One Funeral at a Time

Why smart people become married to stupid ideas

Dec 31, 2024

By the time of his death last year, Sam Zell had accumulated a fortune of $5 billion, the fruits of a lifetime of shrewd real estate investing. What was the secret behind his success? Well, if you believe Sam Zell, it was that he followed an analytical approach that was grounded in Econ 101: basic supply and demand.

Now, that sounds like obvious nonsense, the kind of aw-shucks claim a billionaire might make to distract from his 180 IQ or his wealthy father-in-law or his once-in-a-generation luck. There is no possible way that there is any meaningful alpha that can come from understanding introductory economics, much less enough alpha to make someone the 317th richest man in the world. Zell must have either been spinning a self-serving narrative, or he was completely mistaken. Right?

Just for fun, let’s humor him and take a look at his story. He started managing apartment buildings while he was an undergrad at the University of Michigan, in the early 1960s. He realized that real estate in second-tier markets like Ann Arbor was undervalued, so he leveraged that into a promising career managing and investing in properties in cities like Toledo and Reno.

By the early 1970s, he thought that US real estate was getting irrationally overbuilt, because of the organizational incentives in place at construction and lending institutions at the time (emphasis mine):

First, there were dedicated real estate lenders. Each bank or insurance company would start the year with a set amount of lending capital, say $5 billion to invest. The branch would allocate a certain amount to bonds, a certain amount to equity, and a certain amount to real estate. The rule was that at the end of the year, if each department hadn’t invested its full allocation, the capital had to be returned to the parent company. Not surprisingly, the departments never gave anything back.
…
Other new money had also flooded the industry from the early real estate investment trusts (REITs)...The construction REIT industry went from $1 billion to $21 billion within three years, fueling massive real estate development.
…
This has always been a fatal flaw in US real estate: the volume of development has been related to the availability of funds, not to demand. The industry has a long history of overbuilding when there’s easy money, without regard for who will occupy those spaces once they’re built.
…
My thesis was that over the next five years, we would have the opportunity to make a fortune by acquiring distressed real estate…Everyone thought I was nuts. After all, occupancies were still over 90 percent. Absorption was high. Companies were hiring. It was one of many times I would hear people tell me that I didn’t understand.

Zell raised a fund to invest in distressed real estate, and bided his time. In 1974, the real estate market did indeed crash, and Zell had his first opportunity.

Zell’s Econ 101 analytical framework was really quite simple. He would look for good buildings trading for a fraction of replacement cost — replacement cost being the estimated cost to build a comparable asset from scratch. He reasoned that these assets would be insulated from new competition for a long time: no one can raise $20 million to build an apartment building that will be worth $10 million at completion.

In the absence of new construction, he could be assured that prices would rise, as long as demand kept growing. This second part of the equation was a sure thing in his era, as the Boomers came of age and population shifted toward cities. A growing number of households chasing a stagnant supply of apartments would inevitably push up rents, which would in turn make the buildings more valuable.

He understood that eventually the value of the apartment buildings would surpass replacement cost, at which point developers would start building new apartments, putting a lid on future appreciation and rent growth. But by then he would have made a large profit, both from the appreciation of the asset and the cash flow he collected in the meantime.

His approach was hardly a secret: he laid out this exact thesis in a famous 1975 article in Real Estate Review entitled The Grave Dancer, which Zell later adopted as his nickname. And yet, it did not seem to attract much competition. In the mid-1970s, Zell’s investment firm bought billions of dollars of real estate of all types – residential, retail, and offices — all at a fraction of replacement value. Sure enough, construction stopped, asset values recovered, and he made a fortune.

Zell realized that his replacement cost framework would apply to all long-lived capital investments, not just real estate. In the mid-1980s, his firm bought a fleet of railcars, which were then trading at half of replacement cost, following a construction boom a few years before. Same model, same bet, same result. In 1992 he flipped his railcar fleet to GE for a huge profit.

This is not to say that Zell’s fortune can be entirely attributed to simply buying quality assets trading at a discount to replacement cost. Zell was a savvy operator and a talented dealmaker who boosted his returns through leverage and clever structuring. But at the same time, throughout his career, this was Zell’s big shtick: “to look for opportunities in places where others were ignoring the rules of supply and demand”.

