Nudgelords: Given their past track record, why should I trust them this time? (Don’t call me Stasi)

An economist who wants to remain nameless sent me an email with subject line Hilarious and with the following text:

https://www.amazon.com/Averting-Catastrophe-Decision-Potential-Disasters/dp/1479808482

The link is to a listing for a forthcoming book, Averting Catastrophe: Decision Theory for COVID-19, Climate Change, and Potential Disasters of All Kinds, by Cass R. Sunstein. The book “explores how governments ought to make decisions in times of imminent disaster.”

OK, fine. So what’s Sunstein’s record on giving advice so far?

“The state of New York, with help from Cornell’s Brian Wansink, puts nudge principles to work in its cafeterias. . . . The U.S. Department of Agriculture is planning to award $2 million in research grants for federal food policy inspired by behavioral economics.”

Actually, no, it seems that Wansink faked the data. But, hey, nudges!

“Knowing a person’s political leanings should not affect your assessment of how good a doctor she is — or whether she is likely to be a good accountant or a talented architect. But in practice, does it? Recently we conducted an experiment to answer that question. Our study . . . found that knowing about people’s political beliefs did interfere with the ability to assess those people’s expertise in other, unrelated domains.”

Actually, no, their study said nothing at all about doctors, accountants, or architects. They just made that part up to make it sound more relevant.

“Recent evidence suggests that capital punishment may have a significant deterrent effect, preventing as many eighteen or more murders for each execution. This evidence greatly unsettles moral objections to the death penalty, because it suggests that a refusal to impose that penalty condemns numerous innocent people to death. . . . The widespread failure to appreciate the life-life tradeoffs involved in capital punishment may depend on cognitive processes that fail to treat ‘statistical lives’ with the seriousness that they deserve.”

Actually, no, the data don’t show that each execution prevents eighteen murders (or, as Sunstein and his collaborator charmingly put it, “as many as eighteen or more”). But, hey, anyone who disagrees with him must just be failing in their “cognitive processes.”

“Candidate for coolest behavioral finding of 2019: If a calorie label is on the left of the relevant food item, it has a much bigger impact than if it is on the right.”

I’m more on the side of Elizabeth Nolan Brown, who wrote:

We might disagree on whether federal authorities should micromanage lunchroom menus or if local school districts should have more control, and what dietary principles they should follow; whether the emphasis of school cafeterias should be fundraising or nutrition; or whether school meals need more funding. But confronting these challenges head-on is a hell of a lot better than a tepid consensus for feel-good fairytales about banana placement.

But I’ve saved the best for last. It’s from 28 Feb 2020:

“At this stage, no one can specify the magnitude of the threat from the coronavirus. But one thing is clear: A lot of people are more scared than they have any reason to be. . . . Many people will take precautionary steps (canceling vacations, refusing to fly, avoiding whole nations) even if there is no adequate reason to do that. Those steps can in turn increase economic dislocations, including plummeting stock prices.”

Can you believe those selfish Americans? Canceling vacations, leading to plummeting stock prices? Let’s think of the real victims here: innocent stockholders, whose lives get upended by silly normies who fail to appreciate real-life tradeoffs:

“The underlying problem goes by an unlovely name: ‘meta-cognitive myopia.’ The basic idea is that people are highly attuned to ‘primary information’ . . . By contrast, we are less attuned to ‘meta-information,’ meaning information about whether primary information is accurate. . . .”

As I wrote when I first encountered this particular quote: it’s always good to get a spoonful of jargon with our factoids. The “unlovely” jargon serves a similar function as the stuff in the toothpaste that gives you that tingly feeling when you brush: it has no direct function, but it conveys that it’s doing something.

But don’t worry, just buy this new book. The nudgelords will save us from ourselves. But don’t worry:

“A simple measure of presidential performance takes account of just two variables: approval rating and the Dow. The argument for APDOW, as we might call it, is that public opinion matters, because it captures the wisdom of crowds, and that the performance of the stock market matters, because it provides one measure of how the economy is doing.”

The story is pretty simple. Sunstein and some of his well-connected friends and associates offer “wisdom that should be acted on.” They’re the nudgers that do the acting; the rest of us are the nudgees that do what we’re told. No, not quite: we don’t do what we’re told because we’re full of cognitive biases. Remember: “A lot of people are more scared than they have any reason to be.” So the nudgelords do what’s necessary, following the careful experiments of the Brian Wansinks of the world, to get us to act the way they want us to act. They nudge us to take more vacations to keep the stock market going, they nudge us to execute people, whatever it takes to overcome those pesky biases of human nature.

