The 3DDRR edged up again, to 1.53. Excluding New York, it rose to 1.54, from 1.47 on Monday and 1.50 yesterday. I would say at this point that there is no definite trend, whereas before yesterday it was clearly trending down.
24 thoughts on “The worst is not really behind us”
it’s possible that it is effectively step wise, and that different communities started distancing earlier…
I think it’s fair to say that the spread seems to be peaking now. Most expect the daily confirmed cases (which is lagging) to peak roughly next week, although that will vary by location. Where I am, the peak is expected in early May.
So it’s probably too early to call “the worst is behind us” but we’re almost there. And then the question will be: now what?
Be nice if Arnold updated his 3DDRR post with the list.
I understand all the talk about R (infection growth rates), but it seems to me we are barking up the wrong tree, since we have a naive population in a novel cold virus.
R may go up or down, but when the lockdowns end it will have to go up until some large portion of the public is infected, the experts seem to say 50% to 70%.
I still think a do-nothing policy would have been best, perhaps with voluntary sequestering of the elderly and throw lots of money at hospitals. If the government is to do anything, then Robin Hanson’s ideas about deliberately infecting a large portion of the population, perhaps in exchange for a payment, is a good idea.
Some have pondered if the left-wing or the right-wing or libertarians or conservatives or liberals will come out ahead in this . From what I can tell, some prominent libertarians became statist martinets about two pitches into the first inning of the ballgame. Oil drillers in Texas are talking to the Texas Railroad Commission again, and others want President Trump to connive with OPEC and Russia to artificially inflate oil prices. The roll-call of industries seeking bailouts is too long to catalog.
Well, perfidious pettifogging passes for principles.
For reasons that befuddle me, in general libertarians did not call for large-scale tax holidays and listening to scientists who say (sotto voce) the only way to get over this is herd immunity.
Growth in the number of new cases peaked and has been declining since March 19
Number of new cases probably peaked on Saturday, April 4. The growth rate in deaths seems to track the growth rate in new cases, with a 6-8 day lag. There’s a reasonable chance that number of new deaths will peak in about a week.
Death reporting is a lot more lumpy. IHME (the Bill Gates funded model) notes: “Based on the now multiple iterations of our COVID-19 death model, we have noticed that, for at least some US states, there are massive fluctuations in the number of COVID-19 deaths reported each day. These substantial day-to-day vacillations are more likely due to an artefact in how statewide deaths are being compiled and then reported each day than actual fluctuations in COVID-19 deaths. ”
JPMorgan Halts All Non-Government Guaranteed Small Business Loans
—-
There is no Keynesian free lunch. The bank lenders will not compete against government and cover the seigniorage tax which will likely hit 100 billion or more. The only way they can avoid the tax is to roll up commercial lending and deposit fund in excess reserves until the seigniorage gain is gone. I keep mentioning this and I keep hearing assumptions that the ‘lower bound’ is not a problem.
When retail banking rolls up, it is replaced by shadow banking, and the Fed no longer observes prices with any accuracy, there will be one more step in the process of finding prices. You get one of Hayek’s bulging triangles. Congress cannot budget, they have no information about future government prices as their is the shadow bank interpretation that needs doing. Nore can Congress predict taxes with any accuracy.
PSST shows the problem, banking solves the sustainable pattern problem, if we do not have visible banking we get very erratic goods flow, unsustainable.
If Minnesota is sending out this kind of instruction re: death certificate entry, then how do we know other states are not doing it as well? Or reporting other fake data?
If models are using stinky data like this, then how can anyone trust a model’s output?
Erroneous data, lag time in reporting, lack of general testing of the population — how can anyone pretend to make considered policy, or think they see a trend, based on this junk information?
If my thermometer always read 6 degrees too high, and it tells me the noon temperature has gone up 10 degrees, from 68 yesterday to 78 today, I know that the noon temperature has gone up 10 degrees, from 62 to 72.
In general, one should plan for the worst, and hope for the best. When unknown dangers approaches, the risk of being cautious is usually lower than the risk of being careless.