There is nothing more basic than the rules of supply and demand, so why didn’t more investors see what Zell saw? Here we have to speculate a bit. We should start by observing that the replacement cost model doesn’t feel right, or at least it probably wouldn’t have felt right to most market participants at the time. Consider:

It just seems too simple. How could something as complex as the market for (say) commercial office space could be so well described by such a parsimonious model? What about interest rates, and demographic trends, and shifts in sectoral employment, and international money flows, and oil prices? How could any useful model not consider these important inputs?
All of the traditional indicators of real estate industry health — occupancy rates, rents, and so on — were flashing red. It would seem reckless to buy into an industry that is struggling so badly.
By the same token, you simply shouldn’t buy into an asset class in freefall. Don’t try to catch a falling knife. The trend is your friend.
Zell’s model seems to be based on a paradox. It relies on the market being inefficient enough to not fully understand this model — to first overbuild when money is flowing freely without proper consideration of actual future demand, and then later to shy away from competing for distressed assets trading at a deep discount – but also efficient enough to stop building new assets when prices are low.
Zell is in his early 30s, and anyone that young just lacks the experience to understand what he is doing.

We tend to learn about investing the same way we learn any other skill: we get a sense for what works and what doesn’t based on the limited feedback we receive. This develops into a set of heuristics and intuitive models that help us navigate the world. People of the same generation tend to have similar experiences and similar approaches and often converge on a collection of generally accepted investing folk theories.

On one hand, you can see how these folk theories probably are a little useful. Simple models are often too good to be true. Some asset classes in freefall actually are in terminal decline. Young people really do lack experience and are more likely to make naive mistakes. In a market with less experienced counterparties, these heuristics will be a good guide to making money.

On the other hand, these heuristics are just mental shortcuts, not ironclad models of reality. These objections are all demonstrably false. Interest rates and economic cycles matter for asset values in the short term, but in the long term, we can expect asset values to converge on replacement cost, as long as demand growth is there. It is actually often very profitable to invest in a fundamentally healthy industry experiencing a temporary setback. Market participants really do stop building when asset prices are below replacement cost — you will have a hard time finding a real world counterexample. And Zell’s relative inexperience was actually an asset here: while experienced investors picked up the first four heuristics and were bound by them, Zell was not similarly constrained and came up with a more accurate model.

Investing folk theories are simple and easy to apply, but they tend to be unreliable and contradictory. They often lead to Econ 101-flavored conclusions, but sometimes they diverge, as we saw here. In a competitive market, they can be exploited by someone applying a more rigorous (but still basic) Econ 101 analytical model, as Sam Zell did.

This still leaves one open question: Zell was out raising money behind his thesis for years, and he even went so far as to publish it in a trade journal. Why didn’t other market participants pick up their Spring 1975 copies of Real Estate Review, realize their mistake, and take their properties off the market? Or better yet, jump in and buy with both hands?

One explanation might have to do with institutional imperatives: under the rules of the time, banks had to find a way to restructure their outstanding bad loans. But on the other hand, Zell was adamant that his view was always distinctly in the minority, and he doesn’t seem to have had much competition in the market.

The more likely answer lies in how we model reality. It is very difficult to navigate a world with noisy feedback and uncertain attribution, so our default approach is to extrapolate some useful heuristics from our personal experience. If we have some success using these heuristics in the real world to make money, they become etched in our brains as The Inviolable Laws of Investing. Anyone who proposes an investment approach that contradicts one of our Laws is subconsciously written off as an idiot.

In Richard Feynman’s introduction to his lectures on physics, he reminds his students that what we think of as the “laws of physics” are really approximations, the best models we have based on what we have been able to verify through experiments so far. The best approximations we have to model the physical world have evolved considerably over time, from Aristotle’s notion that all objects have a natural tendency to be pulled toward the center of the universe (which he held to be the center of the earth), to Galileo and Newton and beyond.

But our brains are not wired to think of the world that way. We can accept that Einstein is subject to future revision, but Bob’s 6th Law of Trading must be accepted as an ironclad axiom, because Bob used it one time to trade pork bellies in 1985 and made enough money to buy a Firebird. As Charlie Munger put it:

…[T]he human mind is a lot like the human egg, and the human egg has a shut-off device. When one sperm gets in, it shuts down so the next one can’t get in. The human mind has a big tendency of the same sort.

