Posts
Comments
I originally named the types of knowledge "Type 1", "Type 2", and "Type 3", but was encouraged by early reviewers to actually name them. In light of the conversation here, I think doing that was a mistake. Unless I was sure my names were out of the park correct (and maybe not even then), I should have left it generic so I could get input on more people for what the names, and for that matter definitions, should be.
A thing I really structured to capture was that "i did actual research and had actual models for why masks would help against covid, but it's still not type-3", which is why "know why" doesn't feel right to me.
I tentatively think that some of what you're calling engineering knowledge would fit into what I call scientific (which is a strike agains the names), and/or that I didn't do a good enough job explaining why engineering knowledge is useful.
Ignoring the labels I put on them, do you feel like you have a good sense of what I mean by each kind of knowledge? if so, what would you label them?
Welp, I did not make that deadline. Unfortunately the conditions that led me and the LW team to miss that deadline- high opportunity costs- are not likely to change soon, so instead of holding out for perfection I'm just going to share a couple of thoughts.
I was brought on to lead covid research efforts at LW as an experiment. The hope was that there was significant untapped research capacity, which could be unlocked by providing some structure (hence the research agenda). The structure was not only supposed to give people a sense of what would be useful to research, but reassurance that their research would actually be used, and social reinforcement. This mostly did not pan out- I think I did useful research during the time in question, I think other people produced useful research during that time, but questions I asked tended to be answered by only me.
The experiment was well worth running, and the team got a lot of information on infrastructure useful to support coordinated research (most notably it led to some reworks of Questions). But after 6 weeks it was not achieving its stated goal and had not found something clearly high value to pivot to, so I called it.
I think "prevent" is a strong word but broadly agree that my original approach was not conducive to deep model building. You can read more about how my thoughts have changed over time here.
Ah, I have only read Deep Work.
Cal Newport was specifically who I was thinking of when I said "people who don't understand how useful social media is to me." And I still think he doesn't, I'm just now more aware of the cost as well.
It's pretty wide ranging and I don't have time to find all of it, but off the top of my head
- Jim's Question on long term effects
- The Diamond Princess Data (both initial death rates and the subsequent finding that half the asymptomatic cases had severe chest x-ray abnormalities)
- A baseline expectation that it would be weird for something to be 10x worse in one population and not at all worse in another
COVID-19 isn't that bad for the vast majority of young and otherwise healthy people (no worse than a bad flu).
I disagree with this and disagree that it is the consensus. The risk of serious sequelae is uncertain but potentially quite high- certainly much worse than the flu.
All else being equal you don't want to use antipyretics because fever is beneficial (to a point). Aspirin might reduce your chance of stroke but increase the strength of the infection.
I specifically looked into the smaller recessions and not the Great Depression because it's just too weird. I think you might be right that this one will approach it, but before I look into the Great Depression I want to check out the 1970s oil shocks, which share with the present the common feature that they were caused by an actual thing changing, rather than financial tools misfiring. The Great Depression is also just so well known relative to the smaller recessions (although the Great Recession also has a lot of material on it).
Hey everyone who is following this closely- I've been sprinting madly for the last six weeks and hit my limit. You can expect a retrospective post and perhaps a phase 2 agenda in the next few weeks, but for now I am resting.
http://orbis.stanford.edu/ is an amazing customizable map showing transportation costs across the empire and for routes you choose in particular. You can set travel mechanisms, season, transfer costs, etc.
Article claims beer and soda are endangered by the reduction in driving. CO2 comes mostly from ethanol production, which is on the decline.
Double checking that you've seen this comment on the supply chain thread (which deals with produce)?
I've talked about simulacra levels with Ben a ton and this comment is the single most helpful thing in understanding them or explaining to others.
Not a lawyer, repeating the words of someone else who's not a lawyer.
I talked to my bank (a 3 state credit union) today, specifically asking whether I counted as my own employee. The representative implied the SBA itself didn't know, and if they did they hadn't shared the information yet. She also very strongly implied that as long as I filled out the forms in good faith and my supporting documents were accurate, I would not be punished for fraud.
I think 30X for initial viral load might be too high, because it's comparing symptomatic cases of smallpox to all cases of smallpox. The fair thing is to compare variolation against everyone who was exposed to smallpox, whether they developed symptoms or not.
Internet seems divided on whether or not subclinical smallpox even exists, so this probably isn't a very large factor. I write it out mostly so if someone else has the same idea they don't have to duplicate work.
Where do you get that the cats were sampled from shelters and hospitals? I see
Consistent with the high neutralizing titer, the owners of Cat#4, Cat#14 and Cat#15 were diagnosed as COVID-19 patients. Cat#1, Cat#5∼9 was from pet hospitals, while Cat#2, Cat#10∼13 were initially stray cats and kept in animal protection shelters after the outbreak.
