clear that the models will always skew high, especially here
    where the Deaths (95%+) are all over 50years old with co morbidities and the cases
    are skewed to the youth which are more socially interactive.
    Plus no epidemic modeller wants to err on the low side.

  22. Steven Mosher says:

    By overestimating the outcomes modllers have set themselves to not believed in the
    next important question. Los Angeles county modelers are estimating that if we go back to work
    95% of the population will be infected by august.

    Lunatic assumptions as we reach herd immunity before then
    Lunatic assumptions since the hospital system would be overwhelmed.

  23. Jim Hunt says:

    Mornin’ All (UTC),

    Needless to say neither BoJo or Andrew has got back to me yet, so let’s try a slightly different tack shall we?

  24. Joshua says:

    Anders –

    As for that pod, I’m feeling a little embarrassed for linking it because Kahneman was pretty vague and unfocused.

    But I found it interesting nonetheless because I’ve never really thought before specifically about the distinction and interplay and contrast between noise and bias. I’m wondering if there isn’t a positive relationship between the level of noise in a situation and the potential for people to leverage bias when there is political “motivation” to do so.

    But then there’s variance.

    Oy. This is too complicated for me.

  25. Joshua,
    I did listen. As you say, he didn’t say all that much. However, it is an interesting issue and something I’ve thought about myself without framing it in the way he did. I often think that we’re far worse at assessing things than we think we are. As academics we spend quite a lot of time reviewing other people’s proposals and then have panels who rank these to determine who should get funded. Although I think we’re probably okay at working out which ones are truly excellent, and which ones are dreadline, I think we’re probably very poor at assessing those that are somewhere in the middle (probably good enough to be funded, but neither outstanding nor dreadful). When deciding which of these to actually fund, I think it becomes a bit of a lottery, even though a funding panel would probably not admit this.

  26. jacksmith4tx says:

    RE: Kahneman
    His opinion on replacing human judgment with algorithms was interesting. In previous interviews he has repeatedly stated that due to human nature (future risk assessment) humanity will not be able to avoid some of the more dire outcomes of climate change. My intuition is we will resort to genetic engineering and geo-engineering but on a finite planet somebody (AI?) will have to figure out how to bend the curve on population.

  27. Marco says:

    “When deciding which of these to actually fund, I think it becomes a bit of a lottery, even though a funding panel would probably not admit this.”
    https://www.nature.com/articles/d41586-019-03572-7

    See also
    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4959526/

  28. Jim Hunt says:

    I realise that the UK coronavirus lockdown is no laughing matter, but I’m afraid I’ve been unable to stop myself so doing for several minutes:

  29. Joshua says:

    Marco –

    Thanks for that. That’s interesting. In response to that article, my brother texted me:

    > Interesting — there was a post quite a few years ago arguing that NIH reviewers do a pretty good job of detecting bad proposals, say the bottom 80%? (don’t remember the actual number) but a lousy job of stratifying the best ones, and thus arguing for a randomized choice among those 20%? along with a system where if you make that cutoff but don’t get chosen you stay in the pool for the next round with having to rewrite and resubmit the whole proposal.

    Needless to say NIH was not able to accept that reality. And I have seen a more recent article claiming that they actually do reasonably well even in the top 20? , I don’t recall how convincing it was.

    There was also a study at perhaps the top machine learning / AI conference, then called NIPS and now called NeurIPS, where they set up a parallel review track for some subset of papers to see the level of agreement and found considerable randomness there too. I think it was great that they did the study.

    But what did they learn from it, and how did they change their practice? Not at all, they apparently decided the appropriate response was to keep doing the same thing they have been doing.

    He then lknked this:

    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5866547/

  30. Marco,
    Thanks, I hadn’t seen those. I’ve seen people suggest that some telescope time should be allocated via a lottery, but I didn’t realise that some funding agencies were already doing something like this. I think it does make some sense, as long as you make sure to eliminate those that are clearly not suitable, and maybe have some form of selection to pick out those that are clearly exceptional.

  31. JCH says:

    A second paper, this time on confirmed cased in San Francisco, finds most people with COVID-19 in San Fran have the European variant.

  32. Joshua says:

    JCH –

    Clearly, if you wrote such a thing, you are motivated by nothing but hatred for Trump.

    All those who think that Trump is in any way fallable, are motivated by nothing but hatred for he who leads us.

    Trump is our fearless leader.

    He is big. He is strong. He is powerful, He hates Jim Acosta. And Nancy Pelosi.

    He is a strong daddy.

    All those things make him a dear leader.

  33. Joshua says:

    Sortable:

    Cases per million – US ranks at about 200 of about 220 countries listed:

    https://www.worldometers.info/coronavirus/

  34. anoilman says:

    Well, on the climate science topic, is there a way to model such a massive decrease in carbon emissions? I know that post first gulf war, there was a competition between all the various climate models to see how much and how accurately there would be regional heating from lighting Kuwait’s oil fields. The effect of the pandemic will show as regional variance with different timelines for decrease and the inevitable bounce back.

    In North America, most energy flows into industrial production, and all non-essential businesses are shut down.

    Oh Fudge! Climate scientists didn’t predict future global pandemics! Their numbers will be off! SH*T! Queue the “Models are running hot!” crap. Again.

  35. Steven Mosher says:

    thanks Dave.

    “There have been 10 influenza
    pandemics in the past 250-plus years – two started in the northern hemisphere winter, three in
    the spring, two in the summer and three in the fall. All had a peak second wave approximately
    six months after emergence of the virus in the human population, regardless of when the initial

  36. Joshua says:

    Steven –

    You need to have a talk about population density with your good friend Willis. From his recent post:

    > Meanwhile, out of modelworld and back in the real world, Sweden is doing quite well, thank you very much. I’ve highlighted Sweden’s trail in red/black. Note that Sweden is on the same path as Switzerland, which has had lots of interventions & quarantine regs.

    And that’s not even the dumbest thing he’s written on this topic.

    Are his posts on climate science equally bad?

  37. David B Benson says:

    A longer term modeling study:
    https://medicalxpress.com/news/2020-04-on-off-social-distancing-harvard.html
    with dismal societal and economic implications.

  38. Dave_Geologist says:

    Indeed David. It’s what the March Imperial College study said. Although they didn’t run it so far into the future, and presented a counter-factual of a severe epidemic this winter if we didn’t reintroduce social distancing after a break, it was obvious that it would take years to get to herd immunity without massive deaths among the vulnerable, and that there was a risk we never would if immunity only lasted a season or two.

    Hence the need for a vaccine, even an annual one, and for treatments that allow more people to become infected and not die (and also avoid the need for ventilation and the the months of recovery required after that).

  39. Marco says:

    Joshua, there’s lots more to consider than just population density. Cultural/societal differences can have a major impact on virus transfer, too. Sweden has 50+% single-person households, Switzerland well below 40%. Swedes hug friends, Swiss kiss friends (at least man=>woman and woman>woman; three times, even). It is also my impression that Swedes have a smaller group of people with whom they regularly mingle.

  40. Joshua says:

    Marco –

    No doubt, there are many behavioral and environmental factors. But just the simple difference in population density (567 per square mile versus 64) makes Willis’ comparison as dumb as it gets.

  41. Ben McMillan says:

    Frankly the most pertinent thing on Switzerland vs Sweden is their relative distance to Lombardy.

    But if you want to read something sensible on Sweden, I recommend James’ article:

    Click to access operational.pdf


    Doesn’t look great.

  42. JCH says:

    If they live in single households, it just slows it down. Germany’s death rate on confirmed cases on worldometer keeps climbing. You have to attack this virus to stop it. They welded doors shut. Every time you say “influenza”, hit your fingers with a ruler. There may be no herd immunity. There may be no vaccine. There may be no significant therapeutics.

    Wake up to eradication. If the only way, way bigger than WW3.

  43. Joshua says:

    In all fairness, Sweden has a higher % living in urban areas (a good bit smaller in “intermediate” areas, about the same in rural)

    http://ec.europa.eu/eurostat/documents/2995521/7020151/3-05102015-BP-EN.pdf/bf18a8b3-998c-476d-b3af-58292b89939b

  44. Jim Hunt says:

    James Annan has submitted a COVID-19 themed paper to medrxiv:

    https://bskiesresearch.wordpress.com/2020/04/14/model-calibration-nowcasting-and-operational-prediction-of-the-covid-19-pandemic/

    Wait for the fireworks to begin!

  45. Joshua says:

    Jim –

    Indeed.

  46. Jim,
    Yes, I saw that James and Julia had submitted a paper. I’m interested to see how this plays out. James does make quite a convincing argument that some of the prominent models seem to have used parameters, or made forecasts, that seem obviously wrong. I am, however, always slightly concerned that things are not quite as obvious as they seem, and that there is a reason why the domain experts are doing what they’re doing.

  47. Jim Hunt says:

    I happily admit to being an old cynic Ken. I generally go with the “cock up” theory of history, but this time I’m not so sure. See for example:

  48. verytallguy says:

    On Sweden… Just had an absolutely mind blowing conversation with a Swedish colleague.

    She’s working from home, but only because her daughter is off school sick with a persistent dry cough for the last 10 days. As with the UK, she will only be tested if she is admitted to hospital.

    Her son, meanwhile, rather than being quarantined, is obliged to attend school as he is asymptomatic, and she would be breaking the law if she kept him at home!

    Depressingly, 170 new deaths today in Sweden. I guess inflated by Easter reporting, but that’s over 1,000 pro rata to the UK and more than the entire epidemic in adjacent Norway.

  49. Marco says:

    JCH – indeed, single-person households will just slow down the spread.

    That makes it one of several potential explanations as to why Sweden and Switzerland appear to be (for now) on a similar path despite movement in Sweden being much less restricted than in Switzerland.

  50. Ben McMillan says:

    I think a bit more thought needs to be applied to the claim that these two countries are ‘on a similar path’: Sweden is at least a week behind Switzerland, and the deaths lag by ~15 days, so what the alignment in these curves means is mostly a similar growth rate before social distancing was imposed (in CH). What happens next? The problem, as usual, is choosing data that doesn’t answer the question of interest.

    Sweden vs Denmark, Norway and Finland answers that question.

  51. Everett F Sargent says:

    One knock on their paper would be its timing (very close to 14th April 2020 data in hand) and the fact that each day is a new day in the data set to be rerun with that daily update.

    If they had produced this a month age, then it would be a different situation. in other words, hindsight is 20/20.

    The accuracy of any model is strongly dependent on how well it does without updating.

    In the very early stages of the pandemic we were at 2-3 orders of magnitude accuracy, tens of thousands to tens of millions globally.

    In climate models IMHO, we don’t have anything like those levels of uncertainty.

  52. Jim Hunt says:

    More fireworks from James Annan. His COVID code is now open sourced on GitHub:

  53. Willard says:

  54. Joshua says:

    Ouch!

    I stumbled across a RPJr. tweet about his death (without noticing the date) and so then Googled it and came across that article (without noticing the date)…..

    I guess I should do a better job of staying current…

  55. Joshua says:

    Doh!

    Should have bothered to look at the tweet he was responding to…

  56. Joshua,
    Strange. When Roger returned to the Twitterverse he decided to block me. I thought I wouldn’t be able to see embedded tweets too, but it seems that I can.

  57. William Gray commanded this to his students: “Thou Shalt Not Bow Before Computer Terminals Nor Involve Thyself With Numerical Models”

    That explains a lot

  58. Joshua says:

    https://www.medrxiv.org/content/10.1101/2020.04.14.20062463v1

    (Iaonnidis a contributing author).

    > These prevalence estimates represent a range between 48,000 and 81,000 people infected in Santa Clara County by early April, 50-85-fold more than the number of confirmed cases.

    Good to see the results of a proper investigation with an attempt at a representative sampling. The limitations in that regard should be noted:

    > This study had several limitations. First, our sampling strategy selected for members of Santa Clara County with access to Facebook and a car to attend drive-through testing sites. This resulted in an over-representation of white women between the ages of 19 and 64, and an under-representation of Hispanic and Asian populations, relative to our community. Those imbalances were partly addressed by weighting our sample population by zip code, race, and sex to match the county. We did not account for age
    imbalance in our sample, and could not ascertain representativeness of SARS-CoV-2 antibodies in homeless populations. Other biases, such as bias favoring individuals in good health capable of attending our testing sites, or bias favoring those with prior COVID-like illnesses seeking antibody confirmation are also possible. The overall effect of such biases is hard to ascertain.

    Seems odd to me that they didn’t discuss limitations (or adjusted weighting) with respect to SES. I would hope that a weighted adjustment for SES is buried in the methods section.

    I wonder if they adjusted to make the sample nationally representative, or just representative for the Santa Clara community. I suspect it is the later:

    –snip–

    The population density in Santa Clara is 2504% higher than California.
    The median age in Santa Clara is 5% lower than California.
    In Santa Clara 44.09% of the population is White.
    In Santa Clara 3.70% of the population is Black or African American.
    In Santa Clara 40.40% of the population is Asian.

    –snip–

    Median income in 2018 was @ $116,000. About four times the national median income…

    Hmmm. I hope that they are clear to establish that if they were seeking a sampling representative for Santa Clara, it is in no way representative for the nation.

    That would likely be much less relevant for infection rates – but it’s extremely important for mortality rate numbers.

  59. Ben McMillan says:

    As another datapoint, official coronavirus deaths are now 0.15% of New York City and counting (if I’ve got my numbers right!). Probably not everyone has been infected, and you can draw your own conclusions about what infection mortality rates might be. But this, and Lombardy’s figures, seem to put a pretty unhappy lower bound on them.

    That’s what ‘low’ mortality looks like.

  60. JCH says:

    <a href="https://www.ksbw.com/article/stanford-develops-gold-standard-in-antibody-testing/32112632#&quot;Stanford develops ‘gold standard’ in antibody testing

    The Food and Drug Administration allowed the use of COVID-19 antibody tests in March but did not review any of the tests. As of April 7 the FDA had received notifications from 70 developers that their tests were in use but only one from Cellex Inc. had received Emergency Use Authorization.

    The Stanford test does not have FDA approval but it has gone through a clinical validation process.

    “It is very rigorous and FDA approval is not required but there is a separate sort of mechanism in ensuring that all of our results we are getting are accurate,” said Zehdner.

    Zehdner said the level or rigor is what makes Stanford’s test the ‘Gold Standard’ of antibody tests versus the finger prick point of care tests that are more widely available. …

  61. Steven Mosher says:

    Joshua

    ‘Good to see the results of a proper investigation with an attempt at a representative sampling. The limitations in that regard should be noted:”

    the study is shit.

    a start but shit.

  62. JCH says:

    Peer Review of “COVID-19 Antibody Seroprevalence in Santa Clara County, California”

    I can’t see where he says “the study is chit”. He’s not that kind.

  63. Ben McMillan says:

    Just to summarise: almost all the Santa Clara positives could have been false positives. The German study was better here because the proportion infected was so much higher. Also, if you offer free corona tests on Facebook, you are going to get the people most worried that they’ve had it turn up, rather than a ‘random sample’.

    But even if this study’s numbers were right, that’s still an awful lot of dead people if you let the epidemic run unchecked. Which brings us back to the usual issue that the denominator is still not ‘big enough’. Yes, there are a lot of hidden infections, but it doesn’t impact policy much in the short term.

  64. Dave_Geologist says:

    Even if the caveats about the Stanford study are ignored (and it’s only a not-yet-peer-reviewed preprint), it is not extremely important for mortality rate numbers.

    They’re suggesting a 0.1-0.2% mortality rate according to press I’ve read (sorry, no time to read the preprint at the moment). The study comparing the age-stratified Diamond Princess death rate to China found 0.5% overall (including asymptomatic cases in the denominator). That was about a 50% downgrade of the Chinese 1%*, because the DP cohorts were half as likely to die as their equivalent Chinese cohort. I’ve mentioned before that that may be because wealthy cruise ship passengers who are up for a 10,000 mile flight are certainly wealthier and probably healthier than retired farmers or factory workers living on a meagre pension, which would mean wealthier Western cohorts should be somewhere in between (except obesity-related co-morbidities might outweigh smoking so all bets are off).

    So, setting aside the other problems, if it’s representative they’ve reduced the death rate by somewhere between 2.5 and 10 times. Which, if we open the floodgates, would change an unacceptably high death count to an still-unacceptably-high death count. Every year if immunity fades like with cold coronaviruses. So from a decision-making viewpoint, it’s a “meh….”.

    * 4-5% is a naive estimate of mortality rate, comparing deaths now to cases now while the epidemic is still raging. The best studies track case fatality rate; the second best estimate how long ago the dead were infected and compare deaths today to cases 17 days ago; the gold standard would be to compare counts when it’s all over, but we can’t wait for that. The Ferguson team were obviously doing it right, hence 500,000 unmitigated deaths to get herd immunity to 50,000,000 people.

  65. Joshua says:

    JCH and Steven –

    With respect to that review, as I posted earlier…

    If you are looking to find something that happens 1 case in 1000, Assume you always find it when it is there. Assume that your false negative rate is 1%. Then in 1000 cases you find 1 true positive and 10 false positives, so the probability that you have a true positive given a positive test is <10 %

    Of course, those numbers would have to be adjusted per their results, but it seems to me that trying to extrapolate from that survey to a national fatality rate is more an exercise in faith than science.

  66. Dave_Geologist says:

    Ah, I see my DP point partially replicated Joshua’s: a small, relatively wealthy, relatively isolated population is not a representative demographic for a disease which disproportionately kills people with pre-existing health and wealth co-morbidities.

  67. Dave_Geologist says:

    Aargh, and I meant to add that I’d misread the mortality point as being “maybe it’s not so bad after all” rather than “maybe these are folks who are less likely to die of it”.

    My first para still applies, but without the emphasis.

  68. Joshua says:

    Dave –

    Regarding the representativeness of sampling from a cruise – yup. Judith deleted my attempts to get Nic to addtess that question on his sevond post after he flat ignored my question on his first post.

    Regardong that Santa Clara study…

    > Those impressive-looking numbers, though still leave the possibility of a significant number of people having false-positive and false-negative tests. For example, if 5% of the US population actually had the virus, a test with 95% sensitivity and 95% specificity conducted in a million people would correctly detect 47,500 cases, along with 2,500 false-negatives (those who were infected but were missed).

    But it would also produce 47,500 false-positives. So, if only 5% of the population was infected, the number of true-positives and false-positives would be the same.

    “Imagine going to a healthcare worker and saying ‘we’re going to test you for antibodies right now, and if you’re positive, you have a 1 in 2 chance it’s not real,'” says Michael Osterholm, PhD, MPH, director of the Center for Infectious Disease Research and Policy (publisher of CIDRAP News). “Are you actually going to use that test in a meaningful way?”

    https://www.cidrap.umn.edu/news-perspective/2020/04/antibody-tests-may-hold-clues-covid-19-exposure-immunity-its-complicated

  69. Ben McMillan says:

    Joshua: the review JCH linked actually did the exercise you are suggesting, using the data in the Santa Clara paper, and showed that the cases identified could actually all be false positives.

    i.e., when you write “those numbers would have to be adjusted per their results” this has in fact already been done. So I think you are confusing things by talking about a lot of hypotheticals.

  70. Dave_Geologist says:

    Joshua, one of the profs (Bell I think) made a similar point re antibody testing allowing “normality” on the BBC Today programme this morning.

    Paraphrasing, since it looks like at most a few percent are infected so far, testing non-essential workers might free up at best a very few percent of the workforce. Imagine going to a car factory manager and saying “you can restart production, but only with 1-2% of your workforce; and you don’t get to choose which ones”.

  71. Chubbs says:

    Just noodling some #. Taking the ratio of estimated to confirmed cases from this study of 50 to 80X and applying to US gives the following:

    Total cases (706k) = 35 to 64 million

    New cases last week (213k) = 11 to 17 million

    Guess I will skip a trip to the store for TP this week

  72. Dave_Geologist says:

    Another datum on asymptomatic numbers: in a young, presumably healthy population, not a cruise ship. Guess what? About 50% asymptomatic (the article says 60% is significantly more than 25-50%, but that’s silly given Fauci was giving round numbers and that some may yet develop symptoms, and in any case it doesn’t make any policy difference unless it’s 96% not 60%). Coronavirus clue? Most cases aboard U.S. aircraft carrier are symptom-free

    Defense Secretary Mark Esper, speaking in a television interview on Thursday, said the number of asymptomatic cases from the carrier was “disconcerting.”

    And 25-50% wasn’t disconcerting? Or was he just in denial until the Navy measured it? Either way, that is disconcerting. Or perhaps it’s the fact that the Captain was so summarily dismissed and has been proved right that is disconcerting, politically (not just in the scale of the outbreak, but in what might have happened if they’d continued with their eyes shut and their fingers in their ears).

  73. Joshua says:

    Dave –

    > Imagine going to a car factory manager and saying “you can restart production, but only with 1-2% of your workforce; and you don’t get to choose which ones”.

    This is what doesn’t make sense to me about the re-opening that some people envision here in the US. Suppose a governor says that businesses vsn reopen, and a factor or office ties to. But some workers, maybe some of them having preexisting conditions, don’t want to risk their lives by coming in to work. Does the business owner fire them? Hire replacements? Limp along with a depleted work force? And being open but not operating at capacity, the business owner has most of the overhead of normal operation (in other words more than when closed) and so might be losing more money when open than when closed.

    So those conditions really mean a healthy economy? And suppose infections take place, and spread continues more rapidly than it would have under stay at home orders.

    Can people not realize that such a situation could actually mean a limping economy for an extended period of time? More deaths? More serious illness? More Healthcare workers infected? Do people not realize that certain essential entities, such as may not have enough personnel to run effectively or safely?

    Do they not realize that this isn’t merely a choice between sickness/death and economic health?

    Again, we run into the difficulty people have with conditional probability. All some people see is the economic risk of everyone staying home. Thst risk is so big and so scary and so immediate that it fills their entire screen, and they can’t see anything past that

  74. Joshua says:

    Chubbs –

    > Guess I will skip a trip to the store for TP this week.

    Yup. If these extrapolation are right, basically “opening the economy” means a shit ton of infectious people running around with not practical application of contact tracing. Essentially, that means a lot of people being sacrificed. We don’t yet know the exact relative risk increase for older people or people with co-morbidities, but it’s obviously not insignificant.

    How many working people are there out there with co-morbidities or who are over 65? Is the idea that the economy is going to be up and running smoothly with them all staying home? Are they supposed to go to work not knowing if their co-workers are infectious?

    These people are nuts.

  75. Joshua says:

    Ben –

    > So I think you are confusing things by talking about a lot of hypotheticals.

    Maybe. But I’m thinking of the hypotheticals in the sense of the implications of extrapolating from the indivual study. Regardless of the specifics of that Santa Clara study, it seems irresponsible to me to suggest that (1) we can use it to generalize about infections and even more so mortality, across the country and (2) we can really gain much from extrapolating from these studies until we have extremely effective tests.

    Maybe I just don’t understand, but it seems to me that putting out a study such as that one, and allowing it to generate so much publicity that in effect leads to people confidently downplaying the risk to society (its al over Fox News and echo chambers like WUWT), is highly irresponsible unless the authors make sure to present and foreground the uncertsinty and unsuitability for national extrapolation.

  76. Joshua says:

    Another example!

    https://www.foxnews.com/science/third-blood-samples-massachusetts-study-coronavirus

    At least they somewhat tested a random sampling, unlike the Sant Clara sampling…. oh wait….

    They tested in a hotspot. They tested people out and waking around, and who are waking around in a state with shelter run place orders. Thst isn’t random sampling.

    And there’s no discussion of uncertainty due to false positives/negatives.

    Are the potential public health implications of such publicity completely lost on the researchers?

  77. Ben McMillan says:

    Joshua: as usual, I don’t seem to be communicating my point clearly, but such is life, and I’ll just move on.

  78. Joshua says:

    Ben –

    More likely I’m the one not getting clear communication.

  79. JCH says:

    There is a political movement to end the lockdowns. They have soldiers.

    Meanwhile, another meat plant in the Dakotas has a COVID-19 patient. These places are like cruise ships on calm waters. Rumor, CDC already on site. Sioux Falls started with one, so we’ll see how this cruise goes.

  80. JCH says:

    Herd immunity percentage depends on the virus. They fail to thrive when the immune system kills them before they can find a new host. So it could be 50%; it could be 90%.

    If you grew helping Dad the Vet fight viral diseases, and he dealt with some nasty ones, you saw this in real time. The disease usually disappears rapidly. Poof. They stop dying; they stop getting sick; they go back to getting fat.

    So look. What do you see? The percentage of the population that is effectively isolated and unavailable to the virus, call it 70%. You know, because they’re bragging they’ve flattened the curve with social distancing, etc. 327 million Americans, that’s 229 million people with “herd” immunity. 80X times confirmed cases, 713 thousand, is 57 million, so 286 million quasi immune: 87%.

    COVID-19 should all but gone in about 14 days… long before they have time to liberate all the blue states from their socialist masters.