In short, the nudgelords will save us from our “truth bias” and our “meta-cognitive myopia.”

To be fair . . .

OK, so far I’ve taken a bunch of cheap shots at Sunstein—hey, this is a blog, cheap shots is what we do!—which will do nothing to disconfirm his impression of me as “graceless.” And just wait till he hears about my stance on the hot hand!

To be fair, though, there’s nothing necessarily incorrect about the general Nudgelord attitude, which is:

1. There are cognitive biases. People find it difficult to reason logically when their emotions are involved.

2. Uncertainty triggers emotional reactions People are typically upset by uncertainty and will do a lot to avoid it, or to avoid thinking about uncertainty. For some people there is the opposite reaction and uncertainty creates excitement.

3. Put the above two items together, and this is a recipe for avoidable mistakes when making decisions under uncertainty.

4. Ordinary people and ordinary government officials make these mistakes.

5. Social scientists have studied these problems, and Sunstein is well read, well connected, and a good communicator, so he’s well situated to explain these issues to the general public and to suggest policy remedies.

6. If well-placed people in government and the private sector listen to Sunstein and his fellow nudgelords, they (the well-placed people) will achieve personal success and we (everyone else) will be happier by being nudged to make the right decisions.

I see nothing wrong with any of the six steps above, nor do I have any problem putting them together and saying that Sunstein and his friends are making a net positive benefit in the world. I can even agree with some of the blurbs—ok, everything but the bit about “the most brilliant minds in American law,” which makes me think of people like Robert Bork, Laurence Tribe, Alan Dershowitz, and the O. J. Simpson defense team.

Seriously, yeah, people make mistakes, we have cognitive biases, it should be possible for policymakers to make use of the insights of centuries of researchers ranging from Laplace to Piaget to Kahneman and Gigerenzer to whoever’s at the cutting edge of the field right now, and it makes sense for writers such as Sunstein to draw these connections.

So what’s the problem?

Given what I wrote just above, what’s my problem with the nudgelords? Aren’t they a force for good in the world?

My answer is that I’m not sure. The problem is that good advice is good, and bad advice is bad, but Sunstein and his friends doesn’t seem to distinguish between them. They seem to be spewing out research or claims about research or claims about behavior based on what they already want to believe, without regard to the quality of evidence. And they rarely seem to go back and assess what went wrong.

And, when people express skepticism about their claims, these critics are labeled as “Stasi.” I don’t trust someone who wants to get the government to nudge us, when he has a track record of (a) hyping shaky claims, and (b) ignoring, dismissing, or going on the attack against serious criticism. So I’m suspicious, despite that the Milton Friedman Distinguished Service Professor of Economics tells us, “This is wisdom that should be acted on.”

P.S. Bonus Sunstein quote:

I have noted that when confident people are shown to be wrong, people tend to stop believing what they say.

All right, then.

40 thoughts on “Nudgelords: Given their past track record, why should I trust them this time? (Don’t call me Stasi)

  1. The problems we face in policymaking are rarely ones where an “approachable introduction to decision-theory” could possibly function as “a provocative argument for how governments ought to handle risk.” Maybe I should wait to read the book, but as someone who works on decision theory and policy, it’s infuriating to see facile takes about maxmin and Knightian uncertainty, when the issues we have are mostly about ignoring even more basic tools for decisionmaking, (Actually generating new options! Asking what information would be useful! Talking to people and gathering expert opinions! Planning in advance instead of reactionary lurching from disaster to disaster!)

    Oh, and it’s a bit silly to talk about maxmin as a useful approach for policy when we have actual tools for dealing with deep uncertainty in policy, like robust decision making, assumptions based planning, and scenario planning.

  2. With this research program, there are two claims being made. First and foremost, there are cognitive biases that can subvert performance at particular tasks, which can be overcome with research, understanding, and deliberation. Second, there are lots of the aforementioned that apply across broad contexts, have remained heretofore undiscovered, and can be consistently found by a couple weeks of thought and a couple months of experiment.