So hospitals and shelters were certainly part of the sample, but it seems like they also tested some human patients' cats.
Trends in divorce continue basically unabated
This paper, the only long-time-scale survey I could find, reports a minor negative correlation between unemployment rates and divorce. However looking at their graph, the relationship is obviously mild.
For posterity: I expected there to be a large, detectable drop in divorce rates during recessions and then a spike as soon as the recession ended.
Some states (Washington, California) already have mail-in voting available for everyone, I don't think there are any federal legal barriers to expanding it.
Religious attendance is down already, ideally to zero. What are the long term effects of that?
Brainstorm:
- Permanently weaker social ties, especially among the elderly and rural.
- -> More depression, suicide, addiction
- Longer time to notice that someone has died and for the body to be discovered (for suicides or otherwise).
- Families that rely on church assistance go hungry, leading to permanent physical and mental deficits for children.
- Churches receive fewer donations (both because of lack of physical presence and the recession), some are forced to close.
- I wanted to make a numerical prediction for this but couldn't find numbers on # of churches over time except for LDS, which has been growing.
- I expect LDS to make it through this pretty well, possibly even strengthened, because they've led their members to prepare for this and because I expect them to institute social reinforcement over distance.
- Speed up of the transition to digital services. Gains reduce but do not entirely disappear when social distancing is lifted.
- Membership takes a permanent hit, faster than counterfactually.
Religious Service Attendance Stays Flat
I was really surprised to find a single academic paper in the last 40 years on religiosity and economic conditions, which was not available online. It reports a "strong" countercyclic effect in religious participation in evangelical Protestants but procyclic effect in mainline Protestants, in the 2001 recession. Meanwhile a Pew poll and a Gallup poll show no change in religious participation during the 2008 recession.
For posterity: I'd predicted an increase in attendance.
- A laboratory study (preprint) showed that covid could live in cat and ferret respiratory systems, but not dogs, pigs, chickens, or ducks. It further found that covid could be passed to from an infected to an uninfected cat in an adjacent cage (doesn't look like they tested transmission in ferrets).
- A survey (preprint) of Wuhan cats indicated 15% of cats surveyed after the outbreak had antibodies to covid. 0 of 39 cats caught before the outbreak had antibodies. (ETA 4/9: They don't mention what they were sampling from, from what they imply I think 15% is an overestimate but still enough to establish the possibility).
- At least one zoo tiger has test-confirmed covid, and several big cats at the zoo are showing symptoms
So it seems pretty likely cats are vulnerable to covid, and may be able to pass it to humans.
Prediction: Births will decline precipitously (BOTEC: 20%-60%).
A normal recession sees a drop in birth rates of ~9%, although that is typically made up mostly of delays rather than entirely foregone children. Due to fear around interaction with the medical system, I expect it to drop much more than that.
BOTEC: ~40% of births in the US result from unplanned pregnancies. If no one took any additional precautions due to covid and everyone who was planning a pregnancy chose to postpone, that would decrease births by 60%. In reality I expect some "unplanned" pregnancies to be planned out of existence as people take more precautions, and some people to plan pregnancies even given the circumstances (disproportionately older women whose fertility window is running out, although births using fertility treatments will decline), but 60% is still a good upper bound.
I expect at least as many people to prevent pregnancy due to covid as prevent pregnancy during a recession, so there should be a minimum of 2x as many foregone or delayed births. With rounding, that's a 20% floor.
People die a little less often, especially in nursing homes.
Note: data is for the United States only
Deaths go down during recessions; according to Ruhm 2002, a 1% decrease in the unemployment rate is associated with an average 0.4% rise in total mortality (about 13,000 deaths, relative to the average of ~2.8m). This is counterintuitive, because wealth is associated with longevity (e.g. Chetty et al. 2016) . There were a lot of potential explanations for this centering on how work was dangerous and didn’t leave time for health, but it turns out most of the additional deaths are concentrated among groups that were unlikely to be employed in the first place, such as those over 70 (70% of the total) or under 4. Fewer than 10% of the additional deaths occur among those between the ages of 25 and 64 (Stevens et al 2011).
Why does employment of working-age adults have such an impact on elderly mortality? Stevens et al make a compelling case that it’s because widespread unemployment increases the relative number of people willing to take unpleasant, low-paying nursing home jobs, particularly entry level “aide” positions, and this improves care of residents.
For posterity: I googled this before making a prediction, but do not believe I would have predicted the results.
I'm curious why you don't expect this to bounce back when the pandemic or recession is over?
Fixed, thank you.
This is fantastic. Can I encourage you to make it a top level post?
Do you know if outside-the-home includes hospital transmission? That could skew things severely.
Found my source.