  81. JCH says:

    Statement from Prof Edward Holmes on the SARS-CoV-2 virus

    Unfounded speculation on the origins of the SARS-CoV-2 virus that causes COVID-19 has prompted the following statement from Professor Holmes, who was involved in mapping the genome of the virus. …

  82. Joshua says:

    Ben –

    Here’s what I don’t get about these estimations of massively more infections than the identified infections. Granted, there must be more with the high rates of asymptomatic infections, but along with more of those we’d also have many more cases of pneumonia or other illnesses associated with the disease prior to when the testing started. Additionally, I’d think there would be many more deaths. Where was the spike in mortality, cases of pneumonia, etc., prior to the onset of widespread testing and prior to the mandated shelter in place orders?

    Presumably, these spikes would have occurred on top of the typical morbidity rates associated with the flu? I have yet to read of any surveys to detect a signal in prevalence of pneumonia prior to testing and shutdowns.

  83. John Hartz says:

    Something to keep front and center…

    The most difficult thing for an epidemiological model to predict: human behavior.

    The above statement is the lede for the following analysis by Brian Resnick, a top-drawer science journalist who does his homework when drafting an article.

    Why it’s so hard to see into the future of Covid-19 by Brian Resnick, Science & Health, Vox, Apr 18, 2020

    https://www.vox.com/science-and-health/2020/4/10/21209961/coronavirus-models-covid-19-limitations-imhe

  84. izen says:

    Given the wide uncertainty about the sensitivity and specificity of CORV test, and the highly variable coverage and targeting between Nations and locality, the usefulness of numbers for detected positive cases as a measure of the spread and fatality rate is dubious.
    The recent data on asymptomatic cases may make some difference to some models, but the bottom line is the death toll.
    And as all the revisions to include non-hospital deaths makes clear even that has deficiencies.

    The Nations who have dealt effectively with the outbreak and contained the spread like S Korea, Taiwan, and Iceland all applied the same method. Hospitalise the serious cases, trace their close contacts and isolate/quarantine them. Ideally test the contacts and extend the tracing to THEIR contacts if positive.

    Testing is important and useful, but may be unnecessary. If you hospitalise the serious cases, and can remotely diagnose those with symptoms, then they get isolated and effective contact tracing and 14 day quarantine for all contacts can work.
    Without testing, identifying the definite positive cases is difficult. But a triage protocol that gives false positives would be better than a system that misses infective cases.
    But contact tracing is difficult, unless you have a connected society and can utilise the personal smart-phone use to detect everyone’s position and movements 24/7.
    That may raise privacy concerns Nations/societies with a more individualistic, less communal, political ideology.

    At present Apple/Google are working on apps that could do contact tracing, informing any person who has used the active app if they were in close proximity to another app user who subsequently is diagnosed or tests positive.
    Apparently there is some friction between the Apple/Google side and the EU/UK proposals.
    The implication is that Apple/Google are looking at ways for the system to be local and closed. Only the app user has access to the data on their proximity to a positive case. Governments, inevitably, want the system to have a central database and the location and proximity data to be available (although ‘anonymised’!? ) to the ‘appropriate central authority for oversight… and enforcement.

    Perhaps others here can tell me if a purely local, closed system is possible. I suspect it is using a version of the public/private key encryption systems as used in the ‘Pretty Good Privacy’ app. That has the individual generate a public key that enables all who receive is to encode a reply. A one-time ident perhaps, along with their own public key.That can be decoded with the private key of the original key-giver. If a person becomes a positive, infection risk case, then they can send a general message encrypted with their private keys. That can only be de-crypted by those who have the corresponding public key validating the identity of the sender. If each public/private key is generated by each contact, then a general notification encrypted with the private keys, by one app, would be recognised as a warning to quarantine by all who had received the public key by proximity bluetooth.
    All this would be local and anonymous as far as I can see.
    Does anyone know if this would work and if there are any loopholes or problems that would scupper such a system ?

  85. Steven Mosher says:

    “. Also, if you offer free corona tests on Facebook, you are going to get the people most worried that they’ve had it turn up, rather than a ‘random sample’.”

    yep.

    the best evidence of this is the counts of people who showed up with fake IDs and IDs that didn’t match the survey. 101 cases like that.

    Plus the rich white ladies got to bring a child.

    When they get around to doing a proper random sample they might want to also stratify for things
    that matter:
    works in an office? yes or no
    takes mass transit? yes or no

    things like that. but I think they will just focus on race data and gender data
    rather than behavioral stratifications

  86. Steven Mosher says:

    “That makes it one of several potential explanations as to why Sweden and Switzerland appear to be (for now) on a similar path despite movement in Sweden being much less restricted than in Switzerland.”

    Swedes are moving around less than the Swiss in some classes of travel it depends on what type of movement, by destination

    Percent change in movement
    Swedes: Grocery store -15%
    Swiss: grocery store +11%

    In other categories ( like mass transit) both are in negative territory ( -36%,-27%)

    Policy doesn’t matter: Behavior matters.
    The virus doesn’t care about your policy, it cares about getting into your nose, eyes or mouth

    For the most part mobility in Sweden in down dramatically matching other countries
    The only exception is travel to parks. Sweden is +84% in travel to parks.
    It’s a good thing parks don’t have door handles.

    Oh, in china 80% of transmission was family to family.
    sick cases were removed from their homes.

    50% of swedes live alone. I’m guessing they don’t have family to family transmission in
    households of 1

    here is a recipe for trouble

    take a sick person. confine them at home with their healthy family.
    confine the whole family to the house and only go out for groceries.

    Korea? travel to the park is up, but we keep our distance.

  87. Steven Mosher says:

    “Yup. If these extrapolation are right, basically “opening the economy” means a shit ton of infectious people running around with not practical application of contact tracing. Essentially, that means a lot of people being sacrificed. We don’t yet know the exact relative risk increase for older people or people with co-morbidities, but it’s obviously not insignificant.”

    well in Korea 11% of the cases are over 60, 92% of the deaths.
    similar figures elsewhere.

    Opening the economy doesn’t mean a shit ton of infected people running around.

    china opened it’s economy in mid Feb when positive cases were still being reported. they key is
    having an r0 <1. which meant reducing the dominant mode of transmission: family to family.
    Korea is open, we still have cases but R0 <1. the dominant modes of transmission are being
    managed: hospitals, nursing homes, churches, tightly packed business operations.

    HK, opened its economy having an r0 <1. How? again dominant modes of transmission controlled
    Hospitals closed to visitors since Jan 4th, Nursing homes closed, business with human to human
    contact limited.

    in China here was the protocal.

    1. Every business was responsible for ensuring their workers were isolated for 14 days and reported temperature. Personally I was out of luck since the reporting app was in Chinese.
    2. Everyone who could telecommute did telecommute.
    4. On Feb 15th the office was opened to 50% . managers made seating charts and attendance
    charts. flex on, flex off
    5. Housekeeping staff was in full force. Not a dirty surface, door handle, table top in the building.
    Normally they walk around and do spot cleaning, during re opening, busy as heck.
    6. sign into work, get a mask.
    7. temperature taken twice daily by security guards, sign the paper, it gets a red stamp.
    8. All food delivery was downstairs in the lobby. QR code for the delivery guy to enter lobby.
    9. All delivery bags sealed and marked with a red stamp including the delivery persons name
    some stuff about his temperature being normal and hands washed before and after handling
    the package. Red stamps for everything.

    oh wait..
    feb 14th

    https://www.reuters.com/article/us-china-health-fast-food-focus/fast-food-companies-in-china-step-up-contactless-pickup-delivery-as-coronavirus-rages-idUSKBN20A0RS

    so rather than hand wringing they just went about solving the problem.
    Not perfectly, not to ensure maximum safety .

    There's a bunch more to learn from China, Taiwan, HK, Singapore and Korea.

    Here is a funny one. New York has decided it needs to hire and train an army of contact
    tracers. In Singapore they just assigned the phone calling to the army.
    Basically, call the positive cases, get their family names, friends names, credit card info
    and phone location data.

  88. Steven Mosher says:

    “But contact tracing is difficult, unless you have a connected society and can utilise the personal smart-phone use to detect everyone’s position and movements 24/7.
    That may raise privacy concerns Nations/societies with a more individualistic, less communal, political ideology.”

    Korea is kinda cool when it comes to transparency in government.

    Yeah Open data!

    Here is the data you need.

    1. family members names: 80% of the cases an infected person will cause are
    within the family.
    2. Friends names.
    3. Co workers names.
    4. Places you shopped ( credit card location data, not amounts)
    5. Places you visited. Train stops, stores, events, busses

    data 1-3 captures the vast vast majority of infections you will cause;

    data 4 and 5 requires phone data & credit card data,

    Google already has this data. credit card companies already have this data.
    You have all the data you need.

    How is this data used?

    It’s mostly used to warn others. It is not without it’s issues.

    https://www.newyorker.com/news/news-desk/seouls-radical-experiment-in-digital-contact-tracing

    Mapo district is a little more aggressive
    than my district.

    But you don’t NEED phone data to do the most important part of the job. 80/20 rule.

    Citizens accept this not because they are collectivist. it makes sense and google already has the data. might as well put to good use rather than evil use.

  89. izen says:

    @-SM
    “Citizens accept this not because they are collectivist. it makes sense…”

    I suspect your confidence in the rationality of people is misplaced.
    Citizens, and subjects, accept things because they do not care, do not know, and do not abnegate their cultural norms and the hierarchical social order.
    Sometimes they may even think it is in their self-interest.

  90. Ben McMillan says:

    Joshua: there is a big uptick in symptoms/general mortality, now, near the peak of the epidemic, which is what you would expect if there are large proportion of hidden cases (maybe you think this should happen early on?).

  91. Dave_Geologist says:

    Ben (I’m pretending to be Joshua 😉 ): it’s also what you’d expect if social distancing had reduced the R factor below 1, but delayed a few weeks to allow time for people to go to hospital and die.

    The fundamental problems with postulating large numbers of asymptomatic cases are that (a) where you had more-or-or-less closed populations, it was always around 50% and (b) places like New York and Lombardy would already have more than 100% of the population infected.

    Apply meta-analysis rules. Which is objectively a better sampling regime: an isolated ship or a Facebook ad campaign? What do you find with large samples vs. small samples? What have peer-reviewed papers found vs. preprints? Do some of the analyses (Santa Clara) have issues like positive rates which overlap the 95% range expected from the test’s documented false-positive rate? Do some extrapolate from very small sample numbers (50 positives, binned by zip code so only a few per zip code). Do some of the analyses make patently ridiculous out-of-sample predictions (more than 100% infected; the downturn is always coming next week or next year – remind anyone of solar cycles?). Then ask yourself: does the asymptomatic rate go up or down when you compare good/strong analyses with poor/weak analyses.

  92. Ben McMillan says:

    Dave: OK, I think part of the problem is that ‘large’ without putting a number on it is too vague to be really useful (I know I’m guilty of this), so there is a risk of people talking past each other, or spinning any ‘large’ number as meaning we don’t need to worry (I disagree with this!). I guess I mean ‘large’ as in ‘at least 50% of infections in the UK have not come to the attention of the health system’. I’m primarily thinking about symptomatic cases that weren’t tested rather than asymptomatic cases.

    Indeed, the sensible thing would be to try to put together various different pieces of evidence, according to how reliable they are. Agree that what happened in NYC or Lombardy, is inconsistent with very low mortality estimates. If all you want is a lower bound on mortality, this is pretty hard to argue with. What you don’t do, is go hunting for the lowest possible figure available, treat it as gospel, and ignore all the other evidence.

  93. Dave_Geologist says:

    Ah. I wouldn’t call that large Ben. I’d call it exactly what you’d expect. Small compared to seasonal flu. Separated by a common language and all that 😦 .

  94. Steven Mosher says:

    “Of course, those numbers would have to be adjusted per their results, but it seems to me that trying to extrapolate from that survey to a national fatality rate is more an exercise in faith than science.

    Infection is local
    death is local

    I dont think anti body testing is a silver bullet. It might help you plan body bags required.

    Might help in some local areas.

    and this

    https://today.rtl.lu/news/science-and-environment/a/1498185.html

  95. Joshua says:

    Steven –

    > Opening the economy doesn’t mean a shit ton of infected people running around

    I was talking about here!. (I assumed that it was obvious I meant here. )

    Where there is a rather sizeable % who are saying to open the economy with no comprehensive plan for the logistics of contact tracing, in a country where the kinds of policies used in those other countries will meet a lot of resistance. In a country with a more individualistic way of being, without a more collectivistic sense of duty to society. In a county where the federal government is insisting that testing is the resonsiboty of the states, and where state governments are saying thst they don’t have the needed resources, and where many are already staring into deep financial holes. States where a main source of income, sales tax, has taken a big hit. States where people are running around in flack jackets and brandishing machine guns, with the encouragement of the president, saying that mandated social distancing is an infringement of their freedom.

    It is theoretically possible that public health entities and state governments will be able to enact a comprehensive policy of contact tracing, but my guess is thst in this country, opening the economy is going to mean sacrificing a lot of people..

    It is my opinion that in this country, it is likely that a much higher percentage of older people , and people with co-comorbidities, are going to be infected than in other countries, because a lot of people are saying to open the economy. And let people run around, without effective policies to isolate and protect more vulnerable populations. Older people, people with comorbidities, people of lower SES, institutionalized people, native Americans, homeless people.

  96. Joshua says:

    Ben –

    > (maybe you think this should happen early on?).

    Yes, I did mean early on. Prior to when we started testing and identifying cases.

    Presumably, (well, I presume) for the numbers of infected people walking around to be 85x higher than what we have identified, there would have been some signal in the mortality and morbidity prior to what we’ve seen associated with identified cases?

  97. Ben McMillan says:

    Joshua: no, I don’t think that ‘if the infection mortality rate is low, there should be lots of undetected early deaths’ is a logical conclusion (I don’t understand your logic).

  98. Joshua says:

    Ben –

    I was saying if the infection rate is high (I.e., 85x what we have identified). And I was not speaking only of mortality – but morbidity also (i.e., cases of pneumonia 3 weeks or longer prior to when the first cases started to be identified).

  99. Joshua says:

    Ben –

    And I wasn’t referring to infection mortality (rate) but infection rate.

  100. Joshua says:

    Steven –

    I just happened to read this:

    https://judithcurry.com/2020/04/14/in-favor-of-epistemic-trespassing/#comment-914534

    Liberate Minnesota!

  101. Jim Hunt says:

    Not strictly on models of course, but WTF:

    Liberate Florida!

  102. Jim Hunt says:

    Meanwhile back on topic, James Annan’s preprint is up on medRxiv:

  103. Jim Hunt says:

    And whilst I’m here,

    It’s not often that I praise the reporting in the assorted organs of News UK. Usually quite the reverse! However this morning I commend to you this frankly shocking article by the Sunday Times Insight team:

    http://CoV-eHealth.org/2020/04/19/how-britain-sleepwalked-into-disaster/

    I strongly suggest that you read the article from start to finish, always assuming that you have a strong enough stomach…

    Merely business as usual in the age of “Fake News” and “Truth Decay“.

  104. Dave_Geologist says:

    I see some in the USA are competing for the Darwin Awards…

  105. Steven Mosher says:

    “I suspect your confidence in the rationality of people is misplaced.
    Citizens, and subjects, accept things because they do not care, do not know, and do not abnegate their cultural norms and the hierarchical social order.
    Sometimes they may even think it is in their self-interest.”

    like I said. it makes sense

  106. Steven Mosher says:

    “Not strictly on models of course, but WTF:”

    walk at the beach? nothing wrong with that.

    Swedes travel to parks is up 84%
    Korean’s enjoying the park up 24%

  107. dhogaza says:

    “What you don’t do, is go hunting for the lowest possible figure available, treat it as gospel, and ignore all the other evidence.”

    Including a very low mortality rate based on slender evidence of a very high undetected number of infected people.

    The Stanford serological testing exercise is going to be pushed as evidence that distancing directives can be dropped, and the economy opened up, because covid-19 is at worst only about twice as deadly as the flu. Ioannidis (one of the authors on their paper), Senior Fellow at the Hoover Institution, has been opposed to distancing directives from the beginning. He’s poo-pooed this epidemic in the past, using Diamond Princess data to suggest that the US deaths might top out at 10K (we’ll be at 40K by tonight). And before they even began testing another Senior Fellow at the Hoover Institution got a piece published widely in California claiming that the relatively low confirmed case count and mortality in the state couldn’t be explained by social distancing, etc, and that the point of the antibody testing project was test the hypothesis that herd immunity due to a very high undetected infection rate was the cause.

    Of course, the study should stand or fall on its merits, but I’m a bit suspicious of the entire effort. The fact that those being tested were self-selected from within those sent the facebook ad, questions about false positives, no real knowledge how the threshold of detectability of antibodies relates to immunity in practice would suggest they should be very careful about how they present their results. Rather than lay a PR foundation before the testing even began, which they very carefully did.

  108. Jim Hunt says:

    Steven – So you’d fancy your chances strolling PPE free along Jacksonville beach at this precise juncture?

    Will you be “watch[ing] Florida infection figures in 2-3 weeks time”?

  109. Joshua says:

    dhogaza –

    > Ioannidis (one of the authors on their paper), Senior Fellow at the Hoover Institution, has been opposed to distancing directives from the beginning.

    Interesting. Not that it means he’s wrong (up to recently I have a lot of respect for his work) but it would be interesting if he’s an “activist.”

  110. Joshua says:

    Oh – do you have a link thst shows that?

  111. Joshua says:

    As we discuss errors modeling – from Ioannidis:…

    > If we assume that case fatality rate among individuals infected by SARS-CoV-2 is 0.3% in the general population — a mid-range guess from my Diamond Princess analysis — and that 1% of the U.S. population gets infected (about 3.3 million people), this would translate to about 10,000 deaths. This sounds like a huge number, but it is buried within the noise of the estimate of deaths from ‘influenza-like illness.’ If we had not known about a new virus out there, and had not checked individuals with PCR tests, the number of total deaths due to ‘influenza-like illness’ would not seem unusual this year. At most, we might have casually noted that flu this season seems to be a bit worse than average. The media coverage would have been less than for an NBA game between the two most indifferent teams.

    Let’s note that he said if 3.3 million get infected. Total cases now, 741k. So assuming we get to 3.3 million, and assuming the current mortality continues, there would only be about 176k dead.

    10k vs. 176k. Why quibble about details?

    But let’s acknowledge that as Ioannidis said, it’s hard to model accurately when you are working with so much uncertainty.

  112. Joshua says:

    Of course, given the Santa Clara study, Ioannidis seems to think that the current number of infections is more like 37 million to 67 million.

  113. Joshua says:

    Actually, since the Santa Clara study estimated between about 2.5% and 4.1%, maybe Ioannidis only thinks the number infected will be around 8 million to 13 million?

  114. dhogaza says:

    His editorial published on March 20th is actually a bit more nuanced, as he doesn’t explicitly state opposition to social distancing directives.

    But he does claim that such actions were being taken without any meaningful data to back up them up (it was published days after the Bay Area issued its seven county SIP directives). He does some extrapolation of data from the Diamond Princess to support the possibility of a very low mortality rate in support of his claim that there was not sufficient data to support such action. Oddly, he never mentions data from China before and after isolation efforts were put in place, nor does he mention what was happening in Lombardy on March 20th (the shit was hitting the fan).

    Anyway, his editorial:

    https://www.statnews.com/2020/03/17/a-fiasco-in-the-making-as-the-coronavirus-pandemic-takes-hold-we-are-making-decisions-without-reliable-data/

    Regarding his being an author of the paper, it’s easy to find their preprint.

  115. John Hartz says:

    It is my understanding that the Santa Clara study is currently under peer review — by an expert panel, not by the people posting on on this comment thread. Is this correct?

  116. dhogaza says:

    Joshua:

    OK, I see that you’re familiar with his editorial.

  117. dhogaza says:

    John Hartz:

    Is it already in the hands of reviewers? Obviously it does need to be reviewed.

    However, the PR campaign around this study, which began before it was underway, pretty much guarantees that a certain segment of the population (including some political leaders) will run with what’s already shown up in the press, even if peer review suggests that the study is flawed.

  118. Joshua says:

    dhogaza –

    Do you have a link for his Hoover affiliation?

  119. dhogaza says:

    Here’s an interview by the Hoover Institution with another author of the study, Dr. Jay Bhattacharya, on March 27th, following up on his WSJ editorial on March 24th.

    https://www.hoover.org/research/questioning-conventional-wisdom-covid-19-crisis-dr-jay-bhattacharya

    “We get into the details of his research, which used data collected from hotspots around the world and his background as a doctor, a medical researcher, and an economist. It’s not popular right now to question conventional wisdom on sheltering in place, but Dr. Bhattacharya makes a strong case for challenging it, based in economics and science.”

    Again, the study needs to rise or fall based on its merits, but …

  120. Joshua says:

    JH –

    It isn’t only the people posing on this thread that are considering the in-peer reviewed publication that is getting so much attention. Perhaps people discussing it here isn’t such a bad idea?

    I will begin by amplifying an important announcement you should have received yesterday from NYP. Serologic testing for antibody to the SARS-CoV-2 virus is now available for health care personnel. Those eligible have been symptom-free more than 14 days after a confirmed infection, or after a “COVID-19-like illness,” and “were cleared to return to work” by WHS. The statement implies that only those who contacted WHS and were excused from work are eligible. Since a request for serologic testing can only be granted after a screening call to WHS, all eligibility ambiguities should be resolved in the WHS screening process. Even if you’re not certain about your eligibility, call them. While this sounds frustratingly narrow in scope, I assure you that NYP hopes to screen all health care workers as soon as possible. I believe it’s worth waiting for the NYP ELISA assay to be expanded in scope. It is ~99% specific and >80% sensitive 3-4 weeks after infection, significantly superior to the widely hyped quickie assays that are much less accurate and potentially misleading.

    Unless you’ve been without power and internet, and don’t read, you will have heard about the serologic-testing study done by Stanford University investigators on 3300 people in Santa Clara County over April 3-4. They concluded that there were 50 to 85-fold (!!) more cases in the County than had been detected by other diagnostic means (~1000 cases). Please note that this manuscript is still undergoing peer review. If even half-true, at the simplest level, this would greatly decrease the case fatality rate by greatly increasing the denominator. It also implies an enormous pool of asymptomatic infections, which has very important implications for the level of herd immunity we may have already achieved. The greater our herd immunity, the less we need to worry about epidemic surges after relaxation of mitigating measures. That reassurance is qualified by how little we know about the immune response, particularly what level of protection coronavirus seropositivity conveys. Nonetheless, this is potentially great news for reopening, which explains all the attention.

    […]

    Craig R. Smith, MD
    Chair, Department of Surgery
    Surgeon-in-Chief, NYP/CUIMC

    So people are making plans to open up surgery centers, to some degree justifying the decision based on the un-peee reviewed but heavily publicized Santa Clara study.

    Would it be better were we not to discuss it?

  121. Joshua says:

    dhogaza –

    Do you have a link to confirm your statement that he is a senior fellow at the Hoover institution?

  122. Joshua says:

    Oy. peee reviewed.

    OK. Time to liberate the recent comments listing now. I’ll be quiet.

  123. dhogaza says:

    Hmmm, apparently I was confused about Ioannidis and the Hoover Institution, sorry about that. Probably because of the piece put out Victor Hansen who really is a Senior Fellow there. If I could edit my post, I’d do so. But it appears I was wrong, and I apologize.

  124. John Hartz says:

    Joshua: What the practical outcome of gumming the Santa Clara study to death on his thread? Is it just a way to wile away the time?

  125. Joshua says:

    JH –

    One last comment for a while. I don’t happen think there’s much of a practical outcome from pretty much most interactions on blog comment threads. I don’t see discussing the Santa Clara study any differently in that regard.

    Is there a practical outcome from not
    discussing it gumming it to death?

  126. John Hartz says:

    ATTP: Fodder for a new OP? 🙂

    The Mafia is poised to exploit coronavirus, and not just in Italy by Valentina Di Donato and Tim Lister, CNN, Apr 19, 2020

    https://www.cnn.com/2020/04/19/europe/italy-mafia-exploiting-coronavirus-crisis-aid-intl/index.html

    Of course there is a risk associated with publicly probing into Mafia activities as some of the regular commenters of the site would be prone to do.

  127. John Hartz says:

    Joshua: One practical outcome of participating in a comment thread is learning more about the subject matter. Generally speaking, that requires more listening than talking.

    Things are going to hell in a hand-basket throughout the world because of the COVID pandemic and man-made climate. Like you and everyone else commenting, here, I an extremely frustrated and angry that the human race doesn’t have the moxy to collectively deal with these issues in a meaningful manner.

    I apologize to you and other commenters for directing my frustration and anger toward you.

  128. John Hartz says:

    Amplifying what I stated in my prior comment…

    In the San Francisco Bay Area, where the coronavirus is now eclipsing the dangers of drought and fire, the new pandemic is the starkest reminder yet of how connected we all are. To tackle both the virus and climate change, there is no option but global collaboration.

    [My bold.]