The market is made up of a lot of smart people, but a lot of smart people have similar experiences and thus they independently derive similar heuristics with similar blind spots. And what is a market inefficiency, if not a collective blind spot that cannot easily be arbitraged away in a short period of time?

Sam Zell’s story seems especially unrealistic when we hear it today because he came of age almost two generations ago. We are more likely to understand Econ 101, and more importantly, we know the stories of people like Sam Zell who made billions taking this exact approach to distressed investing. But history at least rhymes, and it is almost certain that the biggest money-making opportunities today are sitting in plain sight, in the market’s blind spot.

We talk here sometimes about the high cost of housing in major coastal American cities, like New York, San Francisco and Los Angeles. The story goes like this: At various times between the 1960s and 1990s, most of these coastal cities instituted barriers to new housing construction. Since then, the population of the country has grown, broader economic shifts have pulled people toward big cities (fewer workers in agriculture and manufacturing and more workers in services and knowledge-based industries), and changes in preferences and lifestyles have raised demand for urban living (people getting married later and wanting to live in dense, amenity-rich cities in the meantime).1

In short, the demand to live in these big cities has steadily grown, but the supply of housing has hardly budged at all. Most housing in the US is rationed through market forces – that is to say, it gets auctioned off to the highest bidder – and as a result housing prices in these big cities have exploded over the last three decades. For example, housing prices in the Bay Area have risen 2.7x since 1987, adjusted for inflation.

High housing prices are a policy choice. We can understand this by using a simple replacement cost model, very similar to the one Sam Zell used. Without barriers to housing construction, we would expect housing prices to cap out at around replacement cost. (Yes, land is still finite, but new apartment buildings economize on land by building upwards, so land only makes up 10% - 20% of total construction costs.)

When housing prices rise above replacement cost, developers capitalize by constructing new buildings, and pocket the difference between the market value of the completed building and the cost of construction. New construction sets off a game of reverse musical chairs: property owners need to fill their apartments to generate income, but the number of people who want housing is fixed. As a result, property owners have to undercut each other on price to attract occupants. This continues until price falls to replacement cost, and it is no longer profitable to build new housing.

(This model excludes a prominent feature from our prior story: the tendency of developers to overbuild and push housing values below replacement cost, at least temporarily. As Sam Zell put it, “I think that to stay in business, most developers must get 50 percent of their returns from real cash flow and the other 50 percent from the intangible benefit of seeing their phallic symbols rise out of the ground. Otherwise I can’t see the reward.” For simplicity, we can ignore this little detail – it’s sufficient to know that absent barriers, developers will build like crazy when the conditions justify it.)

This simple model holds up very well empirically. Ed Glaeser and Joseph Gyourko pulled the data for American cities over the last decade, and produced the chart above, with the ratio of housing prices to replacement cost on the y-axis and the amount of housing construction on the x-axis. The only cities with housing prices meaningfully above replacement cost are those that have growing demand and strictly limit new construction, such as San Francisco and Honolulu, where housing prices are triple what they should be.

Other cities build enough to meet demand, keeping housing prices in line with or below replacement cost. Glaeser and Gyourko note that there are two major groups: Sunbelt cities that are attracting new residents but build enough to absorb demand while keeping prices at replacement cost, and Rust Belt cities that aren’t attracting new residents and don’t need to build to keep prices below replacement cost.

It should be no surprise that this simple model matches reality so well: if it didn’t, there would be cities in which there was instant free money available to any developer who wanted to pop in and start building. The world is not perfectly efficient, but it’s more efficient than that.

If the solution to high housing prices is as simple as legalizing construction, and most cities do it successfully, why are there so many major holdout cities? We’ve talked about one theory relating to incentives: the idea is that since current homeowners actually profit from high housing prices, maybe they intentionally restrict construction for their own selfish economic benefit.

As the main explanation for high housing prices, this theory has a few major holes:

It requires an implausible level of economic sophistication from everyday citizens.
It doesn’t make sense if homeowners have kids (which most do), since young adults who will need housing are the main victims.
By this theory, voters are knowingly voting for an anti-social policy, which is pretty cynical.