When data is junk, you can still use models, but the error bars range from terrible to lucky. So, because you have no choice but to act now, act like it’s going to be really bad, but then invest a lot in gathering actionable intelligence, making your information a lot better, as quickly as possible.
Once you have good information, if you played it too safe, you can always adjust fire and relax. If you played it too loose, it’s too late, the cat’s out of the bag.
With a good early warning system, you can nip problems in the bud. With eyes wide shut and garbage-in, garbage-out, now you’ve got to go to war, with the Army that you have.
Unfortunately, the Army that you had turned out to be total crap.
The problem is that it may be impossible to know what is “playing it safe”. What looks like “playing it safe” may be very dangerous long-term, or if you look at both “the seen and the unseen”.
“Playing it safe” may simply be going with whatever the most common prejudice is.
I suppose there are theoretical circumstances of completely unprecedented dangers for which all bets are off.
But when the problem is contagious, it’s not impossible. Humans have literally millennia of experience dealing with such things, and the safe approach until you know better and can do better is always isolation of individuals and groups.
If the bright spot be white in the skin of his flesh, and in sight be not deeper than the skin, and the hair thereof be not turned white; then the priest shall shut up him that hath the plague seven days. And the priest shall look on him the seventh day: and, behold, if the plague in his sight be at a stay, and the plague spread not in the skin; then the priest shall shut him up seven days more.
I completely agree: isolate “symptomatic individuals”. But Americans are doing a lot more than that.
As always, we are on our own. There is no expert knowledge. The best defense is quarantining oneself from the manipulators in the media, academia, and government. This has been demonstrated again and again. 9/11, the banking crisis, Iraq and Afghanistan, ad nauseum. Covid is only the latest display.
And now, one of those researchers has caught the attention of Bloomberg by announcing that it may have underestimated the virus’s velocity by half, meaning it’s been spreading twice as quickly through China – and will therefore likely follow (or has followed) a similar pattern in the US.
This set of projections was produced by researchers at Los Alamos, BBG said.
New assumptions produced by the team including the average number of people infected in the early days of the epidemic in Wuhan: they have been revised to 5.7, more than twice the number the WHO has projected.
5.7 means one person infected 5.7 in the crowded commerce of the city over the ‘early days’. How long is an ‘early days’ and how crowded was it in the early days. I have one person spreading to 15 in a crowded bridge tournament over eight hours. So this must be three trips to the market, 2-3 hours a trip, over three days. Unimpeded, then over over a 30 day period NYC is 100% infected and more. An impossible boundary condition, and we have already passed the mark,March 1 they started, it is two weeks into through April. Something is missing or the derath rate is very low or the shutdown very effective. I think the solution is we get the high spread rate from a select class of high spreaders who need to deal with people in mass, including transit. These were the first thing \shut, and these folks should now be immune, for a while.
“Journalists: please STOP reporting on weekend COVID19 numbers as a ‘drop’ or ‘flattening’. In every city & state MasksForDocs has worked, there are reporting delays due to short-shifting on Saturdays and Sundays.”
– Masks For Docs Foundation https://twitter.com/MasksForDocs/status/1247565862107623424
“This is EXACTLY what @andreafeigl1 has analyzed and been predicting for weekend numbers. Always a weekend reporting lull. Don’t oversensationalize weekend drops!”
– Eric Feigl-Ding https://twitter.com/DrEricDing/status/1247566172276559873
This suggests that 7DDRR might be a better metric to account for weekly seasonality.
If the reported numbers say the worst is over, many people will decide to go out. If the reported numbers say it’s getting worse, more will decide to stay in. I would expect a certain amount of automatic stabilization due to risk homeostasis.
what is the name of the rule that states that any statistic used as a benchmark becomes useless?
The Lucas Critique is one – but I recall there being other names as well?
