written in reaction to Eliezer Yudkowski’s book

In the Hans Christian Andersen classic, “The Emperor’s New Clothes,” a young boy defies the orthodox belief that the emperor is wearing a fine outfit. When the boy shouts, “But the emperor is naked,” everyone else soon agrees, and they discard the orthodox belief.
But overturning orthodoxy is not that simple, as Eliezer Yudkowsky points out in his new book, Inadequate Equilibria. Yudkowsky does not use this fairy tale example, but he might say that what is most unrealistic about the story is that the rest of the people believed the boy. The normal reaction would instead be to dismiss the boy’s challenge as the unqualified opinion of someone who does not even know how to sew. Yudkowsky writes,
if your status is low, then many people will intuitively perceive an unpleasant violation of the social order should you associate with yourself an image of possible future success above some level.
If it succeeds, a challenge to orthodoxy from someone who lacks appropriate credentials represents a threat to the existing social order. Even if they have no personal stake in the beliefs being challenged, individuals who possess elevated status within the relevant hierarchy will instinctively resist the heterodox economic beliefs of a non-economist, the heterodox medical ideas of a non-doctor, or the heterodox management ideas of a low-level employee.
The most significant episodes in my career have been when I stood for heterodox beliefs. For that reason, Yudkowsky’s book raised issues that matter to me, even though I did not always find Inadequate Equilibria to be clear or convincing.
This essay proceeds as follows. I will articulate Yudkowsky’s two major themes in two ways, first using jargon from calculus and statistics, then using plain English. Next, I will tell some stories from my own life that relate to these themes.
The calculus-jargon explanation of “inadequate equilibrium”
Orthodox beliefs and social practices can operate like a hill-climbing algorithm that gets stuck at a local optimum. There is a better social outcome available, but we need to use a different algorithm to reach it. Unfortunately, we often remain stuck at the local optimum. In his words, we are in an “inadequate equilibrium.”
The statistics-jargon explanation of “excess modesty”
When you think about challenging orthodoxy, the null hypothesis is that you are wrong and the orthodox view is right. If you challenge and you turn out to be wrong, that is a type I error. If you fail to challenge but you were right, that is a type II error. Yudkowsky says that we worry excessively about committing type I errors, and as a result we commit too many type II errors. In his words, we are too “modest.”
The English explanation of “inadequate equilibrium”
Social practices can be difficult to change, even when they are wrong. Yudkowsky terms a bad social practice that is difficult to change an “inadequate equilibrium.”
Individuals who challenge orthodoxy are often made to suffer, even when they turn out to be right. Think of Galileo.
More important, Yudkowsky points out that escaping from an inadequate equilibrium might require more than one group of people to change at the same time. This is particularly difficult.
The example that occurs to me is the inadequate equilibrium that prevailed relative to homosexuality in the 1960s. Homosexuality was widely viewed as a psychological disorder, as embodied in the Diagnostic and Statistical Manual of that era.
When the orthodox view was that homosexuality was a mental disorder, psychologists had an incentive to stick with the orthodox view. They were expected to offer treatment, and failure to do so would have been unprofessional.
This orthodox view also affected homosexuals. They had little incentive to assert that they were healthy. Instead, their incentive was either to seek treatment or try to hide their homosexuality.
Getting out of this inadequate equilibrium required both groups to change. If homosexuals did not try to assert that they were healthy, then psychologists would tend to persist in treating homosexuality as a disorder. If psychologists had continued to regard homosexuality as a mental disorder, then it would have been difficult for homosexuals to claim that they were healthy. Both groups had to change at the same time.
In fact, it took more than just these two groups to escape the inadequate equilibrium in which society treated homosexuality as a mental disorder. Some clergy had to challenge institutional authority within their religions. When other clergy stuck with the old authority, religious adherents had to defy their clergy. Judges had to change their behavior relative to legislation, and legislators had to change their behavior relative to past positions.
In short, we should not be surprised that bad social practices prove to be difficult to change. When change requires a combination of new behaviors by two or more groups of people, it is easy to remain stuck.
The English explanation of “excess modesty”
I often criticize mainstream economics. However, when I read other people’s criticisms of mainstream economics, I almost always think that those criticisms are wrong.
Suppose that I were to reason as follows:
Most of the people who criticize mainstream economics are cranks who are obviously wrong. Thus, when I encounter a critic of mainstream economics, the most reasonable assumption is that the critic is wrong. Therefore, when I find myself criticizing mainstream economics, the most reasonable assumption is that I am wrong.
This line of reasoning is what Yudkowsky calls “modest.” He acknowledges that many people who challenge orthodox beliefs are overconfident, which is why I encounter so many critics of mainstream economics who I regard as misguided cranks. However, he says that people who are wise enough to understand the modesty argument tend to err in the opposite direction. He claims that too many intelligent people are under-confident.