    In an Era of Pandemics and Fires, Global Action Is the Only Hope

    https://e360.yale.edu/features/in-an-era-of-pandemics-and-fires-global-action-is-the-only-hope

  129. Joshua says:

    Aforementioned Don Milton is a co-author…

    > We identified seasonal human coronaviruses, influenza viruses and rhinoviruses in exhaled breath and coughs of children and adults with acute respiratory illness. Surgical face masks significantly reduced detection of influenza virus RNA in respiratory droplets and coronavirus RNA in aerosols, with a trend toward reduced detection of coronavirus RNA in respiratory droplets. Our results indicate that surgical face masks could prevent transmission of human coronaviruses and influenza viruses from symptomatic individuals.

    https://www.nature.com/articles/s41591-020-0843-2

  130. Joshua says:

    JH –

    No problem. At least in part, I come here to learn from smart people. Sometimes the best way for me to learn is to through stuff out there and get feedback.

  131. JCH says:

    There are so many papers in “preprint” right now, I doubt that many actually are ever reviewed. That’s not the intent here. There is a political movement to end the lockdown. They have soldiers. Some of them are at Stanford. Some are FOX. One is the President. They went fishing for a group of people who likely thought they were in close contact, and were curious.

    At the most, it is a study of Santa Clara County, which never really needed its own study. It’s ironic that another team at Stanford announces they have produced a Gold Standard. That is far more significant. I suspect all sorts of shocking surprises are in the future of Covid-19 antibody testing.

    One, every American who has traveled in SE Asia since Vietnam has probably brought back Bat CoV and has antibodies that could mess with the test.

    The Chinese are testing 11,000 people in Wuhan. What will that mean? You’ll learn something about Wuhan. My bass player is from Shenyang. They had 149 cases and 2 deaths in millions of people. That’s less than a pork plant in South Dakota.

  132. John Hartz says:

    Joshua:

    “through” or “throw”? 🙂

  133. Joshua says:

    Throw…

  134. David B Benson says:

    https://medicalxpress.com/news/2020-04-team-evolving-strains-sars-cov-.html

    Sequencing the single strand RNA. No wonder coronavirus kinds evolve rapidly; no repair.

  135. Steven Mosher says:

    “Andrew Gelman’s take on Santa Clara:
    https://statmodeling.stat.columbia.edu/2020/04/19/fatal-flaws-in-stanford-study-of-coronavirus-prevalence/

    Yep.

    Glad to see Gelman pick up on the Age thing as well.
    One other thing that bothered me is their Zip code map doesn’t appear to match any known
    map I have looked at for the county.

  136. Steven Mosher says:

    “Steven – So you’d fancy your chances strolling PPE free along Jacksonville beach at this precise juncture?

    Will you be “watch[ing] Florida infection figures in 2-3 weeks time”?

    As opposed to staying inside when my SO returns from testing positive?
    yes.

    Here

    https://www.propublica.org/article/coronavirus-advice-from-abroad-7-lessons-americas-governors-should-not-ignore-as-they-reopen-their-economies?utm_source=social&utm_medium=facebook&utm_campaign=publishdelay#944302

    Here is the thing

    There is no science on this.
    There will never be science on this
    There is only what has shown to work.
    Vast vast majority of transfer is person to person. Family.

    Now go check how many states in the NE have open beaches.
    Hint the press isn’t covering it.

  137. Joshua says:

    Ben –

    Re: that Gelman article – ouch!

    Thanks for that link. I can’t evaluate the statistical critique, although I know that Gelman is considered a stat heavyweight…

    … but what I don’t get at the most basic level is why professional epidemiologists would allow their work with a non-random sampling of an outlier sample to be used widely in the public to inform debate about infection and mortality rates at the national level.

    And with the Diamond Princess analysis, Ioannidis actually participated directly in the debate.

    I have to assume there’s something I don’t understand, but from what I can understand it seems supremely irresponsible.

  138. Steven Mosher says:

    “It is my understanding that the Santa Clara study is currently under peer review — by an expert panel, not by the people posting on on this comment thread. Is this correct?

    Post normal science. extended peer review

  139. Steven Mosher says:

    “In the San Francisco Bay Area, where the coronavirus is now eclipsing the dangers of drought and fire, the new pandemic is the starkest reminder yet of how connected we all are. To tackle both the virus and climate change, there is no option but global collaboration.”

    everyone will find a hobby horse to ride in a PNS situation.

    Infection is local
    testing capacity is local, unless you are Finland overnighting samples to seoul
    isolation of the positive is local, unless you are in the USA ignoring the lessons learned
    in China, Korea, Taiwan, Hong Kong, and singapore.
    Hospital capacity is local
    Age structure is local
    Ventilators can be shipped.

  140. Joshua says:

    > testing capacity is local

    How banal can you get?

    Everything is local. Except if the localities lack the needed resources, funding, personnel, expertise, facilities, and authority to execute the action.

    Saying that contact tracing and testing are local is great rhetoric. As a policy, in this country, it’s going to get a lot of people killed.and stifle the economy.

    In this country it has to be a cooperative “whole government” approach, combining and cooperating at all governmental levels.

    Listen to the White House task force. How often do you hear them talking about a “whole government” approach – right up to the point where they are being asked about accountability for the failure to execute things like contact tracing and testing?

  141. Joshua says:

    https://www.bloomberg.com/news/articles/2020-04-18/seoul-s-full-cafes-apple-store-lines-show-mass-testing-success

    > walk at the beach? nothing wrong with that.

    Swedes travel to parks is up 84%
    Korean’s enjoying the park up 24%

    ++++++++

    Talking about Koreans going to parks has bears no relevance to Americans going to the beach in Jacksonville.

    In Korea they have a functional national government which is effectively executing testing and contact tracing.

    In Korea they don’t have “freedom fighters” running around in flack jackets and carrying assault rifles, whining about “tyranny” when their government enacts policies to fight a deadly epidemic.

    It’s not a good comparison.

  142. JCH says:

    These are Trump’s Thugatarians: armed to the teeth:

    This a hero standing against a caravan of Trump’s thugs:

    We’ve seen this before:

    Also, an infectious disease doctor on the radio right now just said a lot of Americans have antibodies to a lot of coronaviruses that could be, until there is a Gold Standard test, fooling COVID-19 antibody tests.

  143. Jim Hunt says:

    Steven,

    My own focus is on South West England beaches, not those of the North Eastern United States. Over here it has recently been decreed that driving a “reasonable distance” to go surfing as your daily single lockdown busting exercise is not beyond the pale. Fortunately in all the circumstances the beaches of North Cornwall are much more sparsely populated than Jacksonville’s appear to be!

    Attempting to bring this information to the attention of the denizens of the local vicinity my helpful information has recently been deleted from both the Camelford Police Facebook page and Twitter. Hence here is some more “advice from abroad” for you:

    http://Davidstow.info/2020/04/what-constitutes-a-reasonable-excuse/

    Can we perhaps agree that “Covid-19 science” is an extremely long way from “settled” at this juncture?!

  144. John Hartz says:

    Joshua:

    How banal can you get?

    Don’t sweat the small stuff, it’s just Mosher being Mosher.

  145. Joshua says:

    JH –

    Fair enough. I just heard Cuomo talking about the primacy of the state government responsibility in both testing and contact tracing.

  146. Joshua says:

    JH –

    Bear with me:

    Ioannidis talking about the study:

  147. Joshua says:

    IMO, Ioannidis makes some strong arguments – but I still don’t get how he thinks it’s valid to extrapolate nationally (or globally) from Santa Clara data (without controlling for the factors by which those data are not representative)..

  148. Joshua says:

    re: the Ioannidis video.

    In the name of not staying in my lane…

    I have been arguing all along that the mortality data should not be aggregated across age brackets. Although it’s kind of amazing that even as Ioannidis talks about the problems of doing so, he nonetheless does so himself with extrapolating from the Santa Clara data.

    As for his arguments about the impact of mandated social distancing orders – he treats the various outcome pathways as being a binary forking path. In other words, he completely ignores that many of the economic and psychological harms taking place under “lockdowns” will simply not disappear if the economy is opened contemporaneous with high rates of infection. The harmful outcomes are not only a function of the social distancing orders, but from the concerns about being infected irrespective of whether such orders are in place.

    I’ll also note that he talks about the uncertainty resulting from the “dying with” vs. “dying from” Covid-19 issue.

    Ok, a valid issue.

    However, he fails to discuss the uncertainty about number of deaths given that people are dying without being tested (e.g., dying at home).

    Why does he only talk about the uncertainty on the one side of the issue?

    Seems pretty weird to me.

    He also only talks about deaths as an outcome. He doesn’t discuss much the impact of serious illness, ICU admissions, hospitalizations, etc.

    Seems to me like he’s discussing the data in a manner so as to defend a position rather than address all the uncertainties in all directions.

    And with that, I will stop playing armchair epidemiologist.

  149. Everett F Sargent says:

    Joshua,
    Why are you posting UBoob propaganda?

    “The opinions that have been expressed by Knut Wittkowski, discouraging social distancing in order to hasten the development of herd immunity to the novel coronavirus, do not represent the views of The Rockefeller University, its leadership, or its faculty.

    Wittkowski was previously employed by Rockefeller as a biostatistician. He has never held the title of professor at Rockefeller.”
    https://www.rockefeller.edu/news/27872-rockefeller-university-releases-statement-concerning-dr-knut-wittkowski/
    Perspectives on the Pandemic | Professor Knut Wittkowski | Episode 2 (sorry but no UBoob link will be posted by yours truly)

    Journeyman Pictures = propaganda

    Same goes for Katz and Dr. Bad Combover above …
    “Although Katz has Yale University as his Twitter location, Katz is not academically affiliated with Yale and has not held an academic appointment there since 2016.”
    https://en.wikipedia.org/wiki/David_L._Katz
    https://yaledailynews.com/blog/2020/03/30/former-professors-op-ed-controversial-among-health-experts/

    You do need to stop arguing with nutbar deniers IMHO. Please do so immediately if not sooner.

  150. Everett F Sargent says:

    Trumpkin the Idiotarian.

    I have ZERO faith in our current government, particularly now with COVID-19, I have less then ZERO faith …

    If it can be screwed up any more currently then Trumpkin will find a way to do so. POTUS now stands for Propagandist of the United States. Next up? Small Hands waving dildos on teevee.

  151. Joshua says:

    Everett –

    > You do need to stop arguing with nutbar deniers IMHO. Please do so immediately if not sooner.

    Why?

    Maybe it makes a difference to my mental health. I don’t see why it matter much beyond that.

  152. Steven Mosher says:

    “In Korea they don’t have “freedom fighters” running around in flack jackets and carrying assault rifles, whining about “tyranny” when their government enacts policies to fight a deadly epidemic.”

    They have freedom fighters and protestors. They don’t carry guns. And yes they whine about tyranny. and they physically fight with the police.

    Recent clash.

    jeez, did you check before you made the claim that we don’t have freedom fighters?

  153. Steven Mosher says:

    “Talking about Koreans going to parks has bears no relevance to Americans going to the beach in Jacksonville.”

    1. the only data you have about transmission will show you stuff like this

    lets use the latest analysis of a complete contact tracing.

    54% of cases are family to family
    33 % were friends
    24% were co workers
    8% were from a common place visited.

    Of course Connecticut beaches are open as well.

    My point would be this. There won’t ever be science that settles the question.
    There will never be zero risk.
    All the actual evidence we have points to some elevated risk from prolonged contact
    in close quarters.

    Don’t take mass transit to the beach. and keep your hands to yourself

  154. Joshua says:

    > lets use the latest analysis of a complete contact tracing

    1) Contact tracing of people who were, at least to some extent, practicing social distancing? So yeah, they didn’t transmit in public because they weren’t much out in public. Doesn’t translate to hitting the Beach in Jacksonville.

    2) So someone spreads it to 9 a family members they spend a lot of time with. Where did they get it? One transmission from out in public. So you have 10% public transmission leading to 90% family transmission.

    That is the problem with opening up the economy with inadequate testing and contact tracing. You have a bunch of infected people transmitting to uninfected people in public spaces, who then transmit to vulnerable people who can’t perfectly isolate because it’s hard or just impossible.

    If you could perfectly isolate vulnerable people you’d have a ton of morbidity, at a huge cost, but not a lot of mortality. In the real world (in the US) we’ll get both.

  155. Joshua says:

    > There will never be zero risk.
    All the actual evidence we have points to some elevated risk from prolonged contact
    in close quarters.

    I agree with all of that. But the point is that to prevent exponential growth and expanding risk, you minimize a lot of little risks.

    If I analyze each little risk they all seem too little to be bothered. There is so little risk from touching that one doorknob, why bother worrying about it. The problem is that all those tiny risks add up. By uniformly minimizing as many as I can (I can’t be perfect) I significantly reduce my risk. Now multiply that by everyone following the same process

  156. Steven Mosher says:

    “air enough. I just heard Cuomo talking about the primacy of the state government responsibility in both testing and contact tracing.”

    then you also heard him about opening NY on a regional basis.
    Not county level, but by region.
    why? well because he is learning about clusters and patterns of transmission.

    He still hasn’t asked Korea or Singapore how many contact tracers you need per case,

    anyway Mass did

    Funny thing is they are hiring people to make phone calls, when Singapore just uses the
    military. already on the payroll of course.

    I remain unconvinced of Trump’s “all government” response. Unless it means
    block grants to state governments and removing roadblocks to local innovation.

  157. JCH says:

    This lockdown protest sign wins. Game over:

  158. Joshua says:

    > then you also heard him about opening NY on a regional basis.
    Not county level, but by region.
    why? well because he is learning about clusters and patterns of transmission.

    He mentioned regions, but he also talks frequently of the problems with a patchwork approach and talks of coordinating major policies across multiple states because of what will happen if he has one policy in NYC and another in the Hudson Valley, say (where I live).

  159. Joshua says:

    Given the resistance in this country to electronic or cell phone surveillance, seems to me we’ll need armies of people and for that we’ll need at least money from the federal level. I’d love to see a Covid-19 WPA type program, a national health core. A health services infrastructure plan, putting people to work in good paying jobs to address the hole left by the reduction in public health services over the last 15 years or so.

    Yeah, that’ll happen. Trump’s gonna get right on that one.

    Well, doesn’t matter anyway, red states are opening. We’ll see if they will get it together to do the contact tracing and tasting. Needless to say, I have my doubts.

  160. Steven Mosher says:

    “In Korea they have a functional national government which is effectively executing testing and contact tracing.”

    yes, and what that tracing tells you is beaches are not a problem.

    Families : Problem
    Friends : problem
    Co workers ( working in tight spaces–call centers ) Problem
    Gym ( working out touching common equipment ) Problem
    Churches ( sitting for hours in the same confined place) Big problem
    Nursing homes ( visitors and staff infecting residents ) Big problem
    Hospitals & mental institutions ( Visitors and Staff infecting patients ) Big problem
    Internet cafes ( closed space, sharing common computers) Problem
    Government Offices ( closed spaces, shared public surfaces ) problem
    restaurants, nightclubs, wine bars, etc — 4 cases or so
    beaches, parks, shopping malls– no known problem

    Of course you can stay inside and watch TV reports of people on the beach in Florida
    with your family… data suggests that’s more risky.

  161. Joshua says:

    In my town (which is very rural) I can’t even get information on whether there’s been community spread and if so, where it might have taken place. All I can do is gets counts of how many people have been infected – which is basically useless.

    It’s absolutely insane.

  162. Joshua says:

    > yes, and what that tracing tells you is beaches are not a problem.

    That’s because they have contact tracing.

    Sheece. It’s not that difficult a concept.

  163. Joshua says:

    In Korea they don’t have a ton of infectious people waking around in public.

    Here, because we don’t have virtually ANY contract tracing (and testing, and coordinated isolation of people who test positive) we will.

    They effectively gave up on testing and contact tracing weeks ago. Unless you had serious symptoms you couldn’t even get a test. If you were sick they just said go hime sbd stay away from people. No contact tracing. No fever clinics.

    The situation here is not comparable to Korea.

  164. Joshua says:

    Compare the % of the public who are infectious here, and the % who are infectious in Korea. And that’s with a much smaller % who have been treated. Wait until we catch up in % tested.

    Given those differences, policy imitations are vastly different.

  165. Joshua says:

    Smaller % who have been tested, that should say…

  166. Joshua says:

    Compare the % positive tests here to the % of positive tests in Korea.

    When we reach that level, sure, open up the beaches. Now? Nope. The places where they’re talking about opening, they haven’t done enough testing to get a meaningful look at the % positives.

    If they do that testing and it’s low like in Korea, then maybe public spaces like beaches and parks would be OK. Without that info, it’s flying blind through heavy anti-aircraft fire.

  167. Joshua says:

    That all said, now there’s an LA County study with results similar to the Santa Clara study.

    If that all pans out, and immunity really works hsppens, there’s more justification for opening public spaces.

  168. JCH says:

    I suspect you have to be packed pretty tight, and for a long time, to get the disease from a genuine asymptomatic carrier.

    Airport —– deaths
    Logan –
    MA —– 1,809
    CT —– 1,331
    JFK — 18,929
    Newark –
    NJ —– 4,377
    PA ——1,348
    Dulles –
    VA —– 300
    MD —- 582
    DC —- 2817
    Atlanta
    GA —– 775
    Miami –
    FL —– 823
    Detroit –
    MI —— 2468
    OH —— 509
    IN ——- 569
    O’Hare
    IL —— 1349
    WI —– 230
    DFW
    TX —– 505
    OK —–143
    LAX/SFisco
    CA —– 1226
    Seattle
    WA —— 652
    OR —— 75
    ——————-
    40,817 deaths in the service areas of the airports at which the Americans returning from Europe had to land. Or, 96% of 42,517. So it’s likely at least 80 to 90% of all US deaths are related to those 12 airports or Mardi Gras – 1328 deaths.
    ==================================

  169. Joshua says:

    Oops. Actually, we have caught up (and passed) in tests per million.

  170. Everett F Sargent says:

    Joshua,
    ” … LA County study ,,, ”
    Link?

    JCH,

    That’s a bridge too far IMHO, OR should be Portland, pretty much every state has at least one international airport, every state has an airport, 50 state airports explains 100% deaths, trivial correlation.

    Try 12 Monkeys. One passenger. Many vials and flights. Insurance. 😉

  171. JCH says:

    Sorry, wrote down the number for DC wrong.

    The rules were they had to return to 13 International airports if they flew from Europe. They could not fly to Portland unless they connected through one of the 13, and it had to be a domestic flight. They had to process through one of 13. One was Honolulu, and I doubt it saw many passengers, but the other 12 are all hotspots.

    If they flew from the UK, they could use any international terminal.

    Mosher is right. It takes a bit to transmit this thing. Close and personal. Mardi Gras and those 12 airports are basically the same scene: jam packed for hours. On March 14 there were 2770 cases in the US. 14 days later there were ~24,000, and they were concentrated in the cities and regions serviced by those 12 airports.

    Airport —– deaths
    Logan –
    MA —– 1,809
    CT —– 1,331
    JFK — 18,929
    Newark –
    NJ —– 4,377
    PA ——1,348
    Dulles –
    VA —– 300
    MD —- 582
    DC —- 2817
    Atlanta
    GA —– 775
    Miami –
    FL —– 823
    Detroit –
    MI —— 2468
    OH —— 509
    IN ——- 569
    O’Hare
    IL —— 1349
    WI —– 230
    DFW
    TX —– 505
    OK —–143
    LAX/SFisco
    CA —– 1226
    Seattle
    WA —— 652
    OR —— 75
    ——————-
    38105 deaths in the service areas of the airports at which the Americans returning from Europe had to land. Or, 89% of 42,517. So it’s likely at least 80 to 90% of all US deaths are related to those 12 airports or Mardi Gras – 1328 deaths.
    ==================================

  172. David B Benson says:

    JCH — I am persuaded. Please properly write this up for journal publication. And quickly submit it.

  173. Everett F Sargent says:

    JCH,

    Louis Armstrong New Orleans International Airport
    Find cheap flights from London to New Orleans
    https://www.kayak.com/flight-routes/London-LON/New-Orleans-Louis-Armstrong-MSY

    Like I sort of did not say, they all came on a plane, the only other possibility would be as crew from a ship (at least a week to weeks of delay). And, I seriously doubt that MG was the very last large gathering in the US. Mardi Gras 2020 was on Tuesday, February 25, 2020 …
    https://www.louisianatravel.com/articles/new-orleans-mardi-gras-2020
    https://bleacherreport.com/articles/2880569-timeline-of-coronavirus-impact-on-sports

  174. Joshua says:

    Steven –

    Does a megachurch group trying to worship really count as freedom fighters with guns out in a political protest about the tyranny of social distancing?

  175. izen says:

    @-JCH
    The airports as a source of ‘patient zero’ in any particular area is a seductive narrative.
    The subsequent regional deaths may reflect the population density and the social structure affecting the spread/death rate, rather than the airport being a key site of spread.
    Otherwise you have to explain the large disparity between numbers for Dulles VA and DC. Or Seattle WA — OR .

  176. Steven Mosher says:

    wow

    Bill de Blasio gets it. NYC self sufficient in key medical equipment, locally designed and built.

    The word is decentralized… next we just have to get him to open source/ copy left
    the design

  177. JCH says:

    Not really. DFW was at a very low infection level when the flights occurred. First patient was March 9. So why did DFW not take off until after the March 12-14 flights. Washington was already mitigating. It only makes sense that fliers tried to stay in the east if they were from the east, so DFW, LAX, Sea, SFisco would have seen a lower number of travelers.

    A lot of Houstonians go to Mardi Gras. It’s not far.

    In Boston they put 404 homeless people into a shelter. Bam, 140 test positive, no symptoms.

    What you have to explain is why all the other popular international airports did not really become hotspots: Kansas City, St Louis, Minneapolis St Paul, Tampa, Charlotte, Nashville, Cincinnati, Phoenix, Pittsburgh, Cleveland, Milwaukee, Memphis, Raleigh, Orlando, Buffalo, Austin, etc. They were getting European travelers Jan, Feb, Mar through 11th.

    And then there is Honolulu, which gets a huge number of travelers from Asia, and has a very low infection number. How many did they get from Europe on March 12-14? Not many. Trump banned travel to the wrong place. The Chinese had already cut off Hubei. His worry was blue-eyed blondes. How ironic.

    And then there are the two studies so far that conclude the variant of SARS-CoV-2 in NYC and San Francisco came from Europe. It was coming in before they forced everybody back, but you have the hyper crowded incubators on March 12-14.

    Sioux Falls, very small numbers. I’ve been in the John Morrell plant many times. Now Smithfield. It’s crowded as heck. Bam, Sioux Falls, area population 225,000, has more confirmed cases than Minneapolis St. Paul, which has a pretty busy international airport that did not receive passengers from Europe Mar 12-14.

    Beach is relatively safe. Anti-lockdown rallies? Not so much.

  178. Jim Hunt says:

    Steven – Bill for @POTUS?!

    Meanwhile here’s the “Community Shield” concept, hot off the presses of the UK Green Party, for your consideration:

    http://CoV-eHealth.org/2020/04/21/the-green-partys-community-shield/

    Your thoughts?

  179. izen says:

    @-Jim Hunt
    “Meanwhile here’s the “Community Shield” concept…”

    Well, its a concept.
    Putting it into some sort of practical system is another matter.

    I am in contact with people (in the UK) who have been trying to arrange some sort of healthcare/treatment for those with other conditions than CORVID19 that arise for whom it would not be appropriate to attend a hospital or other primary care facility. They have had problems.
    The local health authority has refused any help with facilities or equipment because they say they have insufficient for their own needs.
    When the clinicians involved offered to provide their own equipment and PPE they were told that if they did it would be requisitioned by the NHS…
    It is difficult to see an enthusiastic embrace of an idea that is anything other than the current policy of reducing the spread by maximising lockdown and restricting testing and treatment to those who are seriously ill. The onus of responsibility is then on the individual to obey the social distancing rules, and the limited, and largely ineffective, treatment options are rationed. Both the UK and the US seem to have dismissed testing and contact tracing as a priority. Although there may be a change in this as things develop.

    What might work at the community level along with test, isolate and contact trace is strict local quarantine. Road-blocks prevent travel between communities of village to small town size. Larger population centres are broken down into localities.
    You need food and healthcare deliveries to those isolated communities, but they are the only people who are allowed to travel between the enclaves.
    The if there are no new cases in an enclave for a couple of weeks, the lockdown in that area can be relaxed, but travel between it and other localities that are still not confirmed clear is still banned.

    The US may get away with easing the lockdown to the delight of the Trumpists if they follow a travel ban between and within urban centres.
    The problem with this is that much of our society is built on non-locality. The days when most people lived in walking or bicycle distance of work and shops are long gone.
    It is going to be a while before four million people are crowding onto trains for an hour or more to work in London from their leafy suburbs every day.

  180. Joshua says:

    In case anyone is interested…

    A discussion of the potential for self-selection bias in the Santa Clara study recruitment:

    https://analytica.com/adjusting-the-santa-clara-county-antibody-prevalence-results-for-self-selection-bias/

    The LA Country recruitment was a bit better than just running Facebook ads,

    > A random sample of these residents were invited to participate in the study on April 4 with the goal of recruiting 1,000 participants for testing. Quotas for enrollment in the study for population sub groups were set based on age, gender, race, and ethnicity distribution of Los Angeles County residents. Participation in the study was restricted to one adult per household.

