“Stop me if you’ve heard this one before: Ivy League law professor writes a deepthoughts think piece explaining a seemingly irrational behavior that doesn’t actually exist.”

Under the heading, “Stop me if you’ve heard this one before,” Palko writes:

An Ivy League law professor writes a deepthoughts think piece explaining a seemingly irrational behavior that doesn’t actually exist.

(see here and here)

My favorite bit is this, from the Ivy League law professor in question:

What’s more, macroeconomists have typically spent their careers preparing to understand and respond to crises in the economy. They are acutely attuned to the grave dangers associated with an economy grinding to a halt.

What the hell does that mean? “Acutely attuned to the grave dangers”? One thing we can say for sure, I guess, is that this guy got a good score on his verbal SAT. Also “was a clerk to U.S. Supreme Court Justice David Souter.” So he’s got good connections.

64 thoughts on ““Stop me if you’ve heard this one before: Ivy League law professor writes a deepthoughts think piece explaining a seemingly irrational behavior that doesn’t actually exist.”

  1. What’s even more funny is that if we had proper contact tracing we wouldn’t be having this argument.

    Overall though, epidemiologist or economist – and especially politician – it hasn’t been a good year for expertise. Lots of egg masks for Halloween.

    • There’s a scale at which contact tracing becomes ineffective, right? And given that only 30% of positive cases even provide any secondary contacts in the US, I’m not sure it would work even if cases were low in the US.

      • “The time for contact tracking was Jan/Feb. It has been pointless since then”

        I was going to agree but I changed my mind. The value of contact tracing now is that it gives us many people to study to find out how long people are contagious and learn more in general about how the infection is transmitted.

        I’m not signing up for an app, but if someone called me and talked to me I’d probably be happy to participate.

        • Theres been data on that around for awhile, havent seen it change much since early papers from china. And of course it requires a lot of assuming about who transmitted to who.

          Best to just do a challenge study in healthy young people at little risk and be done with it.

      • Really? I want to be called if I’ve been in contact with an infected person. Also with someone with active TB or Ebola. I don’t want to infect my family, friends or others, not to mention I’d like to get a test before I become symptomatic. I don’t know why people are talking about it like it’s for a research study.

        • Based on the number of cases and estimated missed cases the avg person has probably “contacted” hundreds or thousands of infected people so far. If it was accurate youd be getting spammed with multiple calls each day. But it wont be accurate and will just stress you out for no reason.

        • I don’t think so. I don’t know what data you have on how many contacts with people they can name that the “average” person has normally, not to mention in the current restricted environment. (If I’ve had traceable contact with 100 people since March I would be surprised.) I live in a place often described as the epicenter. All through February and early March we at my institution had to do do-it-yourself contact tracing, telling students in classes when a fellow student or faculty member was infected and even then it was not thousands or even hundreds. And that’s high density — most people don’t have clumps of 20-60 people grouped in multiple classrooms to notify. People who don’t live in urban areas simply do not have thousands of nameable contacts in a month and, really, neither do those in urban areas.

        • All through February and early March we at my institution had to do do-it-yourself contact tracing, telling students in classes when a fellow student or faculty member was infected and even then it was not thousands or even hundreds.

          Were asymptomatic people being tested?

        • I don’t know, but the U.S. government won’t stop looking for any excuse to do it, no matter how uninteresting we are

          https://en.wikipedia.org/wiki/PRISM_(surveillance_program)

          If I were in charge in March, I would’ve just paid people for their contact tracing data. Some people would say no, but that’s fine, we didn’t need everyone, just enough to keep the reproduction number down. Be a good use for all that QE money.

      • Contract tracing has been quite effective where I live in NW California, and where most people are pretty good about wearing masks and social distancing. 65% of cases have been linked to a known local infected person, and another 20% were travel-related. we have knocked back three distinct surges in infections, the largest and most recent last month. From the beginning, on average, fewer than 2% of 30+ thousand tests have given positive results.

        • So something like 29,000 people there were suspected cases? What did they have to cause the symptoms and why arent you worried about those pathogens?

          I really wonder whats triggering the million people per day in the US to get tested that mimics covid.

        • Lots of people are just getting tested without any symptoms… For example Univ of Illinois Urbana Champagne is requiring everyone on campus to get tested twice a week.

          Then there are probably employers like care facilities, where they have to test the entire staff a couple times a month… then there are factories where they’re testing people… and there are people who are tested because their friends tested positive… I’d guess less than 10% of the people getting tested have symptoms.