Chris Elmendorf, Clayton Nall, and Stan Oklobdzija recently published a study that points to an alternative hypothesis: they did a series of surveys and found that almost all renters and a slight majority of homeowners say they want lower housing prices, but two-thirds of all surveyed believe that housing construction actually increases housing prices, or at the very least, has no effect. Most people place the blame for high prices on landlords and housing developers.

In other words, voters want housing prices to be lower, but in trying to achieve that goal, they unintentionally enact policies that actually push housing prices higher. When housing prices go up, they don’t connect it to the policies they just implemented; instead, they attribute higher prices to outside forces (e.g. gentrifiers, foreigners, and greedy landlords), and stick with the policies that just backfired, which causes prices to go even higher.

Is it really possible that people are that irrational? Well, let’s dig in a bit on how people form and change beliefs.

We usually imagine that science progresses in a very straightforward way: scientists do experiments, compile and analyze the resulting data, and use what they have learned to collectively update their theories, marching human knowledge forward.

In The Structure of Scientific Revolutions, Thomas Kuhn argues that this idealized model of scientific progress obscures the messy reality. He explains that scientific facts and theories are not free-floating, but rather are only understood as part of what he calls a paradigm, an accepted, unified predictive framework that explains observed phenomena within a specific field. For example, early Western astronomers explained the night sky within a paradigm that held that the earth was at the center of the universe and that the heavens were fixed, as explained by Aristotle and Ptolemy, and this was widely accepted for centuries, and actually did a good job of predicting the positions of the heavenly bodies.

According to Kuhn, the prevailing paradigm also dictates what kinds of scientific questions can be asked, and what even counts as legitimate evidence. Scientists don’t just do random experiments; rather, they rely on the existing paradigm to guide their research. Scientists devise experiments to extend and resolve puzzles within the current paradigm, never seeking to overturn it. For example, early medieval astronomers compiled tables based on the Ptolemaic model, and mostly did not try to come up with alternative models of the universe.

How, then, do scientific fields ever advance from one paradigm to the next? How do we get scientific revolutions? Well, over time, scientists think up more clever experiments, often with newly developed equipment (e.g. microscopes and telescopes), and their experiments, undertaken with the intent of refining and extending the existing paradigm, start to turn up more and more anomalies that require more and more exceptions and epicycles to be piled onto the accepted narrative.

At some point, newcomers to the field begin to suspect that there might be a better paradigm that explains all of this anomalous evidence. And then a Copernicus or Newton or Einstein comes along and provides a new paradigm that does just that. Eventually, scientists see that the new paradigm fits much better with the existing evidence (or if the evidence is still unclear, sometimes they just decide that the new paradigm is much cleaner and more promising) and abandon the old paradigm.

The feature of the Kuhnian model that is most pertinent to our discussion is his observation that most scientists strongly resist conversion to the new paradigm, even as the evidence piles up. In fact, many leading scientists never accept the new paradigm at all, going to their graves as loyal adherents of the old model.

Kuhn cites the story of Joseph Priestley, the 18th century English chemist who discovered oxygen. At the time, the paradigm chemists used to explain combustion was something called phlogiston theory. The idea was that any combustible substance contained an element known as phlogiston: burning wood would cause phlogiston to be released into the air, where it would be absorbed, and once the phlogiston was gone, you would be left with ash. This theory sounds backwards to us today, but in fact it did very well to explain most of the empirical results chemists were able to obtain at the time.

In 1774, Priestley was able to extract oxygen gas by shining sunlight onto a sample of mercuric oxide. Only, Priestley didn’t understand it as oxygen. He decided that it was air, but “more pure, more respirable”. After seeing how well it supported combustion, he concluded that it was “dephlogisticated air”.

It was a young French scientist, Antoine Lavoisier, who picked up on the real implications of Priestley’s discovery. At the time, Lavoisier was trying to figure out why substances like phosphorous gained weight when burned, a result that appeared inconsistent with phlogiston theory. A meeting with Priestley in Paris sparked his curiosity, and he set about reproducing Priestley’s dephlogisticated air experiments for himself.