What regulators spotted a couple of years ago is that banks were buying very focused packages of insurance that would pay off in exactly the scenarios of the stress tests. They had no commercial reason whatsoever, and were in fact probably quite expensive pieces of insurance to purchase. But it meant that the bank could, with a really straight face say, 'Well, you know what? In this stressful scenario we'd be totally fine.' And what's going on under the table is, 'Yeah, because our bet that this scenario would happen would pay off, and we'd suddenly get an extra half a billion extra dollars.'
I had misremembered the insurance as cheap.
Having savings is boring and unleveraged. Buying insurance against an *extremely specific* scenario does even better on the tests and costs much less.
My impression was that if anything these tests were making it worse, but I don't have the reference to double check.
A journalist suggests (and I find credible) that some shortages are due to demand moving from business-to-business to business-to-consumer supply chains much faster than supply can. E.g.,
- Business toilet paper has different fiber content, the rolls are much larger, and many more rolls are sold at a time.
- Restaurant/cafeteria bananas are small and sold in units of 150, consumer bananas are larger and sold in bunches of < 10.
Presumably these could be retooled eventually, but whether that's even a good idea depends on how long the shut down lasts.
EDIT 4/6 : At least one B2B provider is retooling to allow consumers to purchase from them (not an endorsement, I have not used the service).
I dug into this here and found at least a quarter of US produce harvesting is mechanized, which makes me less nervous.
Effect of Economic Downturns on Fertility
The effect of economic downturns on births is surprisingly complicated. On one hand, people have less money and kids are expensive*, which you would expect to lead to fewer children. On the other hand, a reduction in employment expectations reduces the opportunity cost of children, which you would expect to lead to more.
For the rest of this article, I will by default be referring to WEIRD countries.
Based primarily on Economic recession and fertility in the developed world and spot checking its sources, my conclusion is that modern recessions temporarily decrease per capita births, but by and large do not change cohort fertility (i.e. women have the same number of total children they would have had without the recession, but later). Some trends:
- The reduction in births is seen mostly in younger women (20-24), not older women (30-34), suggesting this is a voluntary decision incorporating knowledge of ability to have children in the future.
- The effect is much larger for first births than subsequent births, suggesting this may be more about union formation than post-union decisions to have children (this could also explain the age-related effects)
- The change seems to be driven more by change in situation than by absolute status, i.e. there isn’t a strict relationship between per capita GDP or unemployment and fertility that holds across countries, but in countries where children and women have the same status, people will react similarly to a change in circumstance.
- Male unemployment is universally bad for fertility.
- Female unemployment depends on the era (used to be positively associated with fertility, now is negatively) and on a woman’s socioeconomic status (richer/better educated women’s fertility is more procyclic than poorer/worse education women’s).
- Generous unemployment insurance or non-employment-linked maternity benefits unsurprisingly raise the birth rate during a recession.
Specific numbers are hard to give because every country, demographic, and recession is different, but as an example, this article estimates ~9% decrease in fertility in 2013 in the US.
* This is in societies where children are economic sinks. In situations where they are assets, you would expect the reverse.
My full notes are available here.
Thanks to Eli Tyre for research assistance.
For posterity: I'd predicted a decrease in fertility but hadn't put numbers on it.
Note that Prime Now, Amazon Fresh, and Amazon Pantry are all different services with different availabilities, and I can often find things on Prime Now that aren't on Fresh (haven't even tried with Pantry). Delivery slots haven't been abundant but can typically be found within 2-3 days.
I think you're being too modest, but I've removed it since you think it's been eclipsed by something better.
Suicide rates rise, primarily in unemployed men
I linked to this study above, but for completeness:
A review found that out of 38 studies:
- 31 of them found a positive association between economic recession and increased suicide rates.
- 2 studies reported a negative association,
- 2 articles failed to find any association
- 3 studies were inconclusive.
Unfortunately they didn't share the effect size for most of these studies. Looking at other sources (notes here), I found anywhere from a 4% increase (across Europe and the Americas during the 2008 recession) to 60% (among men in Russia during the 1991 crisis). Studies typically found a much larger effect in men than women, sometimes finding no change in the female suicide rate at all. Different studies found different effects on different age groups; these felt too subdivided to me and I ignored them. Unsurprisingly, unemployment was positive correlated with suicide.
That 60% increase in Russia corresponded to an additional 30 deaths per 100,000 people per year, at a time when the overall death rate was 1300 deaths per 100,000 people. That 4% Europe/Americas increase represents 5000 deaths total, across three continents.
For posterity: I'd predicted a rise but not the amount or demographics.
My understanding is the stress tests for big banks aren't very good- they encourage the bank to shore up against exactly the scenario they know will be tested, and no other. So I'm not very hopeful about them for grocery stores.