        • Typical interval between testing for people being regularly tested is probably 15 to 30 days, but the distribution probably has at least several hundred thousand who get tested every 4 to 12 days (hospital workers, college students on campus, etc) plus a random selection of people getting tested because they’ve been identified as potentially in contact with a patient, plus a small number of people with symptoms. The positivity ratio is consistently less than 10% most places, so it’s probably substantially less than 10% of people who have symptoms.

        • I think that the Health Dept offers testing to people identified as exposed through the contact tracing, whether they are symptomatic or not. Given the percentage of infected people who are asymptomatic, it would be dumb not to.

        • If it was all contacts, itd mean ~98% of “contacts” (however they define it) test negative. In other words 98% percent of the calls could be considered false positives.

          In reality some of those people called dont show up for a test and some positives were symptomatic people who showed up at the hospital.

          Id guess that less than 1% of these calls were really necessary. Even if the call identifies a positive, there is only ~10% chance they would have transmitted it to anyone unknowingly (most transmissions are superspreader events).

          So something like 0.1% of the calls really did any good.

          And of the people who do get infected down the line, ~.01% will die. The back of the napkin math gets more complicated due to the multple transmissions but point is probably < 0.01% of these calls save a life.

          The NNT for this intervention is about 10,000. Perhaps there are better ways to use those resources.

        • If its about 5 minutes per call and contact tracers get paid $20 per hr. Thatd would be a minimum cost of 50k man hrs and $17k per life saved. Then the tests are about $50 each so add in another $50k for a total of $77k per life saved.

          Then theres the other infrastructure and support staff, etc that I dont know how to estimate. But anyway the dollar cost for this is on the order of $100k per life saved.

          That could fund a lot of useful vitamin c/d tests/treatments, hbot chambers, pulse oximeters, etc for people we know need it wothout unnecessarily stressing out 10 thousand othet people and making them miss work, getting injured somehow by the swab…

        • Sorry 50k + 17k = 67k obviously. I also apologize for the typos at the enf, but it should still be understandable.

        • Ah, made another error switching between percent and proportions. Basically its like a drake equation, plug in whatever numbers you think are accurate.

          Lspt = lives saved per test
          Pp = percent positive
          Pt = percent of positives that transmit to at least one other person
          Nt = Avg number of new positives per transmission event
          Pd = percent of cases that die

          Lspt = Pp x Pt x Nt x Pd
          NNT = 1/Lspt

          I get:
          Lspt = 0.01 x 0.1 x 10 x .001 = 0.00001
          NNT = 100,000

          Then just +/- an order of magnitude gives a range of 10,000 to 1,000,000.

          Anyway, I dont see anyone making the argument about whether to devote resources to widespread contact tracing or testing in general of non symptomatics based on rational cost benefit analysis like this.

          I doubt you can plug in reasonable values and get an NNT under 1,000. Thats probably why its never been considered worth doing before besides in subgroups like nursing home employees.

        • Regarding transmission chains theres 1 – (1- Pt)^Nt = 1 – 0.9^10 = .65 or 65% chance that one of those new infections lead to a superspreader that in turn infected 10 people.

          In general (1 – (1 – Pt)^Nt)^Nc, where Nc is the length of the chain. Then the probability the chain continues for 2 sets of new infections is the solution for Nc = 2. Thats 42%. So on average the transmission chain will end after only 20 total infections with ~ 1% chance it gets to Nc = 10, for 100 total infections from the original detected by contact tracing.

          So divide the NNT (or multiply the Lspt) by a factor of 2 or at most 10 to adjust for transmission chains.

          Hope it all makes sense, glad to hear of any errors.

  2. The great unreported thing here, in the quest to “Listen to Science” (and you have to capitalize it or you’re a flat-earther or something) is that neither epidemiologists not economists are qualified to decide what to do. Epidemiologists know (or at least purport to know) how to stop the spread of disease. Economists know (or at least purport to know) how to efficiently make tradeoffs between different social states. What neither has any expertise in (even purported expertise) is in evaluating and choosing between social (non-market) states. That’s what politicians are supposed to be able to do, but they’ve all abdicated their responsibility to do so in favor of posturing and, much more importantly, criticizing each other. The ability to do this credibly is so much more needed than either another epidemiological recommendation or some macroeconomic forecast and in such short supply that it is quite clear that it isn’t Science that’s the problem. As Walt Kelly said: It’s us.