“Goodhart’s law states that once a social or economic measure is turned into a target for policy, it will [eventually] lose any information content that had qualified it to play such a role in the first place. … The law was named for its developer, Charles Goodhart, a chief economic advisor to the Bank of England. [“Originally, an economic theory stating that if a particular definition of the money supply were to be used as the basis for monetary policy, the stability of its statistical relationship with spending on the economy would break down and the policy would prove ineffective.”] …
The most famous examples of Goodhart’s law should be the soviet factories which when given targets on the basis of numbers of nails produced many tiny useless nails and when given targets on basis of weight produced a few giant nails. Numbers and weight both correlated well in a pre-central plan scenario. After they are made targets (in different times and periods), they lose that value.”
So, I’ve been working a little bit on numbers in my own state and beyond the widely reported under-counting of deaths, there’s also a consistent problem that “deaths reported” on April 9 is not the same as “deaths which occurred” on April 9.
3DDRR would really need the second to be accurate, but my observations of the data is that it’s frequently getting the first.
What’s happening in Indian, at least, is that deaths are reported for a day, get fed into the big statistical aggregations, and then, later, the state is quietly updating the daily deaths numbers. So what happens is there’s a consistent bias. Yesterday always looks worse at first, then yesterday gets spread out into yesterday plus another 5-7 days.
To follow this up, here is the 3DDRR calculated for Indiana using COVIDTracking.com vs. the Indiana Dept. of Health
20200330 IN 1.46 1.60
20200331 IN 1.58 1.41
20200401 IN 2.03 1.48
20200402 IN 2.23 1.41
20200403 IN 2.08 1.32
20200404 IN 1.78 1.23
20200405 IN 1.63 1.21
20200406 IN 1.36 1.21
20200407 IN 1.49 1.17
20200408 IN 1.60 1.11
So what’s happening in reality is pretty encouraging. But you have to be careful with the data.
it’s possible that it is effectively step wise, and that different communities started distancing earlier…
I think it’s fair to say that the spread seems to be peaking now. Most expect the daily confirmed cases (which is lagging) to peak roughly next week, although that will vary by location. Where I am, the peak is expected in early May.
So it’s probably too early to call “the worst is behind us” but we’re almost there. And then the question will be: now what?
Be nice if Arnold updated his 3DDRR post with the list.
Experimental results:
https://reason.com/2020/04/08/mass-antibody-testing-in-this-rural-colorado-county-sheds-light-on-covid-19s-prevalence-and-lethality/
Mass Antibody Testing in This Rural Colorado County Sheds Light on COVID-19’s Prevalence and Lethality
—-
They have a high death rate of .5% and a low of .1%, so there are many more infected out there then we measure, according to this. But only 900 out of 8,000 has been tested, they intend to complete the whole sample. This is a rural, sparsely populated town.
I understand all the talk about R (infection growth rates), but it seems to me we are barking up the wrong tree, since we have a naive population in a novel cold virus.
R may go up or down, but when the lockdowns end it will have to go up until some large portion of the public is infected, the experts seem to say 50% to 70%.
I still think a do-nothing policy would have been best, perhaps with voluntary sequestering of the elderly and throw lots of money at hospitals. If the government is to do anything, then Robin Hanson’s ideas about deliberately infecting a large portion of the population, perhaps in exchange for a payment, is a good idea.
Some have pondered if the left-wing or the right-wing or libertarians or conservatives or liberals will come out ahead in this . From what I can tell, some prominent libertarians became statist martinets about two pitches into the first inning of the ballgame. Oil drillers in Texas are talking to the Texas Railroad Commission again, and others want President Trump to connive with OPEC and Russia to artificially inflate oil prices. The roll-call of industries seeking bailouts is too long to catalog.
Well, perfidious pettifogging passes for principles.
For reasons that befuddle me, in general libertarians did not call for large-scale tax holidays and listening to scientists who say (sotto voce) the only way to get over this is herd immunity.
Growth in the number of new cases peaked and has been declining since March 19
Number of new cases probably peaked on Saturday, April 4. The growth rate in deaths seems to track the growth rate in new cases, with a 6-8 day lag. There’s a reasonable chance that number of new deaths will peak in about a week.