    To take a specific example, the instinct when driving a car and a tire pops is to brake, but that’s actually completely terrible. Of course, because lots of people drive, and because it’s in those people’s best interest to not die, this has been catalogued by driving instructors and is taught to new drivers. The alternative scenario, that people have been driving for a century and nobody’s noticed this negative tendency, seems pretty unlikely.

    Of course, sometimes there’s a $20 bill lying on the floor; to see someone pick one up and deny that it could have been there is anti-empirical. But the nudgelords are just constantly stumbling across fresh $20 bills every other time they go outside, I guess claiming to have a system for finding them. They go into some game like chess or basketball that people have been thinking hard about for a long time, quickly find some fundamental mistake that every player can improve upon, then move onto the next game to do it again for their next paper. I’m not buying it. Given that not that many games have changed, it’s much more likely they’re face-planting over and over in every direction into Chesterton’s fence.

  3. ““At this stage, no one can specify the magnitude of the threat from the coronavirus. But one thing is clear: A lot of people are more scared than they have any reason to be”

    Odd. The only way he could’ve claimed that a lot of people are more scared than they ought to be would’ve been if he had actually successfully specified the magnitude of the threat from the coronavirus, showing it to be low …

    • And the logic of this is precious also.

      > “Recent evidence suggests that capital punishment may have a significant deterrent effect, preventing as many eighteen or more murders for each execution. This evidence greatly unsettles moral objections to the death penalty, because it suggests that a refusal to impose that penalty condemns numerous innocent people to death.

      So all we need to do is execute 18 million people and we’ll eliminate murder of innocents in this country entirely. I think that’s a moral imperative.

      • Joshua:

        1. There are about 15000 murders per year in the U.S., so according to this logic you’d just have to execute 1000 people, not 18 million, to bring the murder rate down to 0.

        2. To be fair, the causal inference refers only to the executions on the margin. If existing executions really did prevent 18 murders each, this would not imply that ramping up the execution rate to 1000 per year would send the murder rate down to 0, as presumably there would be diminishing returns.

        • Andrew –

          > 1. There are about 15000 murders per year in the U.S., so according to this logic you’d just have to execute 1000 people, not 18 million, to bring the murder rate down to 0.

          OK, 18 million, 1,000 – I guess I was off just a tad.

          > 2. To be fair, the causal inference refers only to the executions on the margin. If existing executions really did prevent 18 murders each, this would not imply that ramping up the execution rate to 1000 per year would send the murder rate down to 0, as presumably there would be diminishing returns.

          Sure. I wasnt being serioua.

          But my point was that even accepting the deterrent effed is proven (I think dubious), the return per execution would obviously depend on the merits of the choice as to who to execute. I don’t think it’s valid to project as they did.

        • Joshua:

          Oh, yeah, that estimate of 18 murders being prevented per execution was a classic product of statistical noise, arising from taking a ratio with a very small and noisy number (the number of executions) in the denominator. I guess that Sunstein believed it because it looked convincing and it gave him a result he was comfortable believing. Sunstein’s a liberal but he wants broader influence, so it was convenient for him to be able to take a moderately conservative position (pro death penalty) and dress it up with some charming we’re-smarter-than-you paradoxes. In that way, he’s like an academic economist but without doing any economics, all the rhetoric with none of the substance.

          But . . . if the study had been done well and it really supported the claim of 18 murders being prevented per execution, then, yeah, I think Sunstein would have a fair argument about the utilitarian benefits of the death penalty as it existed at the time. This would not necessarily support a massive increase in death sentencing, but it would be a strong argument in favor of not abolishing the death penalty.

        • Andrew –

          > But . . . if the study had been done well and it really supported the claim of 18 murders being prevented per execution…

          How could that be done well? As you mentioned – there’s the issue of diminish returns. Imagine the first through ten executions had a significant deterrent effect as people realized that capital punishment is a reality (I’m skeptical). But would execution number 401 really 18 lives as compared to the li R’s saved from the first 400? Thst seems to me to be a facile assertion.

          And that’s assuming there wouldn’t be a diminishing return in that at least theoretically, as you added to the number of executions you’d be executing people on the basis of less compelling evidence and/or for less aggregoius crimes. Part of the whole thinking behind the deterrent effect of punishment is that it has to be seen as swift and just. As you move to the edges or the rationale for who is to be executed, seems to me the clarity of the “justice” of a given execution would necessarily become fuzzier.