    • There is a specialty in public health called Health Policy which connects cost of doing and not doing, together with a host of efficiency issues. One of my daughters is getting her PhD in this.

      • Such an expert can *advise* a politician, perhaps even better than an economist or epidemiologist… I don’t know. But the fundamental decision the politician has to make is a value judgment which can’t be made by any expert…. not even a pollster.

        • Actually, if you do a good job, a polling company could easily make value decisions far better than the average politician, if by “better” you mean “aligning best with the actual values in the community”.

        • One fundamental problem of polling is whether you are aggregating preferences (something you seem to be hinting at) or information.
          You could poll people about whether lockdowns are too harsh or too lenient.
          People who say lockdowns are too harsh might think the fatality rate of COVID is ~0.1%
          while people who say lockdowns are too lenient might think the fatality rate of COVID is ~1%.
          when of course, the truth is somewhere in between.
          The challenge being that it is difficult to tease out to which extent people’s information shape their preferences and vice versa. (I suspect the latter has a larger effect than I would like.)

        • +1. Exactly. Listening to people’s stated preferences is not even close to aggregating them properly in many, many situations, though it works fine in others. Polling results in this regard are like the advice of Science: useful but not dispositive.

        • Doing a good job would mean eliciting something that corresponds to a utility function, not something that corresponds to a judgement about facts.

          So “are lockdowns too harsh? strongly agree, agree, not sure, disagree, strongly disagree” is just totally the wrong kind of question… instead something like:

          “Suppose that 10000 people between ages 50 and 70 will die next week of COVID, how much should we spend on a hypothetical plan guaranteed to reduce that number to 5000 people?”

        • Both of those are exclusively about voting in an ordinal system (where preferences are purely ordered not of a given cardinality/size). But the point I was making was you can construct a kind of social average utility function from people’s responses to questions about how much they prefer one vs another solution to things (ie. how many dollars they’d like to allocate to one vs another for example)

          Neither theorem has anything to say about such issues.

        • Would that lead to the nightmare of unrestrained democracy? I wonder! What if the majority voted say to let all the vulnerable fend for themselves and open everything up?

          If you let the masses vote on utility functions you may be surprised what sort of policy descisions come out of it!

        • ah yes, can’t have the unwashed masses trying to govern themselves! The natural order is for the owners of capital and land to decide what’s best for everyone. Why fix something that ain’t broken?

        • The problem with constructing a social utility function is that it tends to make policy decisions an engineering issue, rather than a social issue. The fact is that people’s preferences differ – differ greatly – and the way I want to see policy issues resolved is through meaningful dialogue and civil discourse (I know, how quaint). Constructing a social welfare function, while very useful for analysis purposes, obscures the very dialogue we are sorely missing.

        • Dale, I chose to write up my analysis of each of the current CA ballot propositions on my Facebook. I had multiple meaningful discussions with friends and family because of it. If a polling company sent out 20 questions related to social utility to 1M people, and individuals discussed those questions with their friends and family to come up with answers, I suspect we’d have about a thousand times more meaningful dialog than today

        • Also, the problem with *not* constructing social utility functions is that all the debate becomes about which of the small number of proposals we should choose. But most people probably have little capacity to correctly estimate the real effect of any proposal. So people choose things on the basis of “someone I like told me it was a good thing” or “it makes me feel good to be doing something about X”

          If we focus on what people want to achieve, rather than the highly imperfect means to achieve it that are proposed by politicians, we could then have policy groups whose whole purpose was to come up with proposals that actually have a chance of achieving what most people want. Or at least we could have our debates be about “which of these proposals is most likely to produce the most socially desirable outcome” rather than “Politician X says vote for his plan Y”

        • Daniel said,
          “If we focus on what people want to achieve, rather than the highly imperfect means to achieve it that are proposed by politicians, we could then have policy groups whose whole purpose was to come up with proposals that actually have a chance of achieving what most people want. Or at least we could have our debates be about “which of these proposals is most likely to produce the most socially desirable outcome” rather than “Politician X says vote for his plan Y””

          Rahul replied,
          “The imperfection may be a feature not a bug.”