Lavoisier eventually decided that Priestley had actually isolated a component of air, which he dubbed oxygen. This led him to conclude that phlogiston theory was nonsense, and to offer a rival (and correct) theory of combustion: that it was a reaction between a substance and oxygen in the atmosphere.

Here’s the funny thing: despite being immortalized as the man who discovered oxygen, Priestley stuck with phlogiston theory for the rest of his life. Many of his contemporaries resisted conversion to modern chemistry as well, even as Lavoisier demonstrated the power of his new theory by synthesizing water from hydrogen and oxygen, and then breaking it back down again. The phlogistonists always had some objection or pointed to some anomaly that kept them from abandoning the old paradigm. For his part, Lavoisier seemed to accept this:

Lavoisier did not expect his ideas to be adopted at once, because those who believed in phlogiston would "adopt new ideas only with difficulty." Lavoisier put his faith in the younger generation who would be more open to new concepts. Two years later, in 1791, the results were obvious. "All young chemists," he mused, "adopt the theory, and from that I conclude that the revolution in chemistry has come to pass."

The story of Joseph Priestley reflects a common pattern in science. Kuhn notes that it was many decades before the work of Newton or Copernicus was widely accepted. He also quotes Darwin who in On the Origin of Species accurately predicted: “I by no means expect to convince experienced naturalists whose minds are stocked with a multitude of facts all viewed, during a long course of years, from a point of view directly opposite to mine…[B]ut I look with confidence to the future, – to young and rising naturalists, who will be able to view both sides of the question with impartiality.”

Kuhn also quotes Max Planck on the subject, who said: “a new scientific truth that does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it.” Or, as Planck’s quote is more commonly paraphrased: science moves forward one funeral at a time.

This historical pattern is often misunderstood as relating to the stubbornness of old people, or as describing how human nature generally resists change or new ideas. In fact, age itself has nothing to do with it, except as a proxy for experience: Kuhn explains that older newcomers to a field are just as likely to discover and adopt new paradigms as their younger counterparts. It is only with experience and repetition that we become hemmed in by a paradigm.

And although devotion to old paradigms presents as an irrational resistance to change and new ideas, in fact it is nothing of the sort. Kuhn tells us that it is something inherently psychological, because what we see in the world is mediated by what we expect to see from previous training and experience. For example, he notes that because the Aristotelian paradigm held that the heavens were perfect, early Western astronomers never noted the existence of sunspots, even though they are visible with the naked eye under the right conditions. By contrast, ancient Chinese astronomers were not bound by any such paradigm, and so they regularly commented on the sunspots that they observed.

As an analogy, Kuhn uses the classic sketch of the duck/rabbit – the drawing is ambiguous enough that a viewer can see it as either a duck or a rabbit. In the same manner, we can understand that scientists of one generation can interpret the exact same set of evidence very differently than scientists of the next generation. For example, early astronomers looked at the night sky and saw a duck (the heavens revolving around a fixed earth), but afterward Copernicus and Galileo, astronomers looked at the exact same night sky and saw a rabbit (our perspective from a spinning earth orbiting around the sun) .

Let’s go back to the survey from Elmendorf et al., the one that found that most people think that new construction raises housing prices. Why do people believe this? Well, in fact, new construction is associated with higher rents. People walk around their city and see new construction in neighborhoods with rising rent and think “Ah, gentrification! These new buildings are making the rent too damn high!” Which, to be fair, is the simplest and most intuitive interpretation for what we see.

In reality, higher rents cause new construction, and not the other way around. The cycle goes:

People want to move to a certain neighborhood, pushing up rents. The reason is external: maybe it’s close to growing tech employers, or maybe it offers an increasingly popular lifestyle, such as walkability or transit access.
Higher rents attract developers.
Developers build new apartments, which push prices in that neighborhood back down to some equilibrium where it is no longer profitable to build.
The newly-built apartments in that neighborhood also end up lowering rents everywhere else, as it decreases the number of people chasing the existing supply of housing.

When cities enact supply restrictions, we only get step 1, where rents go up in that city, and to a lesser extent, increase rents and construction everywhere else. For most people, this story is a lot less intuitive, but it is the only story that fits the evidence. Believing that new construction causes higher rents because we observe higher rents in neighborhoods with new construction2 is like believing that hospitals are a major detriment to public health because we observe higher death rates in hospitals.