Death reporting is a lot more lumpy. IHME (the Bill Gates funded model) notes: “Based on the now multiple iterations of our COVID-19 death model, we have noticed that, for at least some US states, there are massive fluctuations in the number of COVID-19 deaths reported each day. These substantial day-to-day vacillations are more likely due to an artefact in how statewide deaths are being compiled and then reported each day than actual fluctuations in COVID-19 deaths. ”
As an example, there’s a big lag in how Louisiana reports deaths: https://twitter.com/Crimealytics/status/1247893989849624578
I’ve noticed that the growth rate in deaths is lower than expected on Sundays and Mondays, with some “catch up” on Tuesday and Wednesday.
https://www.zerohedge.com/economics/just-how-bad-it-going-get-jpmorgan-halts-all-non-government-guaranteed-small-business
JPMorgan Halts All Non-Government Guaranteed Small Business Loans
—-
There is no Keynesian free lunch. The bank lenders will not compete against government and cover the seigniorage tax which will likely hit 100 billion or more. The only way they can avoid the tax is to roll up commercial lending and deposit fund in excess reserves until the seigniorage gain is gone. I keep mentioning this and I keep hearing assumptions that the ‘lower bound’ is not a problem.
When retail banking rolls up, it is replaced by shadow banking, and the Fed no longer observes prices with any accuracy, there will be one more step in the process of finding prices. You get one of Hayek’s bulging triangles. Congress cannot budget, they have no information about future government prices as their is the shadow bank interpretation that needs doing. Nore can Congress predict taxes with any accuracy.
PSST shows the problem, banking solves the sustainable pattern problem, if we do not have visible banking we get very erratic goods flow, unsustainable.
I don’t know how one draws any kind of worthwhile conclusions when the data is suspect. Watch this interview (~7 mins.)
https://www.powerlineblog.com/archives/2020/04/how-honest-is-the-covid-fatality-count.php
If Minnesota is sending out this kind of instruction re: death certificate entry, then how do we know other states are not doing it as well? Or reporting other fake data?
If models are using stinky data like this, then how can anyone trust a model’s output?
Erroneous data, lag time in reporting, lack of general testing of the population — how can anyone pretend to make considered policy, or think they see a trend, based on this junk information?
If my thermometer always read 6 degrees too high, and it tells me the noon temperature has gone up 10 degrees, from 68 yesterday to 78 today, I know that the noon temperature has gone up 10 degrees, from 62 to 72.
In general, one should plan for the worst, and hope for the best. When unknown dangers approaches, the risk of being cautious is usually lower than the risk of being careless.
When data is junk, you can still use models, but the error bars range from terrible to lucky. So, because you have no choice but to act now, act like it’s going to be really bad, but then invest a lot in gathering actionable intelligence, making your information a lot better, as quickly as possible.
Once you have good information, if you played it too safe, you can always adjust fire and relax. If you played it too loose, it’s too late, the cat’s out of the bag.
With a good early warning system, you can nip problems in the bud. With eyes wide shut and garbage-in, garbage-out, now you’ve got to go to war, with the Army that you have.
Unfortunately, the Army that you had turned out to be total crap.
The problem is that it may be impossible to know what is “playing it safe”. What looks like “playing it safe” may be very dangerous long-term, or if you look at both “the seen and the unseen”.
“Playing it safe” may simply be going with whatever the most common prejudice is.
I suppose there are theoretical circumstances of completely unprecedented dangers for which all bets are off.
But when the problem is contagious, it’s not impossible. Humans have literally millennia of experience dealing with such things, and the safe approach until you know better and can do better is always isolation of individuals and groups.
I completely agree: isolate “symptomatic individuals”. But Americans are doing a lot more than that.