    ++++++++++

    Still, seems to me there would be a pretty good chance of self-selection bias.

  181. Jim Hunt says:

    Interesting @izen,

    How to turn a “concept” into “reality”?

    Are you able to hook me up with the “people (in the UK) who have been trying to arrange some sort of healthcare/treatment for those with other conditions than COVID19” you mention?

    Perhaps via a DM to: https://twitter.com/DavidstowInfo/

  182. izen says:

    @-Jim Hunt
    I dont think they would want to be involved in any further interactions on this matter.
    Being told they were REQUIRED to volunteer for redeployment within the NHS reduced their enthusiasm for the whole system.

    But if you want to see how inconsistent and unorganised a localised response can be, compare the provision of urgent dental care, which I see has been established in your area –
    https://www.cornwalllive.com/news/cornwall-news/new-emergency-dental-care-centre-4050850

    With many (most?) other local health authority regions that still have little or no provision despite somewhat larger populations to serve than Cornwall.

  183. Ben McMillan says:

    Looks like about 20% of NYC residents have had COVID.
    https://www.nbcnewyork.com/news/local/new-york-virus-deaths-top-15k-cuomo-expected-to-detail-plan-to-fight-nursing-home-outbreaks/2386556/

    So if this is right, still not enough herd immunity to prevent round 2. Also, IFR more like 0.5% (counting only the ‘official’ COVID deaths).

  184. Steven Mosher says:

    “Does a megachurch group trying to worship really count as freedom fighters with guns out in a political protest about the tyranny of social distancing?”

    ya they didn’t have guns, so they physically fought the police
    Psst that wasn’t the mega church, why did you assume it was?
    that was a regular old protestants.

  185. John Hartz says:

    Utilitarianism vs Ethics…

    A quandary now confronts policymakers across the country (USA): Should we reopen the economy to help the majority or protect the lives of the vulnerable by remaining in lockdown?

    One answer to that question — letting a minority suffer so that the majority may benefit — is known as utilitarianism. And it’s one of the most common, and controversial, ways of making moral choices.

    In matters of life and death, like we’re in now, strict utilitarianism can be downright dangerous.
    “This is a moment when every country in the world is facing the same set of ethical questions and dilemmas,” said Anita Allen, a member of President Obama’s bioethics committee.

    How we answer will be a real test of our humanity and sense of justice.”

    The dangerous morality behind the ‘Open it Up’ movement, Analysis by Daniel Burke, CNN Religion Editor, CNN, Apr 23, 2020

    https://www.cnn.com/2020/04/23/us/reopening-country-coronavirus-utilitarianism/index.html

  186. Steven Mosher says:

    “When we reach that level, sure, open up the beaches. Now? Nope. The places where they’re talking about opening, they haven’t done enough testing to get a meaningful look at the % positives”

    sunlight apparently kills the critter. see the latest DHS science.
    so it could be that keeping folks indoors does two bad things.
    1. if they are vitamin d deficient it can lower the effectiveness of their immune system
    2. it keep them in an environment where the virus lives longer on surfaces and in
    the air.

    opps. ya killed people who would not have died otherwise.

    1. Looking at contact tracing in Korea you see ZERO cases tied to X was walking down
    the street or at the park or at the outdoor cafe. zero.
    2. Looking at the contact tracing you see it all tied to indoors close contact scenarios.
    3. 80% of the cases come from 20% of the patients, super spreaders.
    4. Looking at the SSE for Covid , you find it dominated by shit like
    a) drunken parties, Church groups, face to face business networking
    b) Zero SSE walking on the beach

    here is another lesson

    Singapore had this under control. then they got a few positive cases in their migrant worker
    demographic. What did they do? they sent them back to their dormatories to wait

    What happened next?

    boom. take a sick person and quarantine them with well people and boom.
    New York is doing the same thing by telling people who test positive to go home
    and wait. Ask Chris Cuomos wife how that worked.
    China learned this quickly. What did they do? They took the gyms
    that had been closed ( school gyms) and turned them into “isolation wards”
    Not really that bad since the people are not sick enough to go to the hospital.
    How did they learn it? They sent a patient home to be with his wife and kid.
    later when they went to check on him they found the baby alive and both
    parents dead. Kinda a hard lesson to learn.
    Anyway, see that case load in NY and the slow decline from peak?
    That’s what you get when you don’t isolate the positive.
    There you go. you have health officials who KNOW sending positives to be with their family
    is one of the more risky things you can do, but they do it. meanwhile, while there
    is no evidence of transmission at the beach they exercise an over abundance of caution
    its the weirdest use of uncertainty.

    Look, we will never have a complete detailing of everything we need to know to ensure
    100% safety. The science will always be unsettled, always be uncertain, and always
    have holes of ignorance. But what we do know is this: This thing progresses via
    clusters. case, case, case, cluster, cluster, boom. And so smart countries work by
    focusing on reducing clusters. Korea finds a
    cluster in Gyms. Bam, gyms are closed. They find a cluster in Churches. bam
    They are closed. They find a cluster in tightly packed call centers, BAM every tightly
    packed business gets a notice. Parks, beaches? No evidence of transmission in those
    areas. That’s why China and Korea both allow people to go to the park and beach,
    But, they insist on masks. And in China you can’t set up a tent at the beach or in the park.
    of course in Bejing some stupid foreigners did this, milled around having a party
    at the park without masks. They were dealt with.

  187. Steven Mosher says:

    > yes, and what that tracing tells you is beaches are not a problem.

    That’s because they have contact tracing.

    Sheece. It’s not that difficult a concept.

    ###############

    The goal is reduce transmission, not eliminate it. because you can’t eliminate it.
    The goal is to reduce transmission by eliminating clusters.
    So you identify the situations of cluster transmission: indoor face to face gatherings.
    That’s why china changed rules for restaurants. Cluster in Guandong.
    big dinner party, family style serving.
    So new rules for diners. No face to face seating, no shared chopsticks, no shared
    serving spoons, reduced occupancy limits.
    So new rules for mass transit; 50 % occupancy. Reduce transmission, not
    eliminate it.
    So new rules for work. 50% occupancy. Hint we were back at work early in Feb.
    In Korea they learned that quarantining foreigners at home didn’t work.
    Now you report to a “government facility”. Hotels that don’t have customers anyway.

    reduce, because elimination is impossible in the short run.

    So you can do what works or not.

  188. Steven Mosher says:

    When a civil defense guy did this in Korea they played it repeatedly on the nightly news
    and he was relieved of his public facing duties

    https://hannity.com/media-room/fancy-nancy-pelosi-rubs-runny-nose-smears-bare-hands-all-over-house-podium-while-discussing-coronavirus/?utm_source=socialflow

  189. Steven Mosher says:

    since there is no guidance people make shit up

    https://www.msn.com/en-us/news/us/how-soon-can-businesses-reopen-during-pandemic-a-texas-suburb-takes-the-plunge/ar-BB1371hS

    weirdly both here and in Georgia they wanted to open gyms and massage parlors.

    China has posted rules for restaurants. why folks won’t look at them is beyond me

  190. Steven Mosher says:

    “Beach is relatively safe. Anti-lockdown rallies? Not so much.”

    yup

  191. Joshua says:

    > sunlight apparently kills the critter. see the latest DHS science.
    so it could be that keeping folks indoors does two bad things.
    1. if they are vitamin d deficient it can lower the effectiveness of their immune system
    2. it keep them in an environment where the virus lives longer on surfaces and in
    the air.

    Oh. Ok. I didn’t realize that you have to go to the beach with a bunch of other people to get sunlight Sorry. My bad

    > 1. Looking at contact tracing in Korea you see ZERO cases tied to X was walking down
    the street or at the park or at the outdoor cafe. zero.

    They were wearing masks. They were following social distancing instructions.

    And you’re going to limit those people to walking down the street, in the park, at the outdoor cafe? Yeah, that’s going to happen.

    And again, you just really don’t get it, do you? You’re going to have a shit ton of infected people walking down the street. This isn’t Korea. We have done contact tracing, basically, at all. We haven’t isolated people. We didn’t test until a shit ton of people infected.

    Just keep repeating to yourself and you’ll get it: The US isn’t Korea. The US isn’t Korea. The US isn’t Korea.

    > China learned this quickly. What did they do? They took the gyms
    that had been closed ( school gyms) and turned them into “isolation wards”
    Not really that bad since the people are not sick enough to go to the hospital.
    How did they learn it? They sent a patient home to be with his wife and kid.
    later when they went to check on him they found the baby alive and both
    parents dead. Kinda a hard lesson to learn.

    Yeah. That would be great. Except it isn’t going to happen here, nothing even remotely like it at least for quite a long time.

    Meanwhile, we’re going to have a shit ton of infected people walking around. Here, try this:

    The US isn’t China. The US isn’t China. The US isn’t China.

    > Anyway, see that case load in NY and the slow decline from peak?
    That’s what you get when you don’t isolate the positive.

    Right.

    > There you go. you have health officials who KNOW sending positives to be with their family
    is one of the more risky things you can do, but they do it. meanwhile, while there
    is no evidence of transmission at the beach they exercise an over abundance of caution
    its the weirdest use of uncertainty.

    You’re going to have a ton of infected people walking around. Here, try this:

    You’re going to have a shit ton of infected people walking around. You’re going to have a shit tone of infected people walking around. You’re going to have a shit ton of infected people walking around.

    > Look, we will never have a complete detailing of everything we need to know to ensure
    100% safety.

    Not a terribly deep insight but undoubtedly true!

    > The science will always be unsettled, always be uncertain, and always
    have holes of ignorance. But what we do know is this: This thing progresses via
    clusters. case, case, case, cluster, cluster, boom. And so smart countries work by
    focusing on reducing clusters. Korea finds a
    cluster in Gyms. Bam, gyms are closed. They find a cluster in Churches. bam
    They are closed. They find a cluster in tightly packed call centers, BAM every tightly
    packed business gets a notice. Parks, beaches? No evidence of transmission in those
    areas. That’s why China and Korea both allow people to go to the park and beach,
    But, they insist on masks. And in China you can’t set up a tent at the beach or in the park.
    of course in Bejing some stupid foreigners did this, milled around having a party
    at the park without masks. They were dealt with.

    That would all be great!!!!!!!! But how realistic is all of that happening here anywhere in the near future? We completely lack the infrastructure to get it done.

    Here, look at this:

    20k contact tracers in CA alone. Let’s build an army. Cell phone surveillance would be nice…but…This isn’t Korea. This isn’t Korea. This isn’t Korea.

  192. Joshua says:

    > The goal is reduce transmission, not eliminate it. because you can’t eliminate it.

    If you don’t have the infrastructure to test and contact trace, how the hell are you going to reduce it?

    Limit public interaction while you build the infrastructure to test and trace. *Then* you open stuff up. Don’t freakin’ do it before you have the infrascture.

    > The goal is to reduce transmission by eliminating clusters.

    There is more than one goal, especially when you have no working infrastructure to test and trace and a bunch of freedom fighters running around and calling it a hoax.

    > So you can do what works or not.

    This is not a binary situation. Stop with the binary thinking. You do what works, in a limited fashion, with downsides that go along with it, until you build the infrastructure to do something better.

    You don’t just stick your thumb up your ass and have a shit ton of infected people walking around while you lack the infrastructure to do what works better.

  193. Joshua says:

    > weirdly both here and in Georgia they wanted to open gyms and massage parlors.

    Scott Gotlieb – who I think is quite good:

    > “Gyms, nail salons, bowling alleys, hair salons, tattoo parlors — it feels like they collected, you know, a list of the businesses that were most risky and decided to open those first,” Scott Gottlieb, former Food and Drug Administration commissioner, told CNBC on Tuesday.

    Gotleib is a fellow at the American Heritage Institute. Former member of the Trump administration.

    Here’s an interview with him:

    https://www.vox.com/2020/4/14/21219021/scott-gottlieb-coronavirus-covid-19-social-distancing-economy-recession

    Here’s the Donald on your sunlight idea. Maybe you can work it out with him:

  194. Joshua says:

    She’s going to check and test hitting the body with “a very powerful light.”

    Maybe they can do something with the disinfectant, the injection inside or almost a cleaning!!!

  195. JCH says:

    It was so simple. You just have to get one of these inside. He could have been a General.

  196. anoilman says:

    JCH: UV Sanitizer. Yeah… I had to explain to my mom that she just needed to ingest intense UV and earn a 3 dimensional sun burn. I even had to follow that explanation up with;
    https://www.theguardian.com/lifeandstyle/lostinshowbiz/2019/dec/05/josh-brolin-gets-burned-where-the-sun-dont-shine

    For the record… a cold lasts 7 days outside… flu 24 hours. Nothing new here.

    Mosher: My wife works with an epidemiologist/medical doctor (he doesn’t like people, so he went into epidemiology) who wears a mask in public. He does so because he knows he touches his face less in public. (A tick for Steven Mosher?) As a trained doctor and actively working in the field, he knows that mask efficacy is questionable if its not air tight, and he is also aware of the other two new entry vectors because you wear a mask. Namely, the mask can be exposed, and later touched by your hands, or alternatively his hands may be exposed, then touch the mask rendering the mask exposed (a concern if you re-use, and for longer term use beyond a single activity).

    Either way it results in a super lot more hand washing at each and every single mask removal or risk spreading viruses. Wanna scratch now? A little to the right..

    The reason this subject has been studied so extensively and for so long is that the largest vector for transmission is… Doctor’s and Nurses. Think about it a sec.. Those folks (doctors\nurses) are like busy bees pollinating flowers (patients). They are constantly being measured and monitored for hand hygiene for this reason alone. Use of other gear (masks\gloves) has questionable results without proper training. We know this for a fact.

    Personally, I would think masks are great. My wife responds with, “Meh. What’s for dinner.”

  197. JCH says:

    This all reminded me of something. In the 1970’s my Dad leased a commercial building to an old guy who owned a big laundry. He was in his late 80’s, and he worked hard 6 days a week. All day long, and sometimes into the night. He used DSMO, an industrial solvent, like it was going out of style. Rubbed it on his skin. It was like national fad: DSMO for arthritis. Back then, people were getting it from janitors. Don’t know why, but they often had the stuff.

  198. Joshua says:

    I was listening to a podcast of some modelers talking about modeling in the context of COVID 19.

    One of the modelers said something I found quite interesting (paraphrasing):

    > If our most important goal is reducing the strain on the healthcare system, then you can often get away with lower levels of fidelity in the model.

    +++++

    https://castbox.fm/episode/Epidemiological-Modeling%2C-Policy%2C-and-Covid-19-(Daniel-Kaufman%2C-Eric-Winsberg%2C-%26-John-Symons)-id37106-id254740525

    I think that framing applies pretty well to climate modeling, and I think I agree with what the modeler said – and that what he’s saying is an important factor in the risk assessment process. We all necessarily have subjective prioritizations in how we assess the “fit for purpose” of the modeling.

    Perhaps one of the most important parts of the process is to try to make explicit what our priorities are so we can find common interests there, or negotiate different interests. The bigger problem is that people have a tendency to weaponize those discussions.

    I’d be curious to hear the thoughts of others…

  199. Joshua says:

    I guess what I’m saying is…

    People like to pretend that “fit for purpose” is some kind of objective assessment. It isn’t, because it necessarily depends on subjective assessments.

  200. Joshua says:

    Steven –

    In particular you might like the discussion of school closing at about 8 minutes in.

    They say that closing schools beyond 8 days doesn’t limit spread – but they have no explanation. Only caveat is thst there’s no data on what happens with school closures along with the other interventions.

  201. Joshua says:

    Upon finishing that podcast…

    I found the first part maybe not particularly profound but still reasonably intersting. The second part, however, I thought was particularly relevant to climate modeling even if climate change was not explicitly discussed and even if I that past was supposed to be more tightly focused on COVID 19 modeling. In that second part, they talk about the intersection of tribalism and modeling.

    FWIW, One of the participants is apparently quite involved in climate modeling (Eric Winsberg).

    Again, I’d love to read what people think if anyone takes the time to listen.

  202. Joshua says:

    Oops. That wasn’t 8 minute in….. That was 46 minutes in…. In case anyone cares….

  203. Steven Mosher says:

    “20k contact tracers in CA alone. Let’s build an army. Cell phone surveillance would be nice…but…This isn’t Korea. This isn’t Korea. This isn’t Korea.”

    zero imagination. you realize that you can easily build an app to allow people to voluntarily
    reveal the places they have been to aid in the process?
    you realize that without cell phone tracing you get over 80% of the cases resolved?

    no of course not.

  204. Steven Mosher says:

    “And again, you just really don’t get it, do you? You’re going to have a shit ton of infected people walking down the street. This isn’t Korea. We have done contact tracing, basically, at all. We haven’t isolated people. We didn’t test until a shit ton of people infected.”

    you think before the outbreak in Korea that everyone was wearing masks?
    Nope.
    on average before the super spreader event about 5% of people on the street wore masks..
    ya dont get this walking down the sidewalk.
    After Feb17th yes it increased to about 95%

    Again, look at the actual super spreader events. Close contact. face to face.
    avoid that 80/20 rule.

    Some of the silliness is finally being reversed in places like Michigan where the governor was
    keeping people who do lawn care from working. As if my gardener has a party on the front lawn.

    and yes I am aware that the USA is not China or Korea.
    ya know in China everybody loves the family style eating. We share chopsticks and serving plates
    and serving spoons. Its a cultural thing. It really bugs me. But my friends argued “this is china”
    it’s how we do things.

    You’d be amazed how intelligent people changed that behavior over night.

    Not one of them argued “hey this is china. dont ask us to change”

    I know. In the USA they just send send sick covid patients back to their nursing homes.
    It’s the USA, you can’t change.

  205. Steven Mosher says:

    “Here’s the Donald on your sunlight idea. Maybe you can work it out with him:

    Its not my sunlight idea. Its DHS study of virus half life.

    Now ascribing trumps dumb idea to me is not good faith is it Joshua?

  206. Joshua says:

    Steven –

    > Now ascribing trumps dumb idea to me is not good faith is it Joshua?

    It was intended as a joke. I actuality think that’s an intended form of good faith. But in order for it to be received in good faith you have to see it as a joke.

    Apparently you didn’t. No fault in your part of I made a bad joke. So I good faith I apologize.

  207. Joshua says:

    Here Ioannidis talks, a bit past halfway I think, about how he thinks we’re basically way too far in to make even any kind of contact tracing work.

    He then goes on to make the argument that the only thing that will work is basicsmally to do nothing.

    I don’t agree with him. I more take the middle position. We are too far in for what Korea did to work, even if we could get Americans to accept the kind of surveillance that Koreans accept, which we can’t.

    We need a mixture of column A and column B, and a shit ton of testing and a shit ton of investment in an infrastructure to make it all work. And along with that, once that is in place, but lot before, we need a carefully calibrated sequence of lifting mandated social distancing. Star lifting too soon, and imo, basically only the level of operational infrastructure is what establishes the time line along with the ratio of positive to negative testing (which also has to be seen in relation to the number of tests), and you just make the contact tracing all that much more difficult and less effective – ala Ioannidis.

    There is an inverse relationship, imo, between the level of functional public health infrastructure and the amount of need to mandate social distancing. With enough infrastructure, you need very little mandated social distancing, or perhaps even none.

  208. Steven Mosher says:

    very cool

  209. Joshua,

    FWIW, One of the participants is apparently quite involved in climate modeling (Eric Winsberg).

    Again, I’d love to read what people think if anyone takes the time to listen.

    I’ll try to find some time to listen. In case you’d forgotten, Eric and I wrote a joint post about extreme weather event attribution.

  210. paulski0 says:

    Steven Mosher,

    Parks, beaches? No evidence of transmission in those
    areas. That’s why China and Korea both allow people to go to the park and beach

    I’ve heard it’s not quite so simple. Apparently in some Korean parks authorities have been engaging in extensive re-landscaping, digging up and removing popular attractions like large floral displays in parks to reduce inflow and gatherings.

  211. Steven Mosher says:

    Never mind the science

    Study of 318 outbreaks in China found transmission occurred out-of-doors in only one, involving just 2 cases.

    https://www.medrxiv.org/content/10.1101/2020.04.04.20053058v1

    “Background: By early April 2020, the COVID-19 pandemic had infected nearly one million people and had spread to nearly all countries worldwide. It is essential to understand where and how SARS-CoV-2 is transmitted. Methods: Case reports were extracted from the local Municipal Health Commissions of 320 prefectural cities (municipalities) in China, not including Hubei province, between 4 January and 11 February 2020. We identified all outbreaks involving three or more cases and reviewed the major characteristics of the enclosed spaces in which the outbreaks were reported and associated indoor environmental issues. Results: Three hundred and eighteen outbreaks with three or more cases were identified, involving 1245 confirmed cases in 120 prefectural cities. We divided the venues in which the outbreaks occurred into six categories: homes, transport, food, entertainment, shopping, and miscellaneous. Among the identified outbreaks, 53.8% involved three cases, 26.4% involved four cases, and only 1.6% involved ten or more cases. Home outbreaks were the dominant category (254 of 318 outbreaks; 79.9%), followed by transport (108; 34.0%; note that many outbreaks involved more than one venue category). Most home outbreaks involved three to five cases. We identified only a single outbreak in an outdoor environment, which involved two cases. Conclusions: All identified outbreaks of three or more cases occurred in an indoor environment, which confirms that sharing indoor space is a major SARS-CoV-2 infection risk.”

    “Our study does not rule out outdoor transmission of the virus. However, among our 7,324
    identified cases in China with sufficient descriptions, only one outdoor outbreak involving
    two cases occurred in a village in Shangqiu, Henan. A 27-year-old man had a conversation
    outdoors with an individual who had returned from Wuhan on 25 January and had the onset
    of symptoms on 1 February”

    By all means prevent outside conversations

  212. paulski0 says:

    It makes some intuitive sense to me (perhaps wrongly) that transmission would be more difficult in open air than inside, but I’m not sure how much to read into that study. The key sentence is “between 4 January and 11 February 2020”. That is, Winter. How often do people get together outside in the middle of Winter? The difference could be explained simply by opportunity – in Winter people are inside far more than they are outside. How much applicability does that have now as we head into Summer?

  213. Everett F Sargent says:

    A Tale of Two Cities …

    The Bay area includes Santa Clara County + nine other per …
    https://khn.org/news/is-the-bay-areas-unprecedented-lockdown-the-first-of-many/
    The NYC Metro area is per …
    https://en.wikipedia.org/wiki/New_York_metropolitan_area

    It is somewhat of a no brainer that one can integrate doubling time to return to actual total accumulated deaths (although smoothed due to doubling time OLS log-normal fits of 3-, 5-, 7- and 9-day).

    The Bay area 2010 census = 6,269,434 and The NYC Metro area 2010 census = 26,468,898.

    As of COB 2020-04-24 the Bay area deaths = 158 and The NYC Metro area deaths = 27,306.

    Like it or not, as James Annan states here is TRUE …
    https://julesandjames.blogspot.com/2020/04/blueskiesresearchorguk-5-day-doubling.html
    “A small corollary of the above, added on 21 Apr: It is very straightforward to calculate the effect of a delay to the lockdown. A week of unchecked growth at 3-day doubling corresponds to a factor of 5, meaning that 80% of the total size of the first wave we are currently in could be directly attributed to the Govt delaying by a week, if it was felt that the evidence could and should have supported action that much sooner (ie, when most of the rest of Europe was taking action). That means 80% of the peak strain on the NHS, 80% of total cases and also 80% of all resulting deaths. What this calculation doesn’t account for, is what happens in the longer term. We may all get it in the longer term anyway (well 60%+ of us). But we might not, and even so, the huge peak was 5x bigger than it would have been if controlled just a week quicker.”

    Except NYC started with a doubling time of ~ONE DAY! IMHO tens of thousands of lives could have been saves if NYC lockdowns had started 2-3 weeks before their actual 2020-03-22 lockdown. The Bay area lockdown was the 1st for the US starting on 2020-03-17 but they had not seen a death and confirmed was 168 (NYC Metro had 1117 confirmed and 12 deaths on that date, JHU data) cases on the 16th.

    Oh and I tested negative for COVID-19 (hopefully just a bad cold).

    Moral of this sad story, don’t listen to idiots that spew shit about ‘you can’t do that because we need definitive date 1st’ you know who I am talking about, Lockdowns are actually a very old strategy when shit hits the proverbial fan.

    Stay home and stay safe! 🙂

  214. Everett F Sargent says:

    Joshua and SM,

    Are you two on cell phones? SM your posts are most difficult to read, too disjointed, dangling sentence structure, why not just post in Korean.

    ATTP or Willard, my post, in moderation, I used a acac word twice and four links, please post. TIA

  215. Joshua says:

    paulski0 –

    > That is, Winter. How often do people get together outside in the middle of Winter?

    In winter…. during a *lockdown” with soldiers roaming the streets. (The lockdown was imposed on January 23 – but who know how many people were roaming the streets in the midst of a raging pandemic before that?)

    And 34% occurred in transport settings. Gee, I wonder if anyone takes the bus or an Uber to get to the beach? Probably not, tight?