Now, let’s invert the problem: How do people end up believing in the correct model?

In the first section of this essay, we looked at one group of people who generally get it right: real estate investors. Unlike laypeople, they can draw on a much broader and more detailed experience: they routinely compare price trends across different cities, they see at a micro-level what happens to rents when a new apartment building opens, and so on.

They naturally pick up from their everyday experience that high rents only exist in supply-restricted markets, and correctly conclude that local construction restrictions are the key to maintaining high rents. In fact, they readily publicize this, albeit through the use of passive euphemisms like “long-term supply barriers”, as in this recent investor presentation from AVB, a major publicly-traded apartment owner in California and the Northeast, bragging about the lack of supply growth (and implicitly, higher rent growth) in their markets:

The layperson and the real estate professional generally follow the same intuitive reasoning process. (Zell’s fortune is evidence that real estate professionals don’t necessarily understand Econ 101.) The difference is that the layperson sees a much narrower sample, and draws the wrong conclusion, while the professional sees the bigger picture, and gets it right. Extending the rabbit/duck analogy, the layperson sees only the head and automatically interprets the housing market as a duck, while the real estate professional sees the same head, but with paws and a bushy tail, and reflexively sees the housing market as a rabbit.

Who else believes in the correct model? NYT journalist Conor Dougherty traces one of the earliest “YIMBY” (Yes-In-My-Backyard) pro-housing construction movements to Stockholm in 2007. Stockholm has a rent control system where apartments are allocated by waiting list.

A system that rations housing via a waiting list makes it transparent that housing scarcity can be solved through more construction: just build until everyone on the list has an apartment. A market system that rations housing to the highest bidder makes the nature of the problem more abstract, and misleads people into thinking that the root problem has to do with landlords or developers or capitalism. In fact the solution is the same in the market system as it is with a waiting list: just allow construction until the overhang of demand is satisfied. Continuing our earlier analogy, this group sees a bit of fur and that is enough of a cue for them to conclude that they are dealing with a rabbit.

Dougherty has also extensively reported on the YIMBY movement in the Bay Area, founded by Sonja Trauss, a former economics graduate student. Of course, someone who has studied economics at a graduate level will immediately recognize that the root problem is supply restrictions. Economists know enough about rabbit anatomy to be able to get the rabbit from an ambiguous picture of the head — maybe the ears are a giveaway.

The YIMBY movement is disproportionately made up of young, educated urbanites, whereas people that oppose new construction are predominantly older. This is often attributed to economic incentives, but in fact this is the same pattern that Kuhn observed with scientists: young people haven’t yet been conditioned to see the housing market as a duck. They also come into a different world than the last generation did, one where the price difference between supply-restricted cities and the rest of the country has become substantial. Even if a young person initially sees a duck, if you nudge them into noticing that ducks don’t eat carrots, it will click for them that they are actually looking at a rabbit, and they will only see a rabbit thereafter.

We like to imagine that people arrive at their opinions by weighing the objective evidence, and smart people are more likely to arrive at the right answer because they are better at reasoning through the facts. But there seems to be an element of luck: whether we end up with the right answer has lot to do with the path we take through life, and if smart people are better at coming up with the right answer, it is because they are more likely to have learned more of the relevant facts, or because they have been exposed to the most useful paradigm for thinking about those facts at an early enough age.

This is at odds with our usual conception of the world, but completely aligned with Kuhn’s theory. Remember, the people who got stuck on discarded models of the world were actual brilliant scientists. Many of them are still remembered today for their pathbreaking discoveries. It is just really hard for the human mind to move from one paradigm to another.

Phlogiston theory went away completely because once Lavoisier spelled out modern chemistry and it was adopted by young chemists, it would never again be possible to follow the same intellectual path as the phlogistonists. Phlogiston theory makes sense if you start by knowing nothing about chemistry, and start doing experiments to try to work out basic principles, but every child today learns what oxygen is, and knows that water is H2O.

Economics is different. In a sense, most people do go out in the world and start doing economics to work out the basic principles, without learning them in a textbook first. Even if people encounter the concept of supply and demand early on, it is counterintuitive enough that they usually don’t truly grasp it well enough to properly recognize and apply it in real life.