More of the same here: https://www.americanthinker.com/articles/2020/04/the_cdc_confesses_to_lying_about_covid19_death_numbers.html
As always, we are on our own. There is no expert knowledge. The best defense is quarantining oneself from the manipulators in the media, academia, and government. This has been demonstrated again and again. 9/11, the banking crisis, Iraq and Afghanistan, ad nauseum. Covid is only the latest display.
https://www.zerohedge.com/geopolitical/new-projections-show-virus-spreading-twice-fast-expected
And now, one of those researchers has caught the attention of Bloomberg by announcing that it may have underestimated the virus’s velocity by half, meaning it’s been spreading twice as quickly through China – and will therefore likely follow (or has followed) a similar pattern in the US.
This set of projections was produced by researchers at Los Alamos, BBG said.
New assumptions produced by the team including the average number of people infected in the early days of the epidemic in Wuhan: they have been revised to 5.7, more than twice the number the WHO has projected.
5.7 means one person infected 5.7 in the crowded commerce of the city over the ‘early days’. How long is an ‘early days’ and how crowded was it in the early days. I have one person spreading to 15 in a crowded bridge tournament over eight hours. So this must be three trips to the market, 2-3 hours a trip, over three days. Unimpeded, then over over a 30 day period NYC is 100% infected and more. An impossible boundary condition, and we have already passed the mark,March 1 they started, it is two weeks into through April. Something is missing or the derath rate is very low or the shutdown very effective. I think the solution is we get the high spread rate from a select class of high spreaders who need to deal with people in mass, including transit. These were the first thing \shut, and these folks should now be immune, for a while.
“Journalists: please STOP reporting on weekend COVID19 numbers as a ‘drop’ or ‘flattening’. In every city & state MasksForDocs has worked, there are reporting delays due to short-shifting on Saturdays and Sundays.”
– Masks For Docs Foundation
https://twitter.com/MasksForDocs/status/1247565862107623424
“This is EXACTLY what @andreafeigl1 has analyzed and been predicting for weekend numbers. Always a weekend reporting lull. Don’t oversensationalize weekend drops!”
– Eric Feigl-Ding
https://twitter.com/DrEricDing/status/1247566172276559873
This suggests that 7DDRR might be a better metric to account for weekly seasonality.
If the reported numbers say the worst is over, many people will decide to go out. If the reported numbers say it’s getting worse, more will decide to stay in. I would expect a certain amount of automatic stabilization due to risk homeostasis.
what is the name of the rule that states that any statistic used as a benchmark becomes useless?
The Lucas Critique is one – but I recall there being other names as well?
https://en.m.wikipedia.org/wiki/Lucas_critique
Perhaps you are thinking of Goodhart’s Law.
“Goodhart’s law states that once a social or economic measure is turned into a target for policy, it will [eventually] lose any information content that had qualified it to play such a role in the first place. … The law was named for its developer, Charles Goodhart, a chief economic advisor to the Bank of England. [“Originally, an economic theory stating that if a particular definition of the money supply were to be used as the basis for monetary policy, the stability of its statistical relationship with spending on the economy would break down and the policy would prove ineffective.”] …
The most famous examples of Goodhart’s law should be the soviet factories which when given targets on the basis of numbers of nails produced many tiny useless nails and when given targets on basis of weight produced a few giant nails. Numbers and weight both correlated well in a pre-central plan scenario. After they are made targets (in different times and periods), they lose that value.”
So, I’ve been working a little bit on numbers in my own state and beyond the widely reported under-counting of deaths, there’s also a consistent problem that “deaths reported” on April 9 is not the same as “deaths which occurred” on April 9.
3DDRR would really need the second to be accurate, but my observations of the data is that it’s frequently getting the first.
What’s happening in Indian, at least, is that deaths are reported for a day, get fed into the big statistical aggregations, and then, later, the state is quietly updating the daily deaths numbers. So what happens is there’s a consistent bias. Yesterday always looks worse at first, then yesterday gets spread out into yesterday plus another 5-7 days.
To follow this up, here is the 3DDRR calculated for Indiana using COVIDTracking.com vs. the Indiana Dept. of Health
So what’s happening in reality is pretty encouraging. But you have to be careful with the data.
Thanks for alerting me to this.