    Oh, and don’t forget, Steven is a big porponent of wearing masks. Hmmm. I see lots of pictures of chinese and Korean people wearing masks when outdoors. I wonder if, in fact, they’re far more likely to wear masks when outdoors than indoors? I wonder if that might contribute to that disparity in the tramission stats?

    I think maybe Steven posted that comment to convince us of the *effectiveness* of lockdowns?

    Oh, also:

    > Many existing buildings are crowded, poorly ventilated, and unhygienic. A comprehensive review of ventilation conditions in Chinese indoor environments by Ye et al. showed that the CO2 concentration can reach 3,500 ppm in some buildings. The design and operation of buildings have also been under pressure to reduce energy use and increase human productivity.

    And don’t forget that Chinese families like to sit down to eat with shared biwkscat the center of the table that they all teach into with chopsticks.

    Point being, yeah, transmission is going to be less likely outdoors. But if it occurs outdoors (whether through touching surfaces, through droplets, or through aerosols), it will contribute to exponential growth. That’s why it makes sense to limit public interaction *until we have testing and contact tracing in place.

    Oh, and maybe it’s problematic to draw conclusions about the dynamics of transmission in one environment from looking at transmission in a completely different environment?

  216. Joshua says:

    paulski0 –

    > That is, Winter. How often do people get together in the middle of Winter?

    In winter…. during a *lockdown” with soldiers roaming the streets. (The lockdown was imposed on January 23 – but who know how many people were roaming the streets in the midst of a raging pandemic before that?)

    And 34% occurred in transport settings. Gee, I wonder if anyone takes the bus or an Uber to get to the beach? Probably not, tight?

    Oh, and don’t forget, Steven is a big porponent of wearing masks. Hmmm. I see lots of pictures of chinese and Korean people wearing masks when outdoors. I wonder if, in fact, they’re far more likely to wear masks when outdoors than indoors? I wonder if that might contribute to that disparity in the tramission stats?

    I think maybe Steven posted that comment to convince us of the *effectiveness* of lockdowns?

    Oh, also:

    > Many existing buildings are crowded, poorly ventilated, and unhygienic. A comprehensive review of ventilation conditions in Chinese indoor environments by Ye et al. showed that the CO2 concentration can reach 3,500 ppm in some buildings. The design and operation of buildings have also been under pressure to reduce energy use and increase human productivity.

    And don’t forget that Chinese families like to sit down to eat with shared biwkscat the center of the table that they all teach into with chopsticks.

    Point being, yeah, transmission is going to be less likely outdoors. But if it occurs outdoors (whether through touching surfaces, through droplets, or through aerosols), it will contribute to exponential growth. That’s why it makes sense to limit public interaction *until we have testing and contact tracing in place.

    Oh, and maybe it’s problematic to draw conclusions about the dynamics of transmission in one environment from looking at transmission in a completely different environment?

  217. Joshua says:

    Everett –

    > Are you two on cell phones?

    I am. Often when only half paying attention. And I’m lazy.

    I apologize for all the typos and nonsense words.

    It’s my impresin that peple cn usuly fgre out wht I’m tryn to say anywy.

    Also, typos and nonsense words a few are only a few of the reasons why what I write is incomprehensible. 😕

  218. Steven Mosher says:

  219. Joshua says:

    Everett –

    Nice post. Thanks. I’ll put it to good use.

  220. Steven Mosher says:

    “I don’t agree with him. I more take the middle position. We are too far in for what Korea did to work, even if we could get Americans to accept the kind of surveillance that Koreans accept, which we can’t.”

    again.
    Contact tracing has two main forks.
    1. Collect the names of family
    2. Collect the names of friends
    3. Collect the names of Coworkers.
    4. Grab the visitor log from your workplace.

    NONE of this requires phone dumps.
    the people from this list will represent over 80% ( now 90%) of the people you infect.
    These people are contacted and told to show up for testing.

    The second fork is the PLACES
    1. They collect the places you visited.
    2. They send a text to all the people who live within 5 minutes of the place.
    I would typically get 2-3 texts a day. Patient 567 was in XYX train station on tues march 13th

    “Building X was closed on March 9, 2020, immediately after the outbreak was reported. We offered testing to all occupants (office workers and apartment residents) during March 9–12. We collected nasopharyngeal and oropharyngeal swab specimens from PUIs for immediate real-time reverse transcription PCR testing; the average turnaround time was 12–24 hours. Confirmed case-patients were isolated, and negative case-patients were mandated to stay quarantined for 14 days. We followed and retested all test-negative case-patients until the end of quarantine. We also investigated, tested, and monitored household contacts of all confirmed case-patients for 14 days after discovery, regardless of symptoms. During March 13–16, we sent a total of 16,628 text messages to persons who stayed >5 minutes near the building X; we tracked these persons by using cell phone location data. The messages instructed the recipients to avoid contact with others and go to the nearest COVID-19 screening center to get tested.”

    Now despite getting these texts daily I never report for testing. because I was never in building X.
    I know a lot of people make a big deal about the phone tracing, but when you look at the actual investigations 90% of it is friends, family, coworkers. the last 10% is untraceable.

    USA case

    https://www.cdc.gov/mmwr/volumes/69/wr/mm6909e1.htm?s_cid=mm6909e1_w

    So The question is not “can we do a Korean approach” probably not.
    The question is “what is good enough” and How can it be augmented easily by technology
    dumping all your contacts and location data to a tracer voluntarily, is easy.

  221. Everett F Sargent says:

    Daily multiplier for doubling time = 2^(1/doubling time (days)) = 1+daily compounding rate.
    Given a time series of doubling times, initialize N for day zero, apply daily multiplier to form a chain of daily cumulative deaths (or whatever N is). To get your dailies just subtract N-1 cumulative from N cumulative deaths.

    To get doubling time calculate local OLS slope of log-normal cumulative deaths (centered for 3-, 5-, 7- and 9-days of cumulative death data (you lose a day at the end (and beginning) for 3-day, two days at the end (and beginning) for 5-day, … ). Take log(2)/local OLS slope = doubling time. Zero slope = infinite doubling time (I only use finite series data for any integrals or substitute 1 (e. g. the number one) which just pushes the current N forwards).

    Anyways that is how it is in my spreadsheet updated once per each new day of data.

  222. JCH says:

    Went looking at papers to do with coronavirus antibodies that were written before 2019. Kind of surprising how many people have antibodies to coronaviruses of other types. Also surprising how many regions around the globe have bats that carry CoV similar to SARS-CoV-2.

    So people in a Santa Clara County have antibodies to exactly what?

  223. Joshua says:

    Steven –

    All I’m saying about that is that there will be a lot, lot more resistance to use of phone data here than in Korea. Much of it from people who have a lot of political power.

    In addition to which, we’re going to be starting much later in the process, which means a lot.

  224. Joshua says:

    Steven –

    Do you realize how little contact tracing has been going in here?

    With nearly a million identified cases and many multiples of that of unidentified cases?

  225. Everett F Sargent says:

    Joshua,

    Agree with you 111%. 😉

    We have a decentralized republic with very strong states rights (Civil War anyone?). Voting, schooling, local taxes, etceteras. We have the 2nd and 10th Amendments and a bunch of others. This would be litigated, heck it is already being litigated (e. g. lockdowns).

    We need a very strong leader in such situations, and well, we have Small Hands.

    We got us some damn real Freedom Fighters. Tea Party: Part Deux.

    We are trapped and screwed at the same time.

    But hey, we do have plenty of printing presses for printing money or some such …

  226. Everett F Sargent says:

    After a day of hard ridin’ and rollin’ at your local state capital it’s time to relax at your favorite pub …

  227. Joshua says:

    Everett –

    I think that the notion of hedging against the largest downside risk deserves some due diligence here (not unlike how I think about climate change). In other words:

    Imagine the following scenario: (1) Immunity turns out to not be a best case – either it doesn’t happen or it takes a long time to develop or it doesn’t last very long. (2) in 2 months, someone develops an effective therapeutic that significantly reduces fatality rate.

    Considering that, how much should be bet on the differential economic benefit, and health outcome benefit from things like being less cooped up, from policies based on the idea of potentially letting the virus infect people more quickly and more widely, so as to keep a hobbled economy and hobbled national psyche limping along for those 2 months?

  228. Everett F Sargent says:

    Joshua,

    We are in a very bad position at the moment. Particularly since ours (the US) is, 1st and foremost , a service economy (or call it a social economy). My best answer is a slow opening of the economy, where slow means at least a month if not two months. Unfortunately, we need to see what happens. Upticks in confirmed cases or deaths (right there we are talking like a two week delayed response) will probably occur and I don’t know how best to manage death. If an uptick is significant (e. g. 2nd wave) we are sort of screwed until a cure or we push forwards and watch the bodies pile up.

    I still think that the economy is screwed as people will still adhere to best health practices themselves regardless of how safe they all (authoritarian types like governors and mayors) say it is safe to go out and resume life as so this pandemic had never occurred. There is no turning back and we must adopt an Asian lifestyle or some such. I have grown very weary and tired of my homeland though, as most here might have guessed.

    The answer to your hypothetical is rather easy given a very high probability but I am in no position to consider those odds at the present (right now I have not myself heard of any such two month solution, so I can only put a rather low probability that such a thing could happen).

  229. Joshua says:

    Everett –

    > (right now I have not myself heard of any such two month solution, so I can only put a rather low probability that such a thing could happen).

    Sure. Probably quite low probability. But that’s my point. Sometimes you go with low probability if the potential “cost” saving is massive. Sometimes you take a significant hit to hedge against a massive cost – in other words to hedge against what the “herd immunity” approach would mean if there isn’t immunity or if the immunity is minimal.

    Im the other hand, if we’re going to go down, we can at least go down laughing.

    And

  230. Steven Mosher says:

    Super spreader

    “. What
    determines the heterogeneity of individual infectivity in a certain outbreak is unclear. Stein summarized the factors that
    may shape SSE as follows: host factors including physiological, behavioral, and immunological factors; pathogen factors
    including virulence and co-infection; and environmental factors including crowding, misdiagnosis, inter-hospital transfer,
    and airflow dynamics that influence the heterogeneity of infection [9]. These lists of factors may be incomplete, but they
    seem to explain the SSE in the Korean MERS outbreak of 2015.
    The primary goal of control strategies was to reduce reproductive numbers, and the serial changes of time-dependent
    generational and case reproductive numbers during the
    MERS outbreak in Korea were estimated to evaluate the effect
    of counter-measures [10]. SSE added a further complication
    and suggested inadequacies in the traditional approach [7, 8,
    11]. SSE played a major role in spreading infections like SARS
    and MERS, and the prevention and control measures for SSE
    should be central in controlling such outbreaks [4, 7, 11]. One
    missed super-spreader could cause a new outbreak. LloydSmith et al. showed that individual-specific strategies (for example, isolation of the infected individuals) were more likely
    to exterminate an emerging disease than population-wide interventions such as advising an entire population to reduce
    the behaviors associated with transmission [7, 11]. According
    to the model proposed by Lloyd-Smith et al., isolating infected
    individuals increased the heterogeneity of infectiousness and
    when the variation of infectiousness was large, extinction occurred rapidly [7, 11]. By taking advantage of heterogeneity,
    control measures could be directed towards the smaller group
    of highly infectious cases or the high-risk groups.

    7. Lloyd-Smith JO, Schreiber SJ, Kopp PE, Getz WM. Superspreading and the effect of individual variation on disease
    emergence. Nature 2005;438:355-9.
    8. James A1, Pitchford JW, Plank MJ. An event-based model of
    superspreading in epidemics. Proc Biol Sci 2007;274:741-7.
    9. Stein RA. Super-spreaders in infectious diseases. Int J Infect Dis 2011;15:e510-3.
    10. Park SH, Kim WJ, Yoo JH, Choi JH. Epidemiologic parameters of the Middle East Respiratory Syndrome outbreak in
    Korea, 2015. Infect Chemother 2016;48:108-17.
    11. Galvani AP, May RM. Epidemiology: dimensions of superspreading. Nature 2005;438:293-5.

  231. Steven Mosher says:

    “Steven –

    Do you realize how little contact tracing has been going in here?

    With nearly a million identified cases and many multiples of that of unidentified cases?”

    Do you realize what my argument is?

  232. David B Benson says:

    About the misuse of “models” regarding electrical power generation:
    https://bravenewclimate.proboards.com/thread/727/michael-moore-sponsored-documentary-exposes?page=1#post-6566

  233. Steven Mosher says:

    Interesting study on secondary attack rates and tracing approaches.

    Click to access bbc_contact_tracing.pdf

  234. Joshua says:

    Steven –

    Maybe there’s hope:

    > A coronavirus tracing app released by the Australian government on Sunday has already been downloaded more than a million times, despite privacy concerns.

    Registering with the app, called COVIDSafe, is optional but strongly encouraged by health officials who say it will help speed up the contact-tracing process and allow people to get diagnosed and treated early.

    Using bluetooth technology, the app logs each time that a user comes within 1.5 meters (4.9 feet) of someone else. When an app user later on tests positive for coronavirus, their log of recent encounters will be accessed and everyone who has had more than 15 minutes of close contact with them will be notified.

    https://www.washingtonpost.com/world/2020/04/27/coronavirus-latest-news/#link-BGXI32PGOBFVZJFSK7BYB4LET4

  235. Steven Mosher says:

    “Maybe there’s hope:

    > A coronavirus tracing app released by the Australian government on Sunday has already been downloaded more than a million times, despite privacy concerns.”

    yep. I am helping a developer with a similar project.

    A few things.

    There is a brief study out on how little contact tracing you actually need to make a difference.
    Obviously more is better, and a little gets you return as well. However, people estimating
    contact staffing and tests are all over the map, orders of magnitude differences

    There are other applications t help with social distancing, basically reservation systems
    for retail.
    I dont know how they will re imagine mass transit . In China you had an ID card and they
    limited seats to 50% full. stuff like that.

    One thing that makes it easier is a digital ID. lot of work ahead.

  236. Steven Mosher says:

    On a good note maybe all the people with High BMI
    will lose some weight and be less likely to die.

    https://www.cnn.com/2020/04/26/business/tyson-foods-nyt-ad/index.html

    we should probably use this opportunity to end all meat production.

    gardens, that’s the ticket. And MRE

  237. Joshua says:

    Preprint (after the Santa Clara fiasco, I’m thinking maybe no one should ever be allowed to post a preprint online?

    > This paper presents two epidemiological models that have been developed in order to study the disease dynamics of the COVID-19 pandemic and exit strategies from the lockdown which has been imposed on many countries world-wide. A strategy is needed such that both the health system is not overloaded letting people die in an uncontrolled way and also such that the majority of people can get back their social contacts as soon as possible. We investigate the potential effects of a combination of measures such as continuation of hygienic constraints after leaving lockdown, isolation of infectious persons, repeated and adaptive short-term contact reductions and also large-scale use of antibody tests in order to know who can be assumed to be immune and participate at public life without constraints. We apply two commonly used modeling approaches: extended SEIR models formulated both as System Dynamics and Agent-Based Simulation, in order to get insight into the disease dynamics of a complete country like Germany and also into more detailed behavior of smaller regions. We confirm the findings of other models that without intervention the consequences of the pandemic can be catastrophic and we extend such findings with effective strategies to overcome the challenge. Based on the modeling assumptions it can be expected that repeated short-term contact reductions will be necessary in the next years to avoid overload of the health system and that on the other side herd immunity can be achieved and antibody tests are an effective way to mitigate the contact reductions for many.

    https://www.medrxiv.org/content/10.1101/2020.04.14.20063750v1

  238. Joshua,

    Preprint (after the Santa Clara fiasco, I’m thinking maybe no one should ever be allowed to post a preprint online?

    Or, people should learn not to base their decisions on a single study/paper, especially if it’s a preprint?

  239. Joshua says:

    Yeah, that too.

    Of course I was being facitious about banning anyone from posting them online, but the problem is that we know that preprints will inevitably be used to confirm biases – especially when the authors get out in front of the data as with that Santa Clara study.

    Meanwhile….

    Yet more evidence of why all this certainty about the implications of epidemiological analyses of COVID 19, and the mortality rates in particular, are waaaaay premature and more a reflection of confirmation bias than anythjnf else:

    > In the early weeks of the coronavirus epidemic, the United States recorded an estimated 15,400 excess deaths, nearly two times as many as were publicly attributed to covid-19 at the time, according to an analysis of federal data conducted for The Washington Post by a research team led by the Yale School of Public Health.

    https://www.washingtonpost.com/investigations/2020/04/27/covid-19-death-toll-undercounted/?arc404=true

    The level of certainty we’re seeing at “skeptic ” sites about the implications of epidemiological analyses is pretty amazing.

    Confident pronunciations about the mortality rate when we don’t even know with confidence how many people died?

    Isn’t respect for uncertainty the mantra of climate “skeptics?”

    Where is the due skeptical diligence?

    Paging Mr. Monster. Mr. Uncertain T. Monster, please pick up line #1!

  240. JCH says:

    Spam.

    I bought 24 cans for the “when the power goes out” phase.

    Was hugely disappointed they’ve done away with the windup key lid opener of my youth. Now you pop the top like a beer. No fun Spam, still nutritious.

  241. Steven Mosher says:

    “Preprint (after the Santa Clara fiasco, I’m thinking maybe no one should ever be allowed to post a preprint online?”

    arrg.

    1. Assumed an R0 of 3 with no sensitivity study.
    2. Agent based model with no age dependent contact matrix ( other models use
    empirically based ‘contact” matrices which capture the age dynamics of social interaction.
    3. No age distribution used for hospitalization, use of ICU and death.

    quick read comparing to other agent based approaches.

    I pretty much stand behind what I wrote here
    https://judithcurry.com/2012/08/03/post-normal-science-deadlines/

    science in a rush is not pretty

  242. JCH says:

    Pretty current. Gao from China tells us exactly what my father’s WW2 Marines learned from China: gung ho – work together.

  243. Steven Mosher says:

    “Pretty current. Gao from China tells us exactly what my father’s WW2 Marines learned from China: gung ho – work together.”

    yup. the biggest difference I see is how the media addressed the situation in China
    All for 1, 1 for all. party line? you bet.

    Cuomo comes the closest trying to avoid politics

    A lot of stuff like this

    Some of the best stuff

  244. dikranmarsupial says:

    “Or, people should learn not to base their decisions on a single study/paper, especially if it’s a preprint?”

    One of the first things I try to teach PhD students is that just because something is published in a peer reviewed journal, doesn’t mean that it is right. Getting through peer review is the first step to acceptance by the research community, not the last.

    Might be better to ban papers having press releases, which I suspect is a source of papers getting misrepresented in the media (c.f. Spiegelhalter’s “The Art of Statistics”).

  245. izen says:

    The response to COVID19 in China and the US reveals the underlying social and economic ideology.

    China may have abandoned the strict adherence to central government control of the economy, and allowed a degree of individual entrepreneurial autonomy, such as the breeding and selling of live animals in the wet markets.
    But AFAIKS is still has a shared belief in the role of the central authority to provide food, shelter, and healthcare. Along with the socialised aspects of a modern civilisation like clean water/sewage management, police and legal services and roads/communication infrastructure.

    The US ascribes to the strict linkage of food, shelter, and healthcare to working for a private business. It is seen as a virtue for the central government to do the least possible in terms of provision of basic calories and housing. When such provision is unavoidable it is grudging. The only role that seems to be considered legitimate for central government is to provide support and bailouts to the private enterprises so that they can continue to pay a wage that enables individuals to obtain their needs.

    This is why there is a faction both in government, industry and the working population that are implacably opposed to the social distancing and lockdown measures that are indicated for the reduction in cross-infection. Those precautions are incompatible with the employment that both the workers and the business require to obtain the money for necessities and dividends.

    For the employer the threat of government interference is viewed as un-American, a Threat the to the inherent re-distributive process of the business economy. For the employee on a below average wage, ANY government regulation of their autonomy is seen as an impediment to that link between work and calories/housing/healthcare. It becomes an issue in the culture wars.
    And results in signs like these –

    and

    Obviously there is a spelling error in these two placards.
    The ‘OR’ should be an ‘AND’.

  246. izen says:

    @-SM
    “yep. I am helping a developer with a similar project.”

    Perhaps you can reveal whether the app uses local proximity detection with no central collection of the data, or requires, or is required to provide the identity and location of possible contacts. I asked about this way up-thread.

    https://andthentheresphysics.wordpress.com/2020/04/12/models/#comment-174551

    This seems to be the problem with the Apple/Google app in Europe. Germany has apparently accepted the local proximity with no central database version, the UK is is very opposed and wants a provision for central oversight.

    I take your point that even a little contact tracing goes a long way, and is most useful at the home/work/church level.
    The problem is that if you have a high rate of daily new cases even that limited contact tracing becomes a big job.

  247. Joshua says:

    izen –

    > China may have abandoned the strict adherence to central government control of the economy, and allowed a degree of individual entrepreneurial autonomy, such as the breeding and selling of live animals in the wet markets.
    But AFAIKS is still has a shared belief in the role of the central authority to provide food, shelter, and healthcare. Along with the socialised aspects of a modern civilisation like clean water/sewage management, police and legal services and roads/communication infrastructure.

    I think that’s an important part of it…but I think there’s another factor. Broadly speaking, in comparison to the US, Chinese people have a shared belief in the role of collectivism as a main tenet of their lives.

    I’ll give a generalized example from a classroom comprising largely Asian and American students. The Asian students listen intently to the teacher with the intent of being able to remember what was being said, and a wish not to interrupt the teaching/learning process with an individual focus. They don’t ask questions because they don’t want to focus the classroom on their individual needs. They hold off on formulating opinions because they’re not experts and their opinions are quite likely to be wrong.

    The American students ask a lot of questions and engage the teacher in discussions about the material at hand. They will easily formulate opinions and express them even if they aren’t really that sure that their opinions are correct. They think that by focusing on their individual needs, they will help propel the entire class to a better level of understeanding.

    My Asian students/clients often often told me of the saying that it’s the nailhead that sticks up above the rest that’s the first to get hit with the hammer. That saying, IMO, reflects a very different attitude towards the role of the individual with a societal matrix. It’s easy to over-generalize about these thing’s (and to conclude with attitude is “better”), but there’s a pretty robust literature on these differences. I’d say that some philosophical differences underlie the dynamics you’re describing in your comment.

  248. Steven Mosher says:

    “Perhaps you can reveal whether the app uses local proximity detection with no central collection of the data, or requires, or is required to provide the identity and location of possible contacts. I asked about this way up-thread.”

    Decentralized of course. I could not do a centralized anything and save face.

    The concept is that every interaction can be treated as a signed transaction stored on a
    blockchain.
    party A met party B,
    or Party A “checked in” ( like in facebook) to building X,
    The “meeting” get stored just like a transaction in Bitcoin.
    Everybody can see , A met C, A “met” location X
    But nobody knows who A and B are
    After 14 days all transactions get burned.
    If you test positive the health worker can “unlock” the transaction and “B” will get
    a text notice that he has met someone or passed through the same location as someone who has tested positive.

    For your friends and close contacts and workmates, that is different. Pretty easy if
    the person has facebook, or linked in.

  249. Steven Mosher says:

    “Yet more evidence of why all this certainty about the implications of epidemiological analyses of COVID 19, and the mortality rates in particular, are waaaaay premature and more a reflection of confirmation bias than anythjnf else:”

    people have lost the train of argument

    In the beginning, the argument was LOOK AT ALL THE DEAD BODIES WE WILL HAVE
    of course many making this argument knew that the denominator in deaths/cases
    was low. but hey you go with the data you have.

    the second argument that was made was EVEN IF the deaths are lower, it’s all about the
    hospital collapsing. Vents vents vents, we need vents and transform the javitz center, hurry hurry.

    Now that the peak has passed and we have a little serology to start to BALLPARK
    a lower death rate and Now that we see the hospitals can handle the load…..
    the argument will change again and it will be

    How can we be as safe as possible.

    And more creative folks like Cuomo will say.. we don’t want t restart a broken system
    we need to transform everything.

  250. Joshua says:

    > Now that the peak has passed and we have a little serology to start to BALLPARK
    a lower death rate and Now that we see the hospitals can handle the load…..

    How do you ballpark from a non-random and non-representative sample to even get the most basic requirement to justify a broad extrapolation to a population that’s orders of magnitude bigger and more varied?

    Santa Clara, as an example, has 4x the national median income and isn’t nationally representative for race/ethnicity. SES has a huge influence on health outcomes!

    Maybe if people identified uncertainty ranges and priors…

    but they’re not doing that. They’re just running with their priors. That’s just messed up.

    Respect the uncertainties. Speaking of which.

    > Coronavirus deaths more than twice hospital toll, data indicate

    https://www.ft.com/content/0ed8ea34-ebc5-4425-b86a-7a29447de57b

  251. Joshua says:

    Borrowing someone else’s math:

    > If you know how many people died, and you know how many people were infected, then you can calculate that ratio.