          My first thought was to agree with Rahul. But on second thought, I’m not so sure — especially since something on my agenda for today is to decide how to vote on propositions for an upcoming “election”. What so often seems to be the case is that the proposers of propositions say things like, “This will satisfy this group’s desire for a better transportation system”. But different groups may have different criteria for what is better — and, worse, some politicians may use some pretty “creative” thinking to back up such claims. Or their proposal may throw a crumb to what “most people” want, while it gives them something that is a lucrative boost to some business enterprise of theirs. For many people, the price of the crumbs can be very high.

        • Martha, that’s exactly my point. There are a ton of terrible laws which are on their surface written “to address problem X” but whose main effect is to line the pockets of some group, or just screw up the way things work for everyone except X, or whatever.

          Suppose there’s a concern that transportation in a certain region is not very good… instead of having some damn fool lawyer write up a law, which would sound like “if you are between the ages of 0 and 13 you pay X, and between 13 and 18 you pay Y and 18 to 55 Z and 55+ you pay P… except on holidays and days when the local baseball team is playing the rides will be free, and we’ll allocation an enormous $Q dollar bond to upgrading stations and adding more trains…”

          which “addresses the affordability issue” but 3 years later the transportation district is bankrupt because it can’t cover its own operating costs given the too-low fares, and all the bond money was spent lining the pockets of politician’s friendly local construction company…

          Instead we could ask people questions about how often people plan to ride the train based on different fares, and how much money the local government should allocate to subsidize fares vs having the riders pay their own way, and how much money we should spend on upgrading roadways, etc etc… maybe 10 to 20 questions, maybe not the same question for each person, but rather a random sample of 20 out of 100 questions on each questionnaire…

          And then publish some dataset, and offer a prize to design a new transportation policy, complete with analysis of costs, ridership, social utility, and soforth, as well as a prize for auditing the analyses and providing counter-arguments. finally the top 5 proposals are chosen and the county uses score voting to score the proposals and chooses the one with the highest score.

          Gotta be better than what we have.

        • At the limit then, why not just have a direct democracy. Everyone votes on everything. Get rid of the intermediate layers of elected representatives.

          What use are representatives in Daniels model?

        • I have no answers, only questions. The issue of how to make these decisions is a fundamental challenge to our system. Daniel’s solution seems technocratic to me – it appeals to the analyst in me, but I think it is profoundly anti-democratic. At he other extreme, direct democracy has numerous problems in a complex modern world. The public is not in a position to understand the complexities and design and vote on appropriate policies. While I cherish democracy, I am reminded constantly about what an imperfect system it is. We need something better – policy alternatives should not suddenly appear in front of voters and we need only see which they prefer. We also can’t expect to elicit their utility functions and then use a decision making algorithm to choose (not if we want to avoid handing so much power to the technocrats). So, the only way forward seems to require a more educated populace, serious about civil discourse. Good luck with that.

        • Dale, notice that my proposed system is neither technocratic nor direct democracy but rather a system where we elicit preferences, we engage the analysts and engineers to create proposed solutions, and also to shoot down those solutions and then we engage the voters through score voting to choose one of the proposed solutions (with “do nothing” always an option on the ballot).

          I honestly think there is no role for representatives. However there is a strong role for the judiciary.

          I’d like to point out the importance of score voting. It’s a vastly superior voting system, it expresses preference cardinality, and Arrows theorem doesn’t apply. It tends to produce decisions thought to be reasonably good by almost everyone. None of this “the majority decides” meaning 51% of people eat the other 49% for lunch

        • I like your proposal, Daniel. I would gladly seek citizenship in such a place. Unfortunately, I don’t see it as realistic or even feasible – in my lifetime or yours. That doesn’t mean we shouldn’t discuss it, or even pursue trying to implement it. But let’s not confuse a good idea with reality. I think we were closer to what you are suggesting 200 years ago than today. Sad, but that’s my glass 0.24684 full.

        • Dale, agreed that it’s nowhere near happening… Also agreed that it doesn’t mean we shouldn’t discuss it.

          The biggest frustration I’ve had throughout my life is to listen to people saying things like “democracy is terrible except for all the other systems” and then you simply lay out a way better system, and they immediately shoot it down as infeasible.

          The biggest problem with my proposal is that no-one has ever heard of it, or thought about how things might be made REALLY better. There are tons of proposals to patch this or that little corner of what we’ve got. It’s like a computer program that’s supposed to run an entire country, but it’s mostly bugs that compensate for other bugs.

          Anyway, maybe I should write a book. The time is ripe for people to discuss how our current system is broken, considering that Trump pretty much threw it on the ground and stomped on it.