          I haven’t read his argument but I have to admit that I’m skeptical that anyone can really control for the confounding variables to use longitudinal analyses or cross-locality comparisons to “prove” a significant deterrent effect. But the idea that it has some measure of deterrent effect doesn’t sesm inherently implausible to me. And if someone makes that case in a solid manner, then fine. But that wouldn’t justify that kind of facile projection.

  4. My biggest problem with the nudgelords is that they are rarely satisfied to nudge. The only argument for nudging instead of commanding is unobservable heterogeneity in the “right answer,” but it’s really, really important. But all too often in the nudging literature, that heterogeneity is conveniently downplayed, usually in the service of “effective communication.” At which point the nudges are usually figleafs for commands.

  5. I’ll read the book. As I have mentioned several times, the ones who are schooled in critical thinking, including cognitive biases, reveal their habits of mind in casual conversations. I have discovered that experts too can miss their own biases, especially if the incentives/rewards also foster these biases. So that is the conundrum.

    As Andrew has pointed out, we need to be able to assess the quality of evidence. Requires access to raw data.

    With respect to COVID19, most of the experts and public don’t have sufficient knowledge of the biology of the virus nor of the mechanics of testing. In my non-expert opinion, that is where the more substantive discussion is being waged.

  6. On the Amazon page, there is a list of glowing endorsements for the book from intellectual heavyweights. If you search on Amazon for those heavyweights, you will find the books they wrote with glowing endorsements from Cass Sunstein.

    Now what made me want to do such a cynical thing? Because one of the endorsements described the book as “at once modest and transformative.”

  7. Andrew used the term, “nudgelords,” six times and thus far, several responses brought the count up to thirteen. Naming rights can be fun but “blogjuvenality” should not be carried to extreme.

    • This should be studied in Israel and the Arabic speaking countries. If calorie labeling to the right of the main label was more influential than left labeling, I’d be impressed.

      • Actually, it was studied in Israel and they reported finding a benefit of labeling on the right for Hebrew readers (study 3 in the original paper). The problem with the reported data/conclusions is that every study worked. That’s a problem because, based on the magnitude of the reported effects and their sample sizes, some of the studies should _not_ have worked (just due to random sampling). Thus, the reported results seem too good to be true, which can plausibly be interpreted as indicating some kind of error in data collection, analysis, or reporting.

  8. It’s not surprising that subsets of academics support each other. As I was discussing with some friends, the shelf life for some theories/practices has become shorter and shorter. I was noting that audiences huddle in their echo chambers. This note is born out by Pew Research regarding audience habits on Twitter. Experts don’t engage enough with the public much. A few like Michal Mina and Stefan Barai engage actively to their credit. It enhances one’s credibility to be open to criticism.

    What is most interesting to me is that we are having difficulties in fostering historical narratives that guided post WW 2 policy-making. I thought Richard Posner emphasized the need for accountability for opinions/scholarship rendered. There are so few platforms in eliciting the extent of accountability of prognostications circulated.

    Andrew’s blog is a great platform for engagement among experts.

    Otherwise, I’m mulling over the proper use of ‘who’ and ‘whom’ as well as when do use single quoations. Hmmmmm

    • It seems to me that historical narratives about the pre-WW2 period were partly responsible for post-WW2 policy errors. Prime example: the “Munich syndrome”. (Is that why Dulles refused to shake hands with Chou En-lai at the 1954 Geneva conference?) Perhaps the difficulties are signs of a more thoughtful approach to historical studies. (Confession: I am not familiar with Posner’s work on the topic.)

  9. Cass Sunstein and his wife, Samantha Power (Biden just named her as director of the US Agency for International Development), are [the] Dem / Liberal Power Couple. I would use caution in criticizing them, especially if you’re seeking career advancement within the DC Beltway…….

  10. At this point I’m pretty familiar with Cass Sunstein. I’ve participated in a multi-day workshop with him, read some of his books and many of his articles, and have used his writings frequently in my teaching. I don’t plan on reading his new book since I think I already know what he will say about almost any topic.