    The problem with these studies is that they don’t quantify their uncertainty even though there’s a lot of it. Suppose d percent of the country has died, and q percent of the country has had the virus, then d/q is the infection fatality rate (so far, there are still sick people who will die in the next weeks).

    but suppose d is about 100k people out of 330M = .1/330 = .000303 (this assumes undercounting of about 2x) and we’re trying to estimate q, and q can realistically be anything between .003 and .05, then the infection fatality rate could by anything from .000303/.003 = .101 to .000303/.05 = .00606

    anywhere from 10% to 0.6%

    That’s a HUGE range. Anyone who comes out and says “the infection fatality rate is around 0.2%” is not only ignoring a huge range of uncertainty, they’re claiming a number that isn’t even rationally in the high probability density region

  252. izen says:

    @-joshua
    ” I’d say that some philosophical differences underlie the dynamics you’re describing in your comment.”

    And I might have replied in the distant past that it is the differences in the economic structure of the two societies that underlie the dynamics.

    These days I would settle for the philosophical and economic differences between collectivism/socialism and individualism/capitalism are just two aspects, or sub-systems of a higher level of process. Each interacts with the other in a complex mutuality.

    Each has benefits and disadvantages in effectively providing people with what they need and want.
    The diverse outcomes when faced with a new infective disease are of interest.

  253. Joshua says:

    izen –

    > And I might have replied in the distant past that it is the differences in the economic structure of the two societies that underlie the dynamics.

    Fair enough.

    > These days I would settle for the philosophical and economic differences between collectivism/socialism and individualism/capitalism are just two aspects, or sub-systems of a higher level of process. Each interacts with the other in a complex mutuality.

    I agree.

    > Each has benefits and disadvantages in effectively providing people with what they need and want.
    The diverse outcomes when faced with a new infective disease are of interest.

  254. Joshua says:

    > Each has benefits and disadvantages in effectively providing people with what they need and want.
    The diverse outcomes when faced with a new infective disease are of interest.

    Yeah, that too.

  255. izen says:

    @-SM
    “And more creative folks like Cuomo will say.. we don’t want t restart a broken system
    we need to transform everything.”

    There are, or seem to be, very distinct differences in the mortality rate between different societies dependent on the form of collective response. Trying to discover a specific fatality rate for the virus based on the age/co-factors of the individual, but independent of the society of which they are a part may be an error.

    If we are shoving words into Cuomo’s mouth, perhaps he would say that the system is not broken, but it does need some significant modifications to make it fit for purpose.
    Or at least that the strong link between wages and basic needs is problematic when dealing with a contagious disease that can best be managed with social distancing/quarantine.

  256. JCH says:

    The thing that is going to play out over the next couple of weeks is heat. Will heat slow/stop the spread? I don’t really believe lazy western-style social distancing can be as successful as the numbers indicate; I think it’s heat.

  257. jacksmith4tx says:

    per SM quote of Cuomo’s comment “we don’t want t restart a broken system
    we need to transform everything.”

    I give you my (former) candidate for president of the USA, Sen. Bernie Sanders:
    as published today,
    https://www.politico.com/news/magazine/2020/04/28/covid-health-care-proposal-uninsured-medicare-212973

    JCH:
    It seems we Texans don’t understand how a pandemic works. Texas gets a -D.
    Given the average delay from infection to sickness this is how you make sure the virus stays in circulation.
    This link should be free: https://www.star-telegram.com/news/coronavirus/article242317616.html

  258. JCH says:

    Veterinary scientists explain it well:

    CoVs are characterised by an exceptional genetic plasticity and evolve rapidly, changing their antigenic profile, tissue tropism or host range by means of two distinct mechanisms. The viral replicase (an RNA dependent-RNA polymerase) does not possess a good proof reading activity, therefore the incorporation of wrong nucleotides at each replication cycle and the consequent accumulation of mutations in the viral genome lead to a progressive differentiation of the viral progeny from the parental strain. This mechanism, which is well known for influenza viruses being responsible for the so called antigenic drift, may cause the progressive adaptation of the viral surface proteins to the cell receptors of new animal species, increasing the viral fitness. In addition, the particular replicating machinery of CoVs facilitates recombination events due to the presence of consensus sequences upstream each gene. Therefore, in the case of coinfection by more than one CoV strain, the RNA polymerase can jump from the RNA of a strain to that of the other one, synthetizing a hybrid RNA containing sequences from both viruses. Recombination can occur not only with genomic sequences of other CoVs (homologous recombination), but also with RNAs of different viruses and other organisms (heterologous recombination) (Luytjes et al., 1988; Banner and Lai, 1991; Lai, 1996; Zeng et al., 2008; Huang et al., 2016). Recombination is an alternative mechanism that let CoVs acquire novel biological properties in terms of virulence, host range and tissue tropism, so that CoV strains, which are non-pathogenic or low-pathogenic in the original host, may increase their pathogenicity in the same species or adapt to different species spreading in the new host with exceptional rapidity (Banner and Lai, 1991).

  259. David B Benson says:

    JCH, can you provide a link?

  260. Jim Hunt says:

    David,

    The Great God Google is your friend. Cut’n’paste reveals:

    https://doi.org/10.1016/j.vetmic.2020.108693

  261. verytallguy says:

    “And more creative folks like Cuomo will say.. we don’t want t restart a broken system
    we need to transform everything”

    I don’t know anything about Cuomo’s desires for transformation, but you can be sure people from all sides of the political spectrum will be aiming to use this upheaval to their own ends, whether attempting to attack China, build more bicycle lanes, or hide the consequences of Brexit, to pick a few at random.

  262. Chubbs says:

    JCH, your post triggered a thought. Ability to jump multiple species and thrive has already been selected for.

  263. Steven Mosher says:

    “I don’t know anything about Cuomo’s desires for transformation, but you can be sure people from all sides of the political spectrum will be aiming to use this upheaval to their own ends, whether attempting to attack China, build more bicycle lanes, or hide the consequences of Brexit, to pick a few at random.”

    humans.

    I have resisted the urge to make charts. It is hard with all the uncertainty to not reach out for our comfort blankets.

  264. Steven Mosher says:

    “How do you ballpark from a non-random and non-representative sample to even get the most basic requirement to justify a broad extrapolation to a population that’s orders of magnitude bigger and more varied?”

    simple you ballpark.
    One ballpark is 6.9% CFR. current cases/current case count

    the Santa Clara ball park lowers this. How much, not enough to make it relevant.
    the NY rate also lowers it, not as much as Santa clara.

    so you end up with a range of values.. right now the range is still uninformative.
    and then you realize that maybe serology is pointless. because no politician will make his decision based on deaths. More than 1 has become too many. Maybe serology helps you plan for body bags. Other than that… not sure what it does.

  265. Willard says:

    > I have resisted the urge to make charts.

  266. JCH says:

    This another issue that interests me. Jeremy Farrar commented in one of the videos that there is data from Canada, where they had a SARS-CoV-1 outbreak, that ADE is not happening, which would be good, but it’s still something to think about in the corner cutting, mad rush to develop vaccines:

    News Feature: Avoiding pitfalls in the pursuit of a COVID-19 vaccine

  267. Joshua says:

    > the Santa Clara ball park lowers this. How much, not enough to make it relevant.
    the NY rate also lowers it, not as much as Santa clara.

    Sure. Ballpark Santa Clara. Ballpark New York.

    What you don’t do is ballpark NY based on Santa Clara data. And you don’t ballpark a IFT based on Santa Clara. That’s what Ioannidis et al. did.

  268. Joshua says:

    IFR

  269. David B Benson says:

    Yet another COVID-19 model:
    https://medicalxpress.com/news/2020-04-track-covid-.html
    Has location tracking.

  270. David B Benson says:

    A model of the economy with COVID-19:
    https://phys.org/news/2020-04-economic-worse-lockdown-social-distancing.html
    by a Cambridge economist.

  271. Steven Mosher says:

    Joshua.
    “Ioannidis ?”

    I have 0 respect for his work.

  272. verytallguy says:

    Steve, that forecast strikes me as near useless – it’s based on case numbers, which are strongly dependent on testing policy and capacity.

    I would speculate it overestimates R significantly as a result.

  273. Swedish models.

  274. Joshua says:

    Steven –

    Do you have 0 respect for this?:

    https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5540883/

  275. dikranmarsupial says:

    My respect for Ioannidis is in the interval (0, 1). If you overstate the significance [sic] of your work, it detracts even from the good bits. A lot of the kerfuffle is cause by having an unrealistic expectation of what peer review (or NHSTs) can actually achieve. Babies being thrown out with the bathwater in both cases.

  276. verytallguy says:

    Lomborg misrepresents the article he cites, and the situation in Sweden. Colour me surprised.

  277. Joshua says:

    dikran –

    I just ran into someone over at Gelman’s blog who’s saying NHST are useless.

    I’m chewing on it. My initial reaction was that NHST shouldn’t be assumed as dispositive (especially a single study), but that when they’re done well they add value/inform probabilities (especially through well-conducted meta analyses that stratify by things like statistical power).

    Thus, it seems to me, they can inform cost/benefit analyses. But maybe I’m just resistant to change.

    Maybe meta-analyses just propogate error.

    His response:

    > NHST amounts to comparing a p-value calculated for an arbitrary statistical model to an arbitrary cutoff. What information could it possibly provide?

    Instead come up with some model of the uncertainties in benefits, risks, and costs based on assumptions you put some thought into and make a decision. Eg smaller expected benefit could still be worthwhile for a cheap and safe treatment. But we would require large expected benefit for an expensive treatment with lots of dangerous side effects.

    +++++++

    Your thoughts?

  278. Ben McMillan says:

    Think I like the NZ/Aus model better than the Swedish one. Apart from anything else, it minimises the economic impact.

    I remember a few weeks ago the usual suspects were claiming that Sweden was ‘following the same curve as Switzerland’…

  279. Joshua says:

    If I were God of COVID-19 arguments, I would ban comparing across countries either way.

    Waaaay to many uncontrolled variables

  280. Joshua says:

    Ben –

    Sweden has more deaths per million, but fewer cases per million, and less testing per million than Switzerland.

    But still, I dunno. I just think comparing across countries is mostly an exercise in confirmation bias.

    It’s too tempting to cherry-pick.

  281. dikranmarsupial says:

    Joshua,

    this blog post (and the paper that informs it) is worth reading https://dougmcneall.com/2014/02/03/a-brief-observation-on-statistical-significance/

    NHSTs are a useful tool in enforcing a degree of self-skepticims and are most useful when you get a non-significant result. Using them as a source of support for your argument is risky, as in the quote ” He uses statistics as a drunk uses a lamppost – more for support than for illumination.”

    “> NHST amounts to comparing a p-value calculated for an arbitrary statistical model to an arbitrary cutoff. What information could it possibly provide?”

    I’d say this is a bit of an overstatement if meant to be generally true. The null generally isn’t arbitrary (it is the thing you must negate for your research hypothesis to continue) and neither is the cutoff if you follow Fisher’s actual advice and set it according to the nature of the experiment.

    The real trouble is that people generally don’t put enough thought in to the question they want statistics to answer and often settle for a very indirect answer that they then (mis-)interpret to suit their position, Which isn’t really the fault of statistics,

  282. dikranmarsupial says:

    Just to add, I completely agree with “Instead come up with some model of the uncertainties in benefits, risks, and costs based on assumptions you put some thought into and make a decision”

  283. Joshua says:

    dikran –

    What about the value of stratified (on variables like sample size, statistical power, etc.) meta-analyses?

  284. Ben McMillan says:

    Joshua: I know there have been a lot of comments at this point, but various commentators upthread (including you) were trying to explain why Sweden+Switzerland were following the same curve.

    Of course it was a cherry pick: but it wasn’t me that picked it, and three weeks later, it doesn’t look so sweet.

  285. dikranmarsupial says:

    It depends on the nature of the problem, like all tools they are useful for some problems, but bad for problems for which they were not intended (or at least for which better tools are available – it is possible to hammer in a nail with a screwdriver). The “cookbook” approach to statistics is bad. I very much agree with Gelman that it is about constructing (bespoke) models of the uncertainties and putting thought into the assumptions and limitations.

  286. Everett F Sargent says:

    What RIghtWingNut got wrong …

    /: = slanthead thinking

  287. Joshua says:

    Ben –

    > I know there have been a lot of comments at this point, but various commentators upthread (including you) were trying to explain why Sweden+Switzerland were following the same curve.

    We appear to have a communication problem. I was never trying to explain why they were on the same curve.

  288. Joshua says:

    dikran –

    Just to be clear, in case it seemed otherwise that was a comment from a commenter, not Gelman.

  289. Everett F Sargent says:

    “What RIghtWingNut got wrong”

    should be …

    “What RIghtWingNut MEDIA got wrong”
    Previous comment directed at wingnot media NOT anyone here. Sorry about that one.

  290. Everett F Sargent says:

    wingnot = wingnut = almost the same thing

  291. JCH says:

    So with uncounted bodies, Trump has killed ~70,000 people with his half-assed solutions. Public figuring out how freaking gawd awful the corrupt dummy is at leadership. Polls down.

    So, let’s start a war with China:

  292. Everett F Sargent says:

    Video above: Start = 45:15 Stop = 46:55/

    The word “model” is used once in that section and IMHO is taken out of context wrt Sweden. But watch that section as all three individuals there make some very important points.

    Oh and don’t ever expect the US to adopt the so-called Swedish model because the US ain’t going to e like Sweden, like ever even …

  293. Everett F Sargent says:

    “Video above: Start = 45:15 Stop = 46:55/”

    should be …

    “Video above: Start = 40:15 Stop = 46:55.”

    Not off to a good start today. :/

  294. Ben McMillan says:

    I actually thought that the comparison was not particularly problematic (Sweden and Switzerland are a pretty good match in many ways), but it was just too early to tell three weeks ago, and that you would expect Sweden to end up with a higher mortality rate than Switzerland if strong social distancing was the main thing that matters. And that is in fact what has happened.

    In fact, ‘strong social distancing is the main thing that matters’ appears to be the key to modelling how countries’ mortality rates change with time: James Annan’s analysis demonstrates this pretty clearly.

  295. Joshua says:

    > We describe the epidemiology of a coronavirus disease (COVID-19) outbreak in a call center in South Korea. We obtained information on demographic characteristics by using standardized epidemiologic investigation forms. We performed descriptive analyses and reported the results as frequencies and proportions for categoric variables. Of 1,143 persons who were tested for COVID-19, a total of 97 (8.5%, 95% CI 7.0%–10.3%) had confirmed cases. Of these, 94 were working in an 11th-floor call center with 216 employees, translating to an attack rate of 43.5% (95% CI 36.9%–50.4%). The household secondary attack rate among symptomatic case-patients was 16.2% (95% CI 11.6%– 22.0%). Of the 97 persons with confirmed COVID-19, only 4 (1.9%) remained asymptomatic within 14 days of quarantine, and none of their household contacts acquired secondary infections. Extensive contact tracing, testing all contacts, and early quarantine blocked further transmission and might be effective for containing rapid outbreaks in crowded work settings.

    https://wwwnc.cdc.gov/eid/article/26/8/20-1274_article

  296. Joshua says:

    Ben –

    Consider the following when you compare Sweden and Switzerland:

    > Also, the social distancing policy is not a problem in Sweden, because that is how it works normally. It is more than a joke when quoting the swedish reaction to the distancing rule of 2m: “that close?!!”. We don’t adress or greet strangers. Chit-chatting about nothing in public is not heard of except from the large immigrant minorities and the occasional “boomer” missing the good old days of demonstrations for some obscure communist guerilla, made up of three letters…We are fullfilling the demands of social distancing without effort.

    https://judithcurry.com/2020/04/26/a-sensible-covid-19-exit-strategy-for-the-uk/#comment-915893

    That joke is pretty funny

  297. Willard says:

  298. David B Benson says:

    JCH & Jim Hunt, thank you.

  299. Steven Mosher says:

    Joshua ( I thought I posted this above)

    ‘Of 1,143 persons who were tested for COVID-19, a total of 97 (8.5%, 95% CI 7.0%–10.3%) had confirmed cases. Of these, 94 were working in an 11th-floor call center with 216 employees, translating to an attack rate of 43.5% (95% CI 36.9%–50.4%). The household secondary attack rate among symptomatic case-patients was 16.2% (95% CI 11.6%– 22.0%). Of the 97 persons with confirmed COVID-19, only 4 (1.9%) remained asymptomatic within 14 days of quarantine, and none of their household contacts acquired secondary infections. Extensive contact tracing, testing all contacts, and early quarantine blocked further transmission and might be effective for containing rapid outbreaks in crowded work settings.”

    yep. the case was followed quite closely on the nightly news in Seoul.

    some things to note.

    Tracers did not have to go looking for contacts and co workers who the patient came into contact with. The company turns over the names of employees. They all go for testing and are ordered to stay home for 14 days. During which time they will get multiple tests and have to report
    there temperature twice daily. Then the rest of the building was tested (like floors 8-13)
    both residents and workers.

    As for the office space a large percentage came from one corner of the office. The office was
    pretty crowded
    http://tbs.seoul.kr/eFm/newsView.do?typ_800=O&idx_800=2388019&seq_800=

    Wack a mole

  300. Steven Mosher says:

    Joshua, Cuomo will be requiring business to set their rules for safe opening.

    here is what I would do based n what I’ve seen.

    Offices.
    1. Company APP .
    A) all employees must have the app.
    B) daily recording of symptoms, temperature etc
    C) app sucks in all phone contacts, linked in, FB. ( category family, friends,coworkers, roomates
    D) employes use app to daily check in their actual contacts .
    E) Meetings and attended are logged in the app.
    F) floor plan and seeting map
    2. Building security:
    A) visitor log in at reception. no reception? get one.
    B) temperature check of all who enter
    C) Mask check, free mask if they dont have one.
    D) Building sweep: All employee temps are logged 2 times a day.
    3. Cleaning staff.
    A) constant patrol cleaning public surfaces: buttons, door handles, surfaces
    In china we had 2-3 ladies working all day with disenfect. The smell of beach
    lovely.
    4. Seating and attendance
    A) Work from home is maximized
    B) On restart staff levels are limited to 2m apart. Bascially we started with
    50% showing up in shifts.
    C) Flex time to keep in building load low.
    5. Delivery.
    A) n delivery staff ( food, FedEx ) is allowed in the building
    B) All delivery personnel are logged. date and time.

    It took a day or 2 for the office in china to set this up.

    There were some national guidelines but cities put their own twist on it adding details
    here and there.

    Things that helped.

    1. national ID
    2. Phones tied to real people. no burner phones allowed
    3. QR code system. Everything is paid for digitally by scanning QR codes that
    are tied to your bank account and ID


    I got into digital currency for other reasons, but it is proving useful

  301. Steven Mosher says:
  302. izen says:

    @-SM
    “Cuomo will be requiring business to set their rules for safe opening.”

    I get the strong impression that in many other cases,
    Business will be requiring Governers (?) to set their rules for safe opening.

    Your list of requirements for safe working is quite impressive.
    Does anyone think that anything like them will be imposed on the meat processing plqants that are have been required by the War Powers act to reopen ? Most employ those in the population with the least access to smart phone apps and a digital ID. Many of these factories have been raided in the past to catch the illegal immigrants they tend to employ.

    FRED looks interesting, but I have doubts about its ability to accurately reflect the process of –
    “Individuals interact within realistic household, school, and workplace social networks.”
    without making unstated ideological assumptions. I suspect that any modelling of CORVID19 will have to include the socio-economic structure of the affected population as much, if not more, than the infectivity and virulence of the virus. That does not automatically arise from agent-based modelling as the larger system is a as much a determinate of the choices of agents as it is an emergent property of those choices.

  303. Joshua says:

    Steven –

    > Offices.
    1. Company APP .
    A) all employees must have the app.
    B) daily recording of symptoms, temperature etc
    C) app sucks in all phone contacts, linked in, FB. ( category family, friends,coworkers, roomates
    D) employes use app to daily check in their actual contacts .
    […]

    I doubt many employers would agree to these rules, but even for those that might I’d imagine they would have many employees that would not agree.

    What would happen then? The employees get fired? Companies will fire people for not agreeing to be electronically surveilled?

    Doesn’t seem remotely plausible to me, for a number of reasons.

  304. John Hartz says:

    CNN has done nice job of graphically presenting COVD-19 data for the US, states, and counties.

    Tracking Covid-19 cases in the US

    Since January, novel coronavirus has spread to nearly every state and territory

    By Sergio Hernandez, Sean O’Key, Amanda Watts, Byron Manley and Henrik Pettersson, CNN

    Last updated: May 1, 2020 at 9:45 a.m. ET

    https://www.cnn.com/interactive/2020/health/coronavirus-us-maps-and-cases/index.html

  305. John Hartz says:

    Unfortunately, most of the COVID-19 data generated for the US, states, and counties are like blocks of Swiss cheese.

  306. Steven Mosher says:

    Its may 1st

    Cuomo faced with a continued 1K cases per day asks the question?
    where do these people live and work? do they ride the subway?
    gosh

  307. Steven Mosher says:

    “Your list of requirements for safe working is quite impressive.
    Does anyone think that anything like them will be imposed on the meat processing plqants that are have been required by the War Powers act to reopen ? ”

    Ah I dunno.

    illegals are tough, lemme think. I have some products I am pitching

  308. JCH says:

    Reviewing business-set rules for a Fortune 500.Ω. Closing most bathrooms; forcing everybody on a floor to use the same one??? To force social distancing at sinks, only one sink per floor bathroom will be operational??? Closing break area. Claim they will have hand sanitizers broadly available. Employees will be fined if they do not use the handrails in the stairways??? Again, hand sanitizer will be available. No mask requirement??? Have been working at home and have really gotten more done at home than they were in the office. The instruction indicate some face-to-face time will improve that efficiency. But, they will be working alternate days, so the face-to-face will have one at the office and one at home????

    For the luv of gawd, hire some people from China to run this thing.

  309. Another way to look at it

  310. dikranmarsupial says:

    I think the new thread on outgoing longwave radiation suggests that the physics may only be basic for large values of “basic”. It isn’t clear to me that you can estimate the amount warming with all of the relevant “basic phsyics” without a fairly complicated model. Things like simple energy balance models can clearly get you in the right ballpark, but that doesn’t mean they are not giving you the right answer for a not-so-right reason.

    Things should be made as simple as possible – but no simpler.

  311. Everett F Sargent says:

    So talking about SE for a brief moment …

    I will will briefly describe what the so-called busy ULH graph is about, Where N = 4 = DK, FI, IS and NO there are 2^N-1 = 15 combinations of there four (nordic countries) versus SE. From those original 15, three were removed (IS, FI and FI+SI (although two of these FI and FI+SI look good over the past week)). Each of these divisors accounts for population differences in the various combinations (pro rata ~ per capita) and one divides SE cumulative deaths by these numbers to determine pro rata SE deaths (LLH graph). The LRH graph is just the difference Between SE deaths and SE pro rata deaths and is called SE excess deaths. The dotted blue lines in the lower two graphs form the envelope of the 12 included combinations, the divisors are roughly a factor of two between minima and maxima (say factor of 3 to 6 as shown in the ULH graph).

    Doubling times are only shown for n = 4, SE and the US.

  312. Evidently, individual choice in Sweden is about as effective as government mandate, with the end result being similar economic decline in Sweden as elsewhere.

    As for other models, Ray Dalio has a working theory of long term debt cycles in which late cycles are characterized by currency devaluation, leading to economic inequality which fans the flames of populism and leads to political ‘turmoil’.

    Dalio writes that COVID-19 and the response were just precipitating events, and something else would have precipitated the same results were COVID not to have occurred, because of global debt accumulation.

    The obvious example, perhaps too obvious, is the global depression preceding WWII.

    Does COVID-19 precipitate global depression, and WWIII?

  313. Joshua says:

    Everett –

    SE= ?

  314. Joshua says:

    TE –

    > Evidently, individual choice in Sweden is about as effective as government mandate, with the end result being similar economic decline in Sweden as elsewhere.

    One study rule applies, nonetheless, the argument that government mandates are what have caused/will cause the economic harm, as distinct from hundreds of thousand of people getting infected and tens of thousands of people dying, has always seemed dubious to me – and likely a function of motivated reasoning.

    The toxic mix of binary thinking and identity protective cognition strikes again.

  315. Joshua says:

    TE –

    My guess is that there are probably discrete variances that get hidden within an analysis that aggregates all the effects.

  316. Everett F Sargent says:

    Joshua,

    SE = Sweden
    IS =Iceland
    FI = Finland
    DK = Denmark
    NO =Norway

    The so-called five Nordic nations. All combinations of those four nations versus SE. The doubling time is a 9-day log-normal OLS fit. I included the US as representative of pretty much the worst major country in terms of doubling times (blame Small Hands for that one, starting lockdowns 2-3 weeks later then they should have, aka negative national leadership). Here is another one for you (basically a Santa Clara vs NYC Metro update) …

    I am currently of the opinion that R0 is mostly useless as it assumes exponential growth, which would show up as a constant doubling time in these graphs (e. g. a horizontal line). NYC Metro (with updated JHU time series throughout (e. g. some backdating)) now shows a doubling time right at one day and the doubling time is rather linear in log-normal space (which is an indicator of subexponential growth, heck anything going towards the upper right is an indicator of subexponential growth)).
    https://en.wikipedia.org/wiki/Basic_reproduction_number

    Oh and, in general, lockdowns work! 🙂

  317. Everett F Sargent says:

    Oh and I’m using two digit country codes …
    https://en.wikipedia.org/wiki/ISO_3166-1_alpha-2

  318. Joshua says:

    Everett –

    Thanks.