Produce will have trouble being picked because farms frequently rely on temporary immigrant workers, and the necessary immigration is banned. This is already happening in Europe, and seems likely to affect the US shortly. Even if it were not, migrant worker conditions seem pretty rife for disease transfer, so I'd expect a big hit to productivity even if they did show up.
Long term I expect this to shift consumption to things that require less human touch in harvesting, like grains and legumes.
ETA: Apparently Canada thought to exempt migrant farmworkers from the travel ban, but with air travel shut down is having much the same problem.
ETA 4/2: This paper words things in a confusing way, but: in 1997, 75-80% of vegetable-acres and 55-60% of fruits-acres grew a type of produce that has at least a 50% chance of being harvested mechanically. That's helpful for knowing what could be mechanically harvested in a labor shortage, but I assume it's too late for farmers to change plans this year. So the actual range (in 1997) was somewhere between 37.5-80% and 37.5-60% of vegetables and fruits, respectively, and has presumably gone up since.
Reasons this number could be wrong:
- Mechanization likely to have increased in the last 20 years
- Mechanical harvesting doesn't actually mean zero labor, someone has to run the machine.
- It's likely that some amount of labor will be found, one way or another
My conclusion is that in the US, if labor shortage is the only problem this year variety will fall, prices will rise, and there will be more emphasis on processed produce, but it's not likely to be a crisis.
One example: (I think, but have not hardcore verified, that) recessions tend to noticeably reduce births. This ends up being really great for the kids who were born into sufficient money, because when they mature there's less competition for top colleges, jobs, etc.
Interesting. My impression is that recessions can start many different ways but typically kick off the same self-reinforcing cycle, so that unhappy economies end up being very alike as well.
Economics Explained argues that I'm wrong, and this recession was caused by high leverage on mediocre bets, just like the last two.
UChicago professors use survey data to estimate 1/3 of jobs can be done at home, accounting for 44% of payroll. See here, scroll down to "How Many Jobs Can Be Done At Home?"
There are already different strains identified, seems plausible some are worse than others.
What percentage of the economy is "essential business?", according to most shelter-in-place laws?
Ballpark: 40m, out of 160m total.
I made a spreadsheet, using the CA definition but numbers for all of the US. It's not perfect- some industries I couldn't find at all (communications is the biggest blank space), the divisions used by the statistics don't map perfectly to the divisions used by the CA definitions, most numbers are from 2018, etc. But it's a start.
I suspect there should be numbers somewhere we can look for how many people were laid off in the first 2 weeks,
3.3m people filed for unemployment insurance in America in the last week. The previous record was 700k in 1982, when the population was 230m
Early Appearance of Symptoms
Tl;dr COVID-19 might start with a cough, or a fever, or both, or occasionally maybe neither. It might start suddenly or slowly. It might remit and then come back worse.
This was hard to figure out. Most academic/medical papers start with the person’s first contact with the medical system, which is too late, so I looked at social media and news reports. These are obviously going to be biased towards people with symptoms severe enough to be interesting, but not so severe as to die. I also restricted myself to test-confirmed cases, which because I was also looking at mostly American sources biases things towards severe cases. And I’m counting on people to represent themselves honestly. So there’s a lot going against this sample.
In total I found 11 cases, plus two notes from doctors doing front line work. You can see my and Eli Tyre’s notes on every case here and the symptom tracking spreadsheet here. Both are messy and no work has been spent making them comprehensible to others.
From this very small and biased sample:
- 36% of people started with a cough on their first day (55% if you count two people who had very mild symptoms on day 1 and developed a cough on day 2)
- 64% started with a fever.
- 18% of people started with both on the same day.
- 18% started with neither symptom (but developed a cough on day 2)
- 78% eventually developed a cough
- 91% eventually developed a fever. The only person who didn’t eventually develop a fever I think might be a false positive, because his symptoms were very weird.
- 27% had digestive symptoms (mostly nausea)
- ⅓ of recovered people had been hospitalized.
On the other hand, this pre-print found that only 44% of patients had a fever on admission to the hospital, and 88% ever had one. The fevers can be intermittent. Both that paper and the doctor I just linked to report very high occurrence (85%+) of lymphocytopenia (low white blood cell count), but that is not very useful to diagnosing yourself, and it’s not clear when it starts.
Known scientific source Business Insider says that most cases start with fever. They do not mention their source.
Commonly Repeated Things I Don’t Believe
You may have heard that 80% of cases are mild. Keep in mind that that paper defined mild to include mild pneumonia, which I would classify as at least moderately severe.
Similarly, the paper reporting 50% of patients appeared at the hospital with digestive symptoms counted “loss of appetite” as a digestive symptom. Only 19% presented with a more definitely gastroenteric complaint like nausea or vomiting.