My personal perspective
I have never been shy about challenging orthodoxy. I do not think it even occurred to me that a hierarchy should be respected for its own sake.
One might say that my entire life as an intellectual and business person has been about challenging orthodoxy. In some cases, I turned out to be right, and I look back on those with great satisfaction. In other cases, I turned out to be wrong, and I try to forget about those emotionally but understand the failures intellectually. In other cases, the jury is still out.
When I was in graduate school in economics at MIT, most of my classmates were in awe of, or at least respected, economists with outstanding mathematics skills. I thought that the mathematical models were often silly. I thought that one should give more respect to logical thinking, intuition, and experience. Forty years later, I am still fighting this battle, but I see some signs that things are moving in a direction of greater methodological pluralism in economics. I have an essay in National Affairs that speaks to this (and more).
Also when I was in graduate school, the most exciting idea was something called “rational expectations.” I thought that this concept was interesting but not always correct and not nearly as central as others saw it. I think that many orthodox economists have come to share this view, although it has taken a long time.
I can name many fellow students who wrote “rational expectations” dissertations and ended up with high prestige in the academic/policy world, and I did not. After I obtained my PhD, I got off the academic track, which in the end probably made me a wiser economist.
Jumping ahead to 1991, I was at Freddie Mac, the mortgage giant, looking into ideas for using technology to make credit decisions more efficiently. We were initially looking at rule-based systems. But when I heard a presentation on using credit scoring for second mortgages because they are similar to consumer loans, it struck me that this approach could be applied to first mortgages as well. My position was unorthodox, but after a long, hard battle, I won over the senior executives. By the late 1990s, credit scoring was standard in mortgage underwriting.
In the short run, credit scoring did make underwriting more efficient. However, years after I left Freddie Mac, Wall Street picked up on credit scoring to greatly expand the mortgage securities market, and many lenders took the view that with credit scoring they could get away with loans where they did nothing to verify the borrower’s income, assets, or employment. In hindsight, I would say that intellectually my challenge to orthodoxy was sound. But as things turned out, the financial system would have been better off if nobody had discovered the idea of applying credit scoring to mortgages.
I left Freddie Mac because I was not rewarded for overcoming orthodoxy. In fact, while others took over the implementation of my ideas and received the credit, I was faced with a humiliating demotion. I quit, and I started a business on the Internet.
Starting an Internet business in April of 1994 was heterodox. Almost no one else was doing it. But within five years it was orthodox, and I had just enough luck to come through the experience well myself.
Like Yudkowsky, I have sometimes challenged medical orthodoxy, even though I have no training in medicine or biology. Years ago, a doctor found microscopic blood in my urine and sent me for tests. I protested that he did not understand Bayes’ theorem. He told my wife that I was a crank, and I eventually went for the tests to mollify her and the doctor. The tests turned out to be painful and unnecessary. I did more research, and I decided that I had been correct, and that my Harvard-trained doctor was wrong. I switched doctors. I have since read that doctors do not know Bayes’ theorem. If you know it, and your doctor does not, don’t be afraid to challenge orthodoxy.
Orthodoxy in pediatrics is to insist to parents that newborns sleep on their backs. Basically, this is because one pediatrician was adamant that this would reduce instances of Sudden Infant Death Syndrome. SIDS is an event that is too rare to make it feasible to test using a controlled experiment. But apparently as a matter of fact, the rate of SIDS was lower after pediatricians started giving this advice than it was before. I fail to find this evidence decisive, because (a) many other factors have been changing over time and (b) the causes of death probably have been reported in different ways at different times (maybe now they can identify precise causes of death that previously would have been impossible to classify and thus dumped into the SIDS category).
Meanwhile, I observed that many babies now show delayed development in terms of crawling, and a pediatrician I talked to confirmed that there is evidence that this can be attributed to back-sleeping. I also observe that many babies have sleep disorders that last from ages 6 months to three years. I have no idea whether studies show that this has increased relative to 40 years ago, but my belief based on anecdotes is that it has.
I may be wrong about all this. But I would bet that if the issue of back-sleeping were studied rigorously, the orthodoxy would turn around, babies would be sleeping on their stomachs, and the outcomes would be better for parents and children.
Finally, back to economics. I have many heterodox opinions. They resemble opinions that you can find in the work of Deirdre McCloskey or Russ Roberts or some of George Mason University’s economists, such as Peter Boettke. Some people wrongly believe that I am on the faculty at Mason.
My book Specialization and Trade spells out my heterodox views. At this point in my life, my biggest career goal is to overthrow mainstream neoclassical economics and see it replaced with something closer to the views in my book. That is quite a challenge, and reading Yudkowsky’s Inadequate Equilibria helped me to clarify the nature of that challenge.