        • >>Anyway, maybe I should write a book. The time is ripe for people to discuss how our current system is broken,

          Yeah, probably.

          I don’t think your specific proposal is likely to succeed, because IMO it’s just not going to be convincing to most people (who don’t think in terms of utility functions etc.) The readers of this blog are not a random sample of the population ;)

          But fairly dramatic change is likely possible in the near-to-medium term (say 8-12 years, two or three presidential election cycles) as the US is probably going through another party realignment with the breakdown of the “status quo” that has held since ~1970.

          I don’t think amending the Constitution is likely to be feasible, so the basic representative system will remain. But lots of procedural changes are possible without that. The US government works radically differently than it did in 1930, and most of the changes aren’t related to constitutional amendments.

        • Sometimes a utopian vision is useful just as a way for people to reset how they think about what we have. The people doing the hard work of patching the existing govt can always look at what might be, and try go get closer to that…

  3. Andrew noted:
    One thing we can say for sure, I guess, is that this guy got a good score on his verbal SAT. Also “was a clerk to U.S. Supreme Court Justice David Souter.” So he’s got good connections.

    In addition to a good score on his verbal SAT, he managed to get (1) Rhodes Scholar, (2) PhD from Oxford in oriental studies, (3) Junior Fellow at Harvard, and (4) chair at Harvard Law.
    You can find his bio at the Harvard Law School website.
    Bob76
    (I’ve never had any contact with the guy and I’m not aware of any mutual friends.)

      • I was hounded (by the other kids) in 1st grade because my coloring went outside the lines. That classroom teacher took their side too in the matter. I can still hear it, and this is verbatim, They: “What’re your drawing there?” Me: “It’s a design”, They:”Teacher! Teacher! He says he’s drawing a *design* again! He’s drawing outside the lines!!”

  4. “What the hell does that mean? “Acutely attuned to the grave dangers”?”

    Acutely attuned to the grave danger that their stock portfolio could drop by 30 percent, and not so attuned to the danger that people might end up in their grave.

    • ” not so attuned to the danger that people might end up in their grave.”

      From what I understand the risk of ending up in one’s grave was pretty high even before covid, and tragically unresponsive to policy.

      Just looking quickly at the COVID fatalities per capita, there are several “advanced” countries ahead of us that have had more stringent lockdown policies. I wonder what the correlation of COVID fatalities is with, say, national obesity rates or other conditions. We know for sure the US is one of the fattest nations on earth. It also has very high rates of chronic disease in elderly people compared to other countries.

  5. Andrew,

    Having graduated from the law school on Morningside Heights, I hold no brief for Harvard Law School. But, your snide comments about Noah Feldman are unjust and unkind.

    I only know of Noah Feldman through his testimony and articles. Several months ago, Noah Feldman made a very fine presentation before the Senate Judiciary Committee on the Kavanaugh nomination. His articles in The New York Review of Books are generally excellent. They are analytical and well reasoned.

    One further point: You get to be a Supreme Court clerk on merit alone, not connections.

    • Michael:

      1. Fair enough. I could well believe that this person has done lots of great stuff. Everyone has their bad days, and we shouldn’t judge all his work based on one bad think piece that he wrote.

      2. I think connections help too in getting that clerkship. But I was really saying that the clerkship is itself a connection.

    • -not so sure about connections not playing a role in clerkships–see the Chua kid getting a Kavanaugh clerkship after her mom came out in favor of his nomination for example
      -In a similar vein, Feldman rushed to write an oped supporting the ACB nomination…the usual “despite being a liberal” business…doesn’t say much for his judgment…

    • To push this discussion further: if it’s true that this law professor writes analytical and well reasoned articles for New York Review of Books but also writes hack op-eds, I don’t think that’s such a good sign. I’ve seen this before: academics who are careful in their academic work but sloppy when writing for general audiences, and it doesn’t make me happy.

  6. I, being an epidemiologist, heard a different but similar theory from another epidemiologist. The theory goes like this:

    Medical professionals have a different response to the pandemic depending on whatever they are doctors or epidemiologists. Doctors think only in the numbers of life saved, no matter the cost, so they advocate for the complete lockdown. Epidemiologist think in terms of relative risks, years of life lost, costs etc. So they advocate for a lighter measures.

Leave a Reply to Jonathan (another one) Cancel reply

Your email address will not be published. Required fields are marked *