    There are two central legs (stool minus one) to his thinking. The first is that he fully embraces a conception in which the ideal rational actor is counterposed to the empirical actor burdened by nonrational heuristics. His understanding of rationality is not simply abstract in the sense of consequentialist decision theory but operationalized via orthodox welfare economics. All choices are understood as consumption decisions over which each individual has or should have a consistent preference map. Our truest guide to well-being are the consumption choices people make when not under the influence of heuristics, which is most of them. Thus monetization as a framework for cost-benefit analysis of the remaining choices is theoretically grounded. There is also a libertarian bias to welfare economics insofar as it uses individual utility-maximizing as its benchmark, understanding individuality as possession of one’s own preference map. The honoring of individuals as a moral and political value—liberalism—is therefore no more or less than respecting their preferences. I will admit that Sunstein has helped me understand the limitations of welfare economics to the point where I would now do without it altogether.

    The other leg, which has made him a frequent subject of this blog, is that he shares the delight that Andrew has recognized in many economists (and economist followers) in showing that people are less rational than they think they are. Sunstein and those of a similar bent regard themselves as experts in rationality, equipped to detect lapses among the less knowledgeable. Identifying such lapses and coming up with policy tweaks to remedy them is how they see their job. (Designing nudges is one strategy but hardly the only one, and Sunstein’s advocacy of cost-benefit analysis goes far beyond nudging.) As Andrew has pointed out, identifying the foibles of others gives people like Sunstein enormous pleasure, so much that they seize on every research paper that claims to have found a new irrationality, whether or not the evidence stands up to scrutiny. I won’t stoop to pointing out the irony here….

  11. The last section of this post makes me think that there is some unacknowledged complexity in this issue. Basically – famous/influential people have an impact because they have optimized, at least in part, for becoming and staying famous and influential. This might necessarily (though I don’t actually know that it must) make them worse at getting things right. There was a recent Scott Alexander post on this that I thought was pretty good: https://astralcodexten.substack.com/p/webmd-and-the-tragedy-of-legible.

    I particularly wanted to bring up this quote, which is referring to the fact that the CDC and Fauci were lagging pretty far behind the best evidence at the beginning of the pandemic:
    “[Fauci is] a very smart and competent doctor, who wanted to make a positive difference in the US medical establishment, and who quickly learned how to play the game of flattering and placating the right people in order to keep power. In the end, he got power, sometimes he used it well, and other times he struck compromises between using it well and doing dumb things that he needed to do to keep his position.

    I don’t want to judge him. Everyone has to make their own compromise between morally-pure-but-useless and tainted-but-useful, and I think Fauci comes out better than many. This isn’t about judgment”

    I don’t mean this to come off as defending people who misrepresent science to push a policy agenda or advance their own careers, particularly these really cringey ‘nudge’ ideas. I think criticizing these people is good and helpful! But I do think it should be acknowledged that in these public sphere competitions the alternative is very rarely ‘other people with better scientific principles’… At least if someone is citing research to push their agenda (whether or not they are doing so in good faith) then they can be debated with on scientific grounds.

  12. ““Knowing a person’s political leanings should not affect your assessment of how good a doctor she is.”

    Gee, Dr. Mengele, . . .

    Sorry, I couldn’t resist.

    • Ethics are separate from skill, though. Much of what we know about hyperthermia – rate of cooling of the human body, survival times, and the success of various revival techniques – is due to data collected from Nazis freezing prisoners to death.

      “unethical” doesn’t even begin to accurately describe how horrible these experiments were, but the data is good and whether it is ethical to use that data has been a matter of debate.

      https://theconversation.com/is-it-ethical-to-use-data-from-nazi-medical-experiments-39928

      • Given my (and my friends’) personal experiences with doctors and nurses during this pandemic, I don’t doubt the relevance of the medical professional’s political leaning to the outcome for some diseases. A family friend and physician I know was still touting HQC as a miracle drug in December. A friend of mine went to the doctors office over the summer and nurses there were openly suggesting that going out to bars without masks wasn’t that big of a deal.

        You have to ask yourself if the treatment (or suggested behavior) might have some sort of moral relevance to anyone. If the answer is yes, I’m gonna bet that the moral relevance will correlate highly with political leaning.

        • “doctors office over the summer and nurses there were openly suggesting that going out to bars without masks wasn’t that big of a deal. ”

          Have those nurses changed their tune since?

  13. If we really believe in the full subtle-effects doctrine of nudging, doesn’t that imply it’s an urgent moral imperative to regulate advertising at least as strictly as tobacco, with the same, semiavowed aim of eventually getting rid of it?

    Of course Google, Facebook, WPP etc are significant political donors. Let’s talk about banana placement.

Leave a Reply to Sameera Daniels Cancel reply

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