    Duh (on phone, didn’t expand charts enough to actually read them ).

  319. Everett F Sargent says:

    As usual, I spoke too soon …

    Entire US. First week ~ exponential (roughly constant horizontal line), next ~week+ superexponebtial (NYC kicks in on ~3/16 so that doesn’t immediately reduce doubling times) and the remainder ~3/19 forward to present is subexponential. So, I guess there may be some utility for R0 for larger geopolitical areas (e. g. more people in the pool provided that they are synchronized swimmers)?

    Most of the real early death doubling times have very small counts, often less then say 10-20, so that if another hot spot shows up it can change the doubling times significantly (also note uneven reporting (e. g. weekend effects) for causing the doubling times to jump about for low n-day fits).

  320. Everett F Sargent says:

    Joshua,

    OK, got it, I sized the graphs more for desktop use (right click to see graph in separate window). I just got a 34″ 21:9 aspect ratio monitor, so somewhat the opposite effect. I haven’t figured out how to view web pages in ultrawide mode (or if it is at all possible to do so).

  321. Marco says:

    “Doesn’t seem remotely plausible to me, for a number of reasons.”

    Well…
    https://www.worktime.com/what-are-the-u-s-employee-monitoring-laws-get-updated-in-2020

  322. Joshua says:

    Marco –

    Sure. But are companies going to fire employees for not agreeing to surveillance?

    I think this may be instructive:

    > Stillwater Mayor Will Joyce amended an emergency declaration requiring masks be worn by residents after businesses reported employees being threatened by patrons who didn’t want to wear masks.

    According to the Stillwater News Press, Joyce said the order had to be changed after just three hours.

    “In the short time beginning on May 1, 2020, that face coverings have been required for entry into stores/restaurants, store employees have been threatened with physical violence and showered with verbal abuse,” Joyce said. “In addition, there has been one threat of violence using a firearm. This has occurred in three short hours and in the face of clear medical evidence that face coverings helps contain the spread of COVID-19.”

    The amended declaration makes the masks optional through May 31.

    https://oklahoman.com/article/5661461/stillwater-business-employees-threatened-over-mask-requirement

  323. Everett F Sargent says:

    Marco,

    That works only in cases where the devices are supplied to the employees gratis. Or devices necessary to do work on employer supplied networks (e. g. try using VPN on government networks, you will be busted in less then a day). No employer has the right to install their software on anyone”s personally owned devices (e. g. nope, I don’t own a cell phone (they lie) or tablet (well actually almost no one owns a tablet today)).

    But yeah, not a bad way to get a free cell phone from your employer, which you then turn off outside of working hours (remove the battery and SIM card or place said device in a heavy metal box).

    I’m quite sure there will be zero legal challengers from Confederate flag waving, gun toting individuals in Michigan, for example.

    Our republic was never meant to allow surveillance at the invasive levels allowed for in other countries (local, state and federal laws). I do think that we still have some privacy rights in the US (the ACLU would be so all over this ASAP) such that when in a non-public location no one should be allowed to know where you are and what you are doing, except for illicit actions where Johnny Law has a court order, wiretap etceteras.

    IANAL, but then again neither is SM, so whatever measures are taken the default reaction is bound to be a lawsuit right around a 247 corner. And not everyone has an employer.

    Regardless, at least tens of millions in the US will never be surveilled properly.

    Note to self: I think SM can only think of ways where existing US laws are broken. IMHO

  324. John Hartz says:

    Everett F Sargent: It is high time for the USA to transform our social fabric and laws to embrace the concept that the common good, not individual rights, are paramount. If we don’t, we will never be able to properly cope with pandemics like COVID-19 and, more importantly, with man-made climate change. We did it in the Great Depression and in WWII, we can do it again.

  325. Everett F Sargent says:

    JH,

    It was high time like 40 years ago. So that I’m not exactly holding my breath on that happening any time soon …

  326. John Hartz says:

    Good news from Down Under….

    “I think many people are surprised in Australia at how well we have done. Really, when you look across all the states and territories, this is the safest place to be in the world, perhaps other than New Zealand,” she said.

    Neighboring country New Zealand recently achieved its ambitious goal of “eliminating” the coronavirus, meaning that while the daily infection rate has yet to reach zero, the country is able to track the origins of each new case.

    How did Australia flatten its coronavirus curve? Restrictions easing as infection rate continues to fall< by Nectar Gan, CNN, May 1, 2020

  327. Steven Mosher says:

    I doubt many employers would agree to these rules, but even for those that might I’d imagine they would have many employees that would not agree.

    What would happen then? The employees get fired? Companies will fire people for not agreeing to be electronically surveilled?

    Doesn’t seem remotely plausible to me, for a number of reasons.

    Its is pretty simple

    employers in New york will have to present plans to Cuomo to re open.
    no plan. no re open.

    electronically surveilled?

    If you object then I suppose I can just ask you to write down your contacts.

    Imagine you have a business with 100 people.
    75 agree to use the APP
    they enter their friends family contacts co workers
    25 dont. But the 25 who dont are listed as contacts and co workers of these who do use the app.
    you reach a point that is akin to herd immunity, but it’s herd surveillience

    So you dont use the app. Not a huge problem as your friends, family and co workers will
    have you in the system. It’s like an infection.

    How do you think social media works anyways?

  328. Steven Mosher says:

    “Reviewing business-set rules for a Fortune 500.Ω. Closing most”
    bathrooms; forcing everybody on a floor to use the same one??? To force social distancing at sinks, only one sink per floor bathroom will be operational??? Closing break area. Claim they will have hand sanitizers broadly available. Employees will be fined if they do not use the handrails in the stairways??? Again, hand sanitizer will be available. No mask requirement??? Have been working at home and have really gotten more done at home than they were in the office. The instruction indicate some face-to-face time will improve that efficiency. But, they will be working alternate days, so the face-to-face will have one at the office and one at home????

    For the luv of gawd, hire some people from China to run this thing.”

    Yup. those rulz are stupid. One wonders if they even asked or researched what works
    in other places. With hand santizer I bet people will steal it.

    Just some observations on cultures

    i went for 2 team building exercises in china. VERY different. Given a team task they just get
    to work. no organization, no meetings, no discussion, they just get busy. afterwords
    I pulled my friend aside and explained to her what ugly Americans meant by Chinese fire drill.
    But, in reality it worked. Shit got done. quickly

    Cleanliness. Korea is super clean. hand towels at meals (before covid). bathrooms spotless
    China? probably have gotten food poisoning 2-3 times. bathrooms? no soap, no toilet
    paper, generally a mess. ( that is changing)

    Smoking: China, smoke everywhere even in restaurants where there is a no smoking sign.
    Smoke in the bathroom at work? guards dont care. Korea? little isolation booths
    on the street for smokers.

    Lines: Hong Kong know how to line up. no line cutting. China, every man for himself. Line
    cutting and tightly packed lines are normal. Korea? pretty orderly on lines.

    Hotel staff: China: maid spends 5 minutes, sheets changed whenever. Korea, maid spends an
    hour, everything spotless. They will organize shit if you leave stuff out.

    I am betting that China will learn some lessons and make some changes to improve cleanliness
    they just get stuff done. watching them adjust travel rules into Beijing was amazing.
    reconfigure PEK, then change to route all flights to other cities first, adaptive strategies.

  329. Steven Mosher says:

    The other challenge for the US employers will be the demand from some workers to
    provide a SAFE working environment. we all know the workplace
    must be free of hostility, imagine the confusion when some workers demand
    temperature taking and mask wearing and electronic tracking and disease free co workers
    while other employees object.

    what are you going to do? fire workers who demand a safe workplace?

  330. Joshua Brooks says:

    Steven —

    How many employees were threatened in Korea for trying to get customers to wear masks?

    You’ve been out of the country for too long. Too many gun-totin’ freedom fighter nutjobs here. I’m not talking about the rare megachurch group. I’m talking about most of my neighbors in my rural community. They think that contact racing’s for sissies.

  331. David B Benson says:

    Joshua Brooks — “Trace or die.”

  332. Joshua Brooks says:

    > imagine the confusion when some workers demand
    temperature taking and mask wearing and electronic tracking and disease free co workers
    while other employees object.

    Yes, we have a problem. This is what the “openers” ignore. in their zeal to hate them on some libz. They act as if we’re going to go back to normal tomorrow just ’cause Trump says we should “liberate.”

    Imagine running a business at 50% staff with social distancing, many key personnel missing, and of that 50% that are working, 1/2 demand a safe workplace and the other half fighting for their freedom to infect their coworkers. And your overhead is close to what it was before COVID. And your income stream is maybe half of what it used to be.

    This might take years. Maybe eventually the freedom fighters will come around. But it will take a lot of time and a lot of pain. Oklahoma? Will it ever get there?

  333. jacksmith4tx says:

    It’s only been a few months since the pandemic started but that’s enough time and data to project what the temporary reduction in emissions might mean to long term coupled global climate model projections. This project uses A.I. and a array of data sources to create a range of post COVID-19 scenarios.
    https://www.hypergiant.com/covid/
    ACES: the After COVID Emissions Simulator illustrates the relationship between the current changes in how we live and the long term impact of these changes on the planet’s health. The tool is designed to show what happens if we immediately snap back to pre-COVID lifestyle in terms of consumption and emissions versus maintaining some of the societal changes and adding new interventions to further mitigate climate change.
    The tool takes emissions data from multiple sources including the Energy Information Administration (EIA) and the Environmental Protection Agency (EPA)

  334. Joshua says:

    Speaking of tracing…

  335. Joshua says:

    David –

    They’re refusing to even wear masks. Don’t tread on me. The tree of liberty….

  336. David B Benson says:

    Joshua, you can point out to that crowd the fact that public health officers are commissioned and can enforce quarentine for as long as they find necessary, on the pain of federal punishment.

    I am old enough to remember passing a house with a big QUARENTINE, DO NOT ENTER sign pasted across the front door and three children inside waving out the front window as we passed.

  337. Steven Mosher says:

    “Yes, we have a problem. This is what the “openers” ignore. in their zeal to hate them on some libz. They act as if we’re going to go back to normal tomorrow just ’cause Trump says we should “liberate.”

    Imagine running a business at 50% staff with social distancing, many key personnel missing, and of that 50% that are working, 1/2 demand a safe workplace and the other half fighting for their freedom to infect their coworkers. And your overhead is close to what it was before COVID. And your income stream is maybe half of what it used to be.

    This might take years. Maybe eventually the freedom fighters will come around. But it will take a lot of time and a lot of pain. Oklahoma? Will it ever get there?”

    To be fair the Closers also ignored things when they closed. like how will you re open.

    Last night Cuomo listed the 10 or questions they will start to ask new positives to answer

    Its may 1.
    NY first case was march 1.
    60 days to develop a list of questions to ask positives?

    There is probably an economic tipping point ahead. Interesting data point
    The category leading the pack in lost revenue is HEALTHCARE.
    Yup hospitals empty. ICU wards full, but high dollar elective surgery cancelled.

    hence hospitals want to “open”

    maybe the food system goes, maybe police, mass transit, power grid..
    boom

    Now here is another coming problem.

    NY will open when say hospital intakes go down for 14 days straight ( as an example)

    So you are a doctor, and you think we should open earlier… hmm don’t tell the patient to check in.
    Or you want to stay locked down, tell every one with a cough to check in.

    Sad to say but every metric folks are suggesting for engineering the opening is hackable.

    And will people go back to retail shops? restaurants?

  338. Steven Mosher says:

    “How many employees were threatened in Korea for trying to get customers to wear masks?

    You’ve been out of the country for too long. Too many gun-totin’ freedom fighter nutjobs here. I’m not talking about the rare megachurch group. I’m talking about most of my neighbors in my rural community. They think that contact racing’s for sissies.”

    oh I dont doubt that it’s different.
    heck even in china you had some resistance.

    No shirt,no shoes no service.
    maybe needs an update.

    Yes contract tracing is for sissies, just like bike helmets and motorcycle helmets and seat belts

    this public humiliation thing is real. I’ve seen bosses line up a whole
    department and give them the screaming in the face treatment

    haha

  339. Steven Mosher says:

  340. Steven Mosher says:

    Everett F Sargent

    of course there will be holes in the system. There are holes in Chinas system… or so I am told.
    there are holes in Korea’s system.. or so I am told.

    So privacy freaks will be able to avoid some surveillience.

    Lets recount how contact tracing works.
    You get sick. New work will set up contact tracing.
    The tracer will interview you and ask for your contacts.
    If you cont have an app or have turned if off, I can still get some of your contacts unless
    you and all your tribe refuse to give contacts.

    Employers can make it voluntary:
    here is our app. use it and you get bonus X.

    So your business has 100 people. 25 use the app. They will enter all their coworkers
    bam, you are captured even if you refuse to use the App.

    When Party B who uses th app enters you as a contact at work, then when you test
    positive B will be contacted. the data doesnt need to be reciprocal.

    The only part that is hard to do is getting where you travelled. But that’s not that critical.

    So in general participation doesnt need to be 100% or even mandatory.

  341. Ben McMillan says:

    One of the annoying things in the UK has been that it looks like cross-infection in the hospitals has been significant. Anecdotally, most of the staff working in Covid wards have either ended up testing positive, or been off sick with typical symptoms. At least some of the medics working on the Covid wards are also working on other wards. And patients with suspected Covid have been moved into non-Covid wards if they test negative.

    Hopefully this means that at least this route to infection is now closed, so that the community infectivity is a bit lower.

    In general there is an interesting property of real world networks (e.g. contacts that lead to infection) that some people are better-connected (come in contact with more people) and act as ‘hubs’ (the idea of scale-free networks comes out of this). Dropping these people out of the network (because they are now immune) can significantly reduce the overall connectivity of the model.

    You can model this in a smooth way by dividing the community into a high-R group and a low-R group with a weak connection. You expect to see in that kind of scenario a bit of herd immunity come into play a bit sooner than in a single-pool model (like the one James is using).

  342. Jim Hunt says:

    Ben – “Anecdotally” front line NHS staff here in the once Great Britain have been, and are still, suffering from a wholly inadequate supply of personal protective equipment.

    That being the case may I commend to you the “Meet NicHolaS” crowd funded project to mass produce protective visors here in North Cornwall:

    https://www.justgiving.com/crowdfunding/meetnicholas

    As for “real world networks” in North Cornwall, please see:

  343. Ben McMillan says:

    Well, there’s been a lot of public/media focus on PPE, but in my opinion not enough on other measures to prevent spread in places like hospitals and supermarkets.

    For example, the medic I know working on a Covid ward has to regularly traverse other areas in the hospital to do their job. There aren’t separate break rooms. Let alone putting up staff members in hospitals for a few weeks. There was no attempt to systematically test other members of the team when one of them was infected. And they were sitting around in break rooms together without full space-man.

    The supermarket stuff also: the best protection is not being there at all. PPE is a poor second best. Getting deliveries scaled up much faster is mega-important. Also, there are a whole bunch of companies producing food in catering-style packs/quantities that could have been mobilised to distribute basics. How many people were infected after making 5 trips in a week to try to obtain toilet paper?

    I was near Barnstaple a week before the official lockdown and was much less worried about contacts in the water than elsewhere (although I was pretty well set up for not having to come into contact with anyone)…

  344. Jim Hunt says:

    Afternoon Ben,

    Living in “SilicInny Valley” on the outskirts of Bodmin Moor Kasia and I always used get our groceries delivered. No such luck any more, since we are not deemed “vulnerable” enough.

    Allegedly people are getting infected on their (weekly?) trips to the supermarket:

  345. Ben McMillan says:

    Hi Jim.
    Yeah, the medic I know can’t get deliveries either. They aren’t vulnerable, but they are much more likely to be infected. Luckily someone else can get their shopping for them.

  346. Steven Mosher says:

    “Allegedly people are getting infected on their (weekly?) trips to the supermarket:”

    I have been once to the local market. It was good……

    1. hand sanitizer at the entrance
    2. Tissues by the carts for wiping the handles of the cart

    THEN I walked around and watched people. So many door handles, picking up packages, putting them back. christ. ah dirty hands everywhere…

    This is crazy. I lived with an OCD buddy n grad school for 2 weeks. It was hell, and now I
    remember everything he did.. it was crazy the constant vigilance is exhausting.
    So I order delivery.
    UNlike china where the delivery person signs and seals the bag and attests to his temperature
    ( it written on the bag) in Korea they dont do that. So …. clean everything that comes in.

  347. verytallguy says:

    “Yes contract tracing is for sissies, just like bike helmets and motorcycle helmets and seat belts”

    Motorcycle helmets and seat belts work.

    Bike helmets – “plastic hats” don’t.

  348. Ben McMillan says:

    https://www.theguardian.com/world/2020/may/03/uk-lockdown-must-not-be-lifted-until-covid-19-transmission-is-understood-say-scientists

    From the article:

    This point was stressed by epidemiologist Anne Johnson at University College London. She said cutting transmissions of Covid-19 to health and social care workers had now emerged as a major priority. “Half of all new infections reported last week were among healthcare workers,” she told the Observer. “This has now become the leading edge of the spread of the disease.”

  349. Joshua says:

    Ben –

    Not just care workers. Grocery store workers, transportation workers, delivery workers etc., as well.

    We all bring something to the math and epidemiology of the pandemic.

    I bring the belief that people who are putting their lives on the line to take care of the ill get extra weighting. As do the people who go to work in capacities that enable vulnerable people to reduce their risk.

    Unfortunately, I fear that in much of this country the risk of infection is pretty abstract. It’s like climate change in that respect.

    Isolating, quarantining, washing hands, social distancing, testing, and tracing are for sissies and libz in government who want to steal our freedoms.

    Sound extreme? Maybe, but maybe not. A straw man? Nope. Check this out.

    https://rosebyanyothernameblog.wordpress.com/2020/05/01/petty-tyrants/

    I hope that in heavily infectwd parts of the country people can maintain diligence, but I fear that the pressure from watching other parts of the country party on will erode compliance and discipline. I hope I’m wrong.

  350. Steven Mosher says:

  351. Joshua says:

    Quail eggs, seaweed, and stir fried anchovies won’t go over so big in Oklahoma, methinks.

  352. verytallguy says:

    Mosh ,

    Dumping links into threads uncommented is rude

    https://pubmed.ncbi.nlm.nih.gov/21376924/

    (I won’t be responding to further on the topic so as not to derail, so you’re most welcome to the last word should you desire it)

  353. Joshua says:

    Willard –

    Did you already post this?

  354. dhogaza says:

    So the IHME “Murray” model has nearly doubled their projected deaths from the first wave of covid-19 in the US.

    Some of this is due to re-opening of businesses etc in some states.

    But buried in their update detail are a couple of interesting statements regarding their “death model”:

    “Since our initial release, we have increased the number of multi-Gaussian distribution weights that inform our death model’s predictions for epidemic peaks and downward trends. As of today’s release, we are including 29 elements, a substantial increase from our original seven and then 13 (which was introduced for our April 17 update). This expansion now allows for longer epidemic peaks and tails, such that daily COVID-19 deaths are not predicted to fall as steeply as in previous releases.”

    “Overall, these modeling improvements have resulted in considerably higher projections of cumulative COVID-19 deaths through August, primarily due to longer peaks and slower declines for locations that have passed their peaks.”

    So their previous parameterized curve fitting algorithm couldn’t generate long, slowly declining tails for daily deaths regardless of the data. Huh.

  355. Steven Mosher says:

    yes dhog, the changes made to IHME are substantial from the modelling side

  356. Steven Mosher says:

    “The supermarket stuff also: the best protection is not being there at all. PPE is a poor second best. Getting deliveries scaled up much faster is mega-important. Also, there are a whole bunch of companies producing food in catering-style packs/quantities that could have been mobilised to distribute basics. How many people were infected after making 5 trips in a week to try to obtain toilet paper?”

    I’ve said it before and it bears repeating.

    In China ( and less so in Korea) Delivery is everything

    In china there are 2 made modes of delivery.

    1. Fedex Like SF express. Except the delivery mechanism is more nimble
    see this
    https://www.alamy.com/stock-photo-a-delivery-courier-from-sf-express-uses-cell-phone-to-contact-client-138009592.html
    2. delivery boys
    https://www.washingtonpost.com/graphics/2019/world/amp-stories/food-delivery-dangers-china/

    of course they describe it as chaotic

    You can have everything delivered for cheap.

    This helped china do a real lockdown.

  357. Steven Mosher says:

    Meanwhile In chicago

    https://www.fox32chicago.com/news/mayor-outraged-over-parties-in-chicago-we-will-arrest-you-and-take-you-to-jail

    Changing workplace rules will be hard and people wont allow surveillience.

    haha

    Petty Tyrants is pretty close to true.

  358. dhogaza says:

    “yes dhog, the changes made to IHME are substantial from the modelling side”

    It goes beyond that. It’s an admission that their previous parameterized death model was shit. Mostly they failed to PR the uncertainties and assumptions involved in their earlier modeling iterations. They have a good PR machine working, though.

  359. Steven Mosher says:

    “Dumping links into threads uncommented is rude”

    I could say making unsubstantiated claims is rude.

    My comment on the publication is simple “FWIW”

    I form no opinion.

  360. Steven Mosher says:

    “It goes beyond that. It’s an admission that their previous parameterized death model was shit. Mostly they failed to PR the uncertainties and assumptions involved in their earlier modeling iterations. They have a good PR machine working, though.”

    The amount of stuff they added was pretty shocking.

  361. Ben McMillan says:

    Agree that the IHME model changes are a huge (rushed?) rethink.

    “We previously implemented a three-day average of the natural log of cumulative COVID-19 deaths to smooth the input data. While this update helped, it did not fully mitigate the effects of volatile input data. As of today’s release, we now apply this algorithm 10 times in a row, which smooths daily death trends for a longer period of time.”

    I mean, I guess that would kinda work, but not exactly deadly-ninja levels of data-science (they do something slightly clever with the last data point, too, but don’t seem to have thought about what happens when you do this ten times).

    I have no idea how their 29 element multi-Gaussian works. Somehow this is also now fed into a SEIR model (an actual dynamical model rather than just a curve fit) that they’ve added in the last couple of weeks.

    Nice that they’ve responded to criticism, but they’ve thrown something pretty complicated and messy together and you have to wonder if it is an improvement. i.e., is it less clearly wrong only because it is less clear?

  362. Ben McMillan says:

    Also, It is worth thinking about what happens when you smooth the log of a curve that goes to nearly zero on weekends… (IHME does account for zeros at least)

  363. Joshua says:

    VTG –

    > “Dumping links into threads uncommented is rude”

    Why?

    I’m with Steven on this.

    Doesn’t seem at all rude to me. You dump a link becsause you think people might find the article relevant or informative. If you don’t find the article relevant or informative, so what?

  364. dhogaza says:

    Moshpit:

    “The amount of stuff they added was pretty shocking.”

    Elsewhere I wrote, just after the model update notes became available, “they have totally rewritten the model. Before they were essentially curve fitting. Now at the heart is an SEIR model.”

    So it’s actually goes beyond adding a lot of stuff. It was toss-and-replace in regard to the portion of the model used to project the number of infected people. They’re still using curve-fitting for their “death model”, which is why they call it a hybrid model, apparently.

    What struck me most, though, was the bit I quoted above in which they’re saying the earlier parameterized function being fit to the deaths data was incapable of generating long tails. Keep in mind that the shorter tails generated due to what was essentially a model artifact were being used by them to project dates by which state governments could safely begin to re-open. They made it clear that they felt governments should be using those projections for planning when to re-open.

    That’s dangerous …

  365. izen says:

    @-Joshua
    “Doesn’t seem at all rude to me. You dump a link becsause you think people might find the article relevant or informative.”

    It is basic netiquette.
    Often the contents of a link can be judged from words within the link. In the case of SM’s link many (depending on browser settings) will have seen the Youtube thumbnail page of the video it linked. It was indeed interesting to see how a civilised Nation is dealing with the pandemic, if a little depressing to reflect on how far otters are falling short.
    But it generally considered polite to give some indication of the content of the link and its source site as the link may be in a form that gives no hint that it is a paywalled site, or a source you have no interest in its content, exchanging cookies, and subsequent targeted adverts.

  366. Joshua says:

    > It is basic netiquette.

    Yah, but I’ve never understood why.

    > But it generally considered polite to give some indication of the content of the link and its source site as the link may be in a form that gives no hint that it is a paywalled site, or a source you have no interest in its content, exchanging cookies, and subsequent targeted adverts.

    Ok. That kind of makes sense

    I guess I’ve never really been ticked off at anyone for dropping off a link that I clicked through on and found uninteresting…but there’s certainly no harm in giving a basic explanation of what’s contained in the link. So I’ll be sure to do so in the future.

  367. Bob Loblaw says:

    “I guess I’ve never really been ticked off at anyone for dropping off a link…”

    I guess I’ve never really figured why someone who can’t be bothered to take the time to at least say why they think the link is worth looking at would expect me to take time to click on it and try to figure it out. If it’s so unimportant to the person posting the link, why would I expect it to be important to me?

  368. Joshua says:

    Bob –

    > If it’s so unimportant to the person posting the link, why would I expect it to be important to me?

    I’ve dropped off links without an introduction merely because I thought the links were interesting or relevant and thought others might find them interesting or relevant also.

    I’ve seen people call it rude but never quite got why. It never bothered me when anyone else did it.

  369. Steven Mosher says:

    “They’re still using curve-fitting for their “death model”, which is why they call it a hybrid model, apparently.”

    At first I thought it was a god rough and ready way of predicting hospital load.
    Start with a curve fit to deaths and you can back out ICU load and general ward
    load. Just a rough and ready planning tool for the hospital system.

    And I defended it n that basis. As time went on it became apparent that they didn’t even have
    the available bed data correct ( for NY at least ) and then the focus became predicting deaths.

    The whole predicting deaths thing is fraught with uncertainty because of the way demographics
    interacts and I am not seeing any of that applying. Plus, there is no acceptable level of death
    Probably best just to focus on a dont crush the hospitals approach. At least thats what I
    thought the plan was. Now however people seem to focusing on a zero deaths
    approach, which simply wont ever happen

  370. Steven Mosher says:

    “I guess I’ve never really been ticked off at anyone for dropping off a link that I clicked through on and found uninteresting…but there’s certainly no harm in giving a basic explanation of what’s contained in the link. So I’ll be sure to do so in the future.”

    I usually make a comment if the link is off topic.
    I am not bothered by folks dropping random links in a discussion. I figure they are
    making a contribution. I click, read, move on.

    Meh, for vtgs sake I can change and indicate why it might be interesting

  371. Steven Mosher says:

    of course I doubted you and read it anyways.

  372. Bob Loblaw says:

    “It never bothered me when anyone else did it.”

    I wouldn’t say it bothers me. I just don’t think it’s worth my time following the link if the person posting it doesn’t think it’s worth their time to suggest why it might be worth my time. It comes across as an attitude that their time is important, but mine isn’t. I can understand why someone might think that attitude is rude.

  373. Ben McMillan says:

    I think the ‘no more deaths’ strategy is pretty good: several countries are at or near this point already (within a month or two). Of course, it is a good destination, but hard to get to from where the UK+US are at the moment. Alternatives look worse: if you can get R<1, and be done in a few months, why settle for R~1 forever?

  374. Joshua says:

    Ben –

    > if you can get R<1, and be done in a few months, why settle for R~1 forever?

    Yes. And so one approach is that modeling could focus on projecting how interventions might get R below 1, rather than in projecting deaths.

    Of course, even there, peple pursuing a political agenda will carp from the bleachers.

  375. dhogaza says:

    Ben:

    “why settle for R~1 forever?”

    Especially when uncertainty in computing R means that R ~ 1 could mean cases will drop or could mean that cases will rise exponentially (with a long doubling time unless the uncertainty is quite large). The difference between up and down is very important. So it would make sense, I would think, to push the estimated value of R well below 1 because even if the estimated R is a bit low, there’s a chance that the actual R is still less than one and you’ll still see case numbers drop over time.

  376. Ben McMillan says:

    Well, I agree that the main important thing is figuring out how to reduce R rather than predict deaths accurately (although “how big a morgue am I going to need” is a valid question). But models largely can’t do that: you need to make comparisons between other countries and other epidemics. Or case studies of how transmissions took place. Mostly this is going to tell you to do the obvious 19th-century stuff.

    Also agree that R<1 is preferable because the risk profile is so asymmetric. I guess it might be reasonably safe to hover at R~1 if you could bring the number of cases down by a factor of 10 at least, so you have time to respond if things start going pear-shaped, but it doesn't seem an attractive option.

  377. izen says:

    @-SM

    The real ‘Ouch’ of that report that 66% of the new cases are not essential workers but are people ‘staying at home’ are the comments posted in response.
    Obviously those people were going out to public places to shop and exercise. The lockdown does not require total isolation, and without the sort of food packages as supplied in S Korea that is not possible.

    But the majority of the responses are that this is evidence that staying at home is useless and that therefore we should cancel the lockdown and all get back to work as soon as possible.

    I do wonder how such commentators manage the cognitive dissonance between the results of the S Korea method of dealing with this which kept deaths in the low hundreds and the UK which is now up in the thirty thousands.
    Presumably there is a model that ‘explains’ it…

  378. Steven Mosher says:

    ““why settle for R~1 forever?”

    why even look at R .
    what matters, we are told, is to not overwhelm hospitals.
    NY has 330K positive cases.
    In reality if we look at serology, there were 2,666,000 cases
    That 2.3 Million people who were positive but never got tested by PCR.
    Actual R is much higher than the R calculated from positive tests.
    What do those people look like? demographically
    next
    About 21% of the 330K required hospitalization. what did they look like?
    About 30% of the hospitalized died.
    The biggest concern of new york was overwhelming the system. Field hospitals were built
    Vents delivered, PPE etc. The system strained but did break.

    About 16 Million new workers are infectable.
    Are they more like the 2.3 million who got infected and resolved without even getting tested?
    or are they more like the 330K who required hospitalization?

    R1 wont tell you anything about the real potential hospital load until you actually
    collect the right data when yu do serology tests. Who didnt get sick, who is apparently
    safe? We know those most at risk for death. Who is most safe? weirdly that data is missing.

    Any way simple point. Suppose the 16 million look much more like the 330 K than
    the 2.3Million. you hospital could get overwhelmed at a lower R0.
    Suppose the 16M look nothing like the 330K, Well then you really dont care.
    You could let it rip.

    Point is the thing that matters–detailed data on who needs the hospital and who doesnt even
    present– more than r0 isn’t known very well. I mean seriously NY tested 15K people for
    antibodies. ~14% were positive. Were they all under 30? all under 40% all thin? all with
    healthy immune systems? who didn’t get sick? why? and what about the 16M who
    remain infectable?

    The logic behind flatten the curve was, we were told, to protect the health system.
    It was critical. We knew and know we cannot reduce death to 0. and since every life
    is priceless we dont want to be calculating “acceptable deaths”. we already accept
    that death is inevitable.It cant be prevented, so we choose to keep the level of infection
    at a level where hospitals are not overwhelmed.

    To do that effectively you need to know about the likelihood of the 16M needed a hospital
    if they catch a case. If they look 1 way, you’d be very concerned about R0 ~1.
    if they look differently, then you’d care a little less about R0

    The symmetry I find odd is the lack of reporting on the characteristics of those who are known
    to be safe.

  379. Steven Mosher says:

    But if you want uncertainties on R0

    https://rt.live/

  380. dhogaza says:

    Moshpit:

    “But if you want uncertainties on R0”

    Do you not know the difference between Rt and Rt, t=0 (R0)

    There is uncertainty for both, of course.

  381. Ben McMillan says:

    SM: “Actual R is much higher than the R calculated from positive tests”: that isn’t how it works. Maybe try making up some curves and doing the calculation yourself. I get the impression that you are confused about what R is.

    “Don’t crush the health system” is important, especially in the early stages of the pandemic, but it is a criterion, not a strategy. It especially doesn’t determine your ‘exit strategy’.

  382. Everett F Sargent says:

    Ben McMillan, dhogaza or anyone else …

    I’ll admin much confusion as to R0 and Rt. So, for example, can one calculate R0 and/or Rt from just the JHU time series? TIA

    I’m thinking an actual epidemiological model is necessary with requisite equations and calculations.

  383. JCH says:

    On my block, during the stay at home: at least 3 parties, including one where they rented a 1960’s convertible that held multiple groups of eight for a drive around the neighborhood (I live in a stunning historic neighborhood;) multiple construction/remodel projects; maids in and out, some more than once a week; frequent shopping trips; no masks. So far, nobody has gotten sick, but case count is just getting large.

    So this why I think temperature is part of what leveled April. Stay at home USA style is a joke.

  384. Steven Mosher says:

    “The real ‘Ouch’ of that report that 66% of the new cases are not essential workers but are people ‘staying at home’ are the comments posted in response.
    Obviously those people were going out to public places to shop and exercise. The lockdown does not require total isolation, and without the sort of food packages as supplied in S Korea that is not possible.”

    I think there is a lot more work to do on these types of surveys.
    It would be nice if they ask the person “can I get your google locations data”
    Just to see, are the “at home” but going out to the store?
    ya know when i first started to see mass delivery in Beijing I thought.. man that will infect people
    One delivery guy could wreck havock.

    i was wrong.

    I’m not so sure they were going anywhere outside their homes. maybe they had visitors?
    maybe their spouse was asymptomatic.

    They need to go deeper.

    yes the comments are bad. I try not to read them

  385. Steven Mosher says:
  386. Steven Mosher says:

    “v“Don’t crush the health system” is important, especially in the early stages of the pandemic, but it is a criterion, not a strategy. It especially doesn’t determine your ‘exit strategy’.”

    really? tell that to the CDC and NY.

    tell it to germany.

    They are all planning exit strategies SUBJECT TO the condition of
    THOU SHALT NOT crush the hospital system.

    hence, in NY they make easing the lockdown subject to a buffer of ICU beds

    NYC has crazier benchmarks and wants to see a decline in deaths and admissions to
    ICU.

  387. Ben McMillan says:

    SM: I think “these exit strategies are subject to the condition of not crushing the hospital system” is a good way of putting it, but for some reason I have the impression you are angrily disagreeing with me.

    Yes, great that Nature paper has a pretty good explanation of R and Rt. Probably there is something a bit less heavy around.

    EFS: Why not try asking James?

  388. Joshua says:

    > “The real ‘Ouch’ of that report that 66% of the new cases are not essential workers but are people ‘staying at home’ are the comments posted in response.
    Obviously those people were going out to public places to shop and exercise. The lockdown does not require total isolation, and without the sort of food packages as supplied in S Korea that is not possible.”

    What a dumb stat. Of course most of those infected were “staying at home,” because there is a “lockdown” going on and the vast majority of people are staying at home!

    Yes, “staying at home” doesn’t mean hermetically sealed off. That’s not possible for many people.

    Ordering in has been a nightmare for groceries. Yes, if 4 people in a family might be able to always stay home, the fifth has to do the shopping or go to work. That person gets infected and likely it means 80% if the infected were “staying at home.”

  389. Joshua says:

    From the comments:

    > The lockdown should apply to age and for those with good fitness in older age brackets could claim a risk assestment based on thier personal wellbeing as I know some 70’s are fitter than some of those in their 40’s.

    What world do these people live in?

    > the cases would have been much worse had they been leaving home. You’d have to be subnormal not to understand the risks.

    Right.

  390. dhogaza says:

    SM:

    “really? tell that to the CDC and NY.

    tell it to germany.”

    Exit strategies vary. CA:

    “On Thursday, state health officials said regions can move further into Phase 2 of the reopening plan when they meet the following conditions:

    –No more than one coronavirus case per 10,000 people in the past 14 days

    –No COVID-19 death in the past 14 days” plus testing and tracing criteria.

    That’s Phase 2 of a four phase plan. Phase 1, starting today, allows some retailers to open and accept orders to be left curbside (as restaurants have been allowed to do all along), but no customers inside and masks, spacing, disinfectant etc for workers inside subject to closure if you don’t play along.

    Of course some rural counties are starting to rebel (though a couple like Modoc may actually meet the criteria) and Newsom’s probably going to have to cave to some degree, but the current exit strategy is virtual eradication here, followed by intense efforts to whack-a-mole new eruptions during Phase 2 and beyond. Not simply “don’t crush the health care system”.

    We are, of course, more than matched by Georgia and Florida etc (Florida has been seeing an uptick of new confirmed cases for a week now). And the non-mask wearing White House itself, with its new case last Wednesday …

  391. dhogaza says:

    EFS:

    Like Ben, I suggest James. If you haven’t read his entire set of blog posts where he started questioning the Imperial College modeling efforts, right through his explaining the development of his own simple (according to his definition of simple) model and putting up a paper preprint, you should. Also his twitter thread on his model and the mainstream modeling efforts by epidemiologists …

  392. Joshua says:

    In NY – what % of the public were “locked down at home?”

    If it was 80%, then what would it mean if 66% of those hospitalized were “locked down at home?” Especially, when you consider that those “locked down at home” may well be more likely to be older.

  393. Jim Hunt says:

    Thanks Steven,

    Vastly more useful than a 100th percentile Josh cartoon!

  394. JCH says:

    No way it was 80%. 80% would have crushed the curve:

    8 weeks of China-style lockdown:

  395. Joshua says:

    JCH –

    It was just as an example.

    My point is that you can’t evaluate that 66% metric unless you have a comparison for what % of people are staying at home.

    Even if 40% of people are staying at home, 66% of those infected staying at home takes on a very different implication than if only 10% are staying at home.

  396. Joshua says:

    Also, JCH –

    Remember that self-report on such surveys is highly unreliable. Especially since ‘stay at home” is such a vague descriptor.

    Conditional probability is hard. Don’t try to draw ANY conclusions from that 66% number.

  397. JCH says:

    Agreed. But the sad thing, which the US has totally blown with its catastrophically bad response to the pandemic, is we could be just about be China right now. Instead we’re screwed for one to three years, and that would be lucky. It could be for decades.

    With the dismantling of the mutual science efforts with China on viral spillover research, the next pandemic could be Avian Flu, and this is walk in the park when compared with the lethality potential there. China has had several Avian Flu outbreaks recently, with the worst killing 60% of the people who caught it.

    We have an intelligence community that does not even know what the answer “we cannot confirm human-to-human transmission” means. The freaking morons think it means China was lying to the world. Good Gawd they’re dumb. Avian Flu, which causes respiratory illness, is primarily transmitted from direct contact with infected birds. But it is also transmitted human to human. If you ask a Chinese scientist if it is transmissible human to human, they’re is going say exactly what an American scientists would say: no. Because the issue, is is it sustainable H2H transmission. To date, despite a large number of deaths from H2H transmission, H2H transmission of Avian Flu goes nowhere. It stays in China. So far.

  398. Joshua says:

    JCH –

    > But the sad thing, which the US has totally blown with its catastrophically bad response to the pandemic, is we could be just about be China right now. Instead we’re screwed for one to three years, and that would be lucky. It could be for decades.

    Korea is another example of how we might have done it.

    Basically I agree. I was thinking this morning, when hearing about the unemployment rate, about the long-term economic implications of us opening up in areas where they haven’t even met the basic requirements of the CDC guidelines.

    I suppose it’s theoretically possible that in rural areas they can limp along economically under those conditions. And it’s theoretically possible that in the larger urban areas they won’t open up unless certain basic infrastructure requirements for testing, tracing, and isolation have been met.

    But I can also see, every easily, how those more rural areas will suffer long term and where the larger urban areas will open up prematurely – and given the % of our national economy that locates in those larger urban areas….

    Yeah – this could be a pivot point in this country’s standing on the world stage. I guess I have some mixed feelings about that – but bottom line it could mean that a lot of people are in for some really bad times. And there’s no way to see that in a positive light.

  399. Willard says:

    Meanwhile, James’ model continues to evolve:

  400. dhogaza says:

    Joshua

    “Korea is another example of how we might have done it.”

    For those who can’t abide Asian examples (not suggesting anyone here does!) we could also be Austria.

    https://www.worldometers.info/coronavirus/country/austria/

    Check out their new cases bar graph.

  401. Joshua says:

    Jesus.

    Wtf did they do in Austria?

  402. Steven Mosher says:

    It pretty much comes down to what we already know.

    1. test.
    2. trace
    3. isolate.

    you wont be able to do these perfectly. If you just test, but dont trace anyone or isolate anyone
    you get new york. If U just test and trace but dont isolate positives you get new york.
    basically if you do nothing on trace and isolate because you can’t do it perfectly, you get
    new york. Sharp rise, flat top, slow decline.. long pain.

  403. Steven Mosher says:

    “–No COVID-19 death in the past 14 days” plus testing and tracing criteria.”

    Putting deaths into the exit criteria is stupid, its a lagging indicator.
    FFS when the put a person on a vent ( basically a 20% chance of survival) that person
    can stay on the vent for weeks without dying.)

  404. Steven Mosher says:

    “Thanks Steven,

    Vastly more useful than a 100th percentile Josh cartoon!

    Oh shit I forgot to make extensive comments about the link I dumped

  405. Steven Mosher says:

    “Also, JCH –

    Remember that self-report on such surveys is highly unreliable. Especially since ‘stay at home” is such a vague descriptor.

    Conditional probability is hard. Don’t try to draw ANY conclusions from that 66% number.”

    The failure of basic data collection and reporting is mind numbing.

    The same goes for serology. Who is designing the data collection systems?.

  406. Steven Mosher says:

    sorry. contains some interesting data on who tests positive

    https://news.yahoo.com/armed-swabs-covid-hunters-stalk-100029759.html

  407. dhogaza says:

    SM:
    “Putting deaths into the exit criteria is stupid, its a lagging indicator.”

    Actually I agree entirely, and have said so on a couple of CA-focused venues.

    I didn’t post it saying I agreed with it, simply to point out that you can’t generalize exit strategies based on NY … or CA, either, of course.

    IMO it should be driven by the new case criteria for CA. We’re so aggressive here that we’re not going to pop off the top of hospital criteria.

  408. Steven Mosher says:

    “IMO it should be driven by the new case criteria for CA. We’re so aggressive here that we’re not going to pop off the top of hospital criteria.”

    I have been struggling to put it in words but I think I agree with you.
    But I Tried to look at in the following way.

    1. you dont want to over run ICU.. say you have 1000 Open beds
    2. you have to pick some Buffer. Like, dont go above 700 beds.
    3. 20% of those going into the hospital will end up in ICU, so
    you dont want more than 3500 in hospital beds full
    4. 20% of those who test positive end up in the hospital.. so… blah blah blah

    anyways after thinking about this thing like it was a machine of buffers and queues……

    5 for every person who tests positive you will tract and test 50.. and you will need xx isolation
    wards…and blah blah blah… blah blah..

    I came to the conclusion.

    can we get a do over for January?

    after thinking I got kinda depressed. not sure anyone knows how to engineer the way out of the mess. I have seen what works.. what works for two specific cultures/geographies . It’s easy to
    argue this and that.

    in any case my visa is up soon, so I have to fly somewhere. not very happy about that.
    suddenly digital nomad is not so cool.

  409. Dave_Geologist says:

    We have an intelligence community that does not even know what the answer “we cannot confirm human-to-human transmission” means.

    JCH, I suspect that (at least some of) the intelligence community know that difference (I presume there are also some Cold War reds-under-the-bed paranoiacs).

    But that politicians are lying about the content of their intelligence briefings. Or to be more generous, don’t understand them or are doing their equivalent: “politicians that do not even know what the answer “we cannot rule out escape from a lab” means”.

  410. dikranmarsupial says:

    Thought this was a rather interesting talk about “scientific machine learning”, i.e. learning from data in a way that builds on the scientific knowledge we have about the nature of the problem. Particularly relevant is the slide at about 6:20, which presents the key components as “respect physical constraints”, “embed domain knowledge”, “bring interpretability to results”, “integrate hetrogenous, noisy and incomplete data” and “get predictions with quantified uncertainties” (the last bit being of particular interest to me, but sadly not greatly explored in the talk).

    ISTR we have discussed similar ideas here before (emulators for climate simulations and galaxy modelling?)

  411. dikranmarsupial says:

    In my hierarchy of

    physics > statistics >= chimps pulling numbers from a bucket

    Sometimes a pure physics model is not feasible (computationally too expensive) or because we don’t know all the relevant physics, but the further to the left we can get, generally the happier I am.

    Sadly, it doesn’t make everybody happy

  412. Everett F Sargent says:

    SM,

    Nice cherry picked graph, although I still believe the US to be 3rd world status (one red, the US, surrounded by 20 green, then surrounded by 21 yellow, in a six-by-seven pattern = 42 countries).

    Note the misleading implication given by showing only one red country, replace with UK or SE or BR or CA or … about fifty other red countries. Cherry picked = RED! Nope, no irony there at all …

    I can pick red too, but at least I’ve combined the US with the rest of (3rd) world (or RoW – black line, note as of now CA is in RoW, so that RoW+US includes most of NA, Mexico is also in RoW).

  413. Joshua says:

    Plus, charting cases (without considering rates of testing) is misleading.

  414. Joshua said:

    “Wtf did they do in Austria?”

    I was in Austria on January 20 for the Euro championships and there were many super fans from the far east packing the arena. At the time China had less than 100 new cases a day. So they essentially dodged a bullet with the timing of the event.

    Also, almost submitted an abstract to the EGU for April which is held in Austria every year. Didn’t but I did attend the virtual conference that was rescheduled and finished up just yesterday. From the feedback, lots of people were disappointed with it. Some of it was maddening (like the Q&A often was limited to a few minutes per abstract) but overall very useful.

  415. Everett F Sargent says:

    I just know there is a relationship between deaths and confirmed cases …

    Well almost, like 3+ orders of magnitude for a single death (zeros are excluded here, but that is 0-3785 or ~3.5 orders of magnitude) and like one order of magnitude for ~1000+ deaths. So that we can only make order of magnitude arguments using confirmed cases. But we already knew that.

    Florida stopped testing, therefore zero confirmed cases, but Florida still has COVID-19 deaths, therefore > infinite mortality rate. Florida sez spontaneous human combustion is the real cause.
    https://en.wikipedia.org/wiki/Spontaneous_human_combustion
    Small Hands sez not a conspiracy and not a hoax.

  416. Everett F Sargent says:

    “3785” should be “653” for zeros or < three orders of magnitude (wrong (sum of zeros) column, go figure). The 1,1 value is for the Diamond Princess (I consider it an error in their database), there are 14 deaths total 13 in JP and one transferred to AU before death.

  417. JCH says:

    “Earth probably harbours a million times more virus particles than there are stars in the observable Universe. …

    But, so far, just 4,958 virus species have been formally described.”

    Classify viruses — the gain is worth the pain

  418. Joshua says:

    I see that bogus 66% number hit WUWT. It will be interesting to see how many “skeptics” completely fail this test of due skeptical diligence. So far only one commenter has passed the test – but my guess is that he isn’t a “skeptic.” I left a comment but doubt it will get past the “censorship.” Someone call a whamublance.

    Joshua Your comment is awaiting moderation.
    May 9, 2020 at 3:29 pm
    First, self-report data on something like this is notoriously unreliable.

    Second, the definition of “stayed at home” is way to meaningless to be of any value.

    Third, unless you know what % of NYers were “staying at home,” the 66% doesn’t really mean anything at all. The implications of that 66% would be completely different depending on whether 80% of NYers are staying at home, or 40%, or 10%.

    Henry Pool raises this point above. What’s amazing that any self-styled “skeptics” wouldn’t catch this immediately

  419. Steven Mosher says:

    Here is comes again

    silent super spreader.

    http://www.koreabiomed.com/news/articleView.html?idxno=8190

    “Chung called for the health authorities and local governments to mobilize all means to trace about 1,900 people who were at the clubs in Itaewon at the same time with the possible super-spreader and test them. He told officials to keep the clubbers’ names and other personal information classified so that they can voluntarily take the test.”

    there is a back story here. its the gay clubs

  420. Joshua says:

    Itewon, huh? Could easily mean spread to the American military base.

  421. Joshua says:

    Steven –

    You’ll like this

    https://twitter.com/i/status/125 787843570380801

  422. Joshua says:

    Oops.

    Trying again.

  423. Steven Mosher says:

    Yup Joshua that’s the chinese.

  424. Steven Mosher says:

    “Meanwhile, James’ model continues to evolve:”

    yes Willard. it is interesting to watch other scientists defend themselves against James by appealing to consensus.

    I got a link to the SAGE consensus documents. it is quite disappointing.
    Basically look like a summary of the news

    I don’t want to bash the epi guys. Must be a horrific job. for years and years you just play with numbers with no global pandemic data to test your model. Then bam.

  425. dhogaza says:

    SM:
    “I don’t want to bash the epi guys. Must be a horrific job. for years and years you just play with numbers with no global pandemic data to test your model. Then bam.”

    I’ve been reading the relevant Ferguson paper. First of all, total deaths is really kind of a byproduct of the paper and the model built as described in the methods. They were exploring the geographical and temporal spread of an avian flu jumping to a human host and mutating or recombining in a human that can transmit human-to-human. in order to try to model effectiveness of rapid intervention with tamiflu and social distancing to stomp it out before it became a pandemic.

    “How can we stop it?” not “if we sit on our ass or do half-assed measures, how bad will it be?”

    Obviously (at least AFIAK) this didn’t happen, at least in a way to let them test their model results.

    So while I respect what James Annan has been saying about model calibration, my impression is that the kind of data needed wasn’t really available before this all happened.

    Now I also think Annan’s right about the fact that they held on to a low value for R0 and the resulting doubling time after it obviously was off. Slow to react is my first reaction. Maybe the data was really there early on and they should’ve reacted a couple of weeks earlier, but seriously, I’m guessing this landed in Ferguson’s lap like a grenade with the pin pulled.

    The team working on the current, partially refactored and cleaned up version that’s on github, state that they run it frequently and as data becomes available, use it to calibrate the model. While the source is available, thus far at least the latest model projections aren’t made public. I could be wrong, of course about the projections not being made public.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.