The thing about working out broader principles from personal experience is that personal experience is almost always very narrow and unrepresentative. What you think is a broader principle is often actually just a statistical artifact: it’s a failure to control for a confounder, or a selection effect, or Berkson’s paradox, or one of a dozen other common statistical illusions, or (as in this example) a simple case of getting the causal relationship backwards.

A smart person who tries to deduce fundamental principles from limited experience will usually end up in the same place as every other past smart person who did the same thing: with some version of phlogiston theory. Which is actually very smart! But also not very useful if you do so at a time when modern chemistry is already well understood.

This is one reason why the economics discourse is afflicted with so many “zombie ideas”, ideas like supply restriction that are easily disprovable and thus rejected by experts, but still enjoy wide acceptance among the masses. People independently rediscover the equivalent of phlogiston theory over and over again, and once they do so, no amount of evidence and reasoning can sway them to modern economics.

(When people learn that their best-loved ideas have been rejected by the experts, they are actually encouraged, and congratulate themselves for their independence of thought; reasoning by analogy, they imagine themselves as having progressed beyond the frontier of expert knowledge, rather than considering they have actually managed to regress to the 18th century.)

All of this is to say that housing opponents aren’t dumb, or if they are dumb, it has very little to do with their beliefs about housing. It is perfectly natural for smart people to adopt narratives based on their first-hand experience and then subsequently to be unable to get away from them when confronted with conflicting evidence.3

Ok, but what about that conflicting evidence? Why doesn’t that convert people? There isn’t any real ambiguity in the data here, since the two models make completely opposite predictions, and the effect size is fairly large. You can look at housing prices by city as Glaeser and Gyourko did, you can do a carefully controlled study at the neighborhood level, or you can look at the success of cities that loosened restrictions. You would think that all of this data would be persuasive enough to win converts, but when these studies are presented, housing opponents usually react like the robots in Westworld: “That doesn’t look like anything to me”.

Kuhn explains that part of what makes converting to a new paradigm so hard is that we see what we expect to see. For example, housing opponents argue that developers will never build enough to lower prices, reasoning that it would be against their collective self-interest. According to them, it is just universally known and beyond question. Many of them also assert that prices are set based on the greed of the business owner, not by any market forces. On these fundamental topics, they cannot see any evidence to the contrary, in the same way that ancient Western astronomers could not see the sunspots that were so obvious to the Chinese.

Kuhn also points out that the standards for what qualifies as evidence is itself contained within the paradigm. Some housing opponents will just say that “I know it to be true”, or “I’ve seen it with my own eyes”, others will spell out intelligent-sounding arguments that upon close inspection are built on circular reasoning (“if we assume my model is true, my model turns out to be true”), and still others will even come up with “evidence” that turns out to be fabricated or built on logical error. Economists will insist on certain standards around logical consistency, proper controls, and statistical significance, but why would housing opponents defer to economists on standards for evidence, if they don’t accept the economists’ paradigm?

Once you have been accustomed to using one paradigm to explain a set of facts, it is very difficult to even properly grasp a totally different paradigm. What even is replacement cost, to someone unfamiliar with it? It is hard to correctly state a foreign paradigm in your own words even when it has been correctly explained to you, and it is hard to even isolate a correct explanation with so many wrong versions floating around. It is very easy to dismiss a new paradigm by taking an incorrect version of it, conjuring up a counterexample to the incorrect version, and declaring it disproven.

Finally, a potential convert also has to cope with consistency bias. Once you announce your position on a topic, it gets wrapped up with your ego and identity and your brain doesn’t want to consider a different interpretation of the facts. Even if you start to have your doubts, you will naturally resist conversion.

In a way, the best evidence against the folk paradigm is the kinds of arguments that are made in its favor. They are usually either appeals to heuristics (the Econ 101 model is said to be “trickle-down”, or the proposed solution of repealing supply restrictions is too simple for a problem that is so complex), or else they try to poke holes in the Econ 101 paradigm. But is it not suspicious that in a $50 trillion dollar market with rich, detailed data dating back a century, no one can point to a single rigorous study demonstrating the accuracy of this theory? Supply restrictions have been tried so many times in so many places, so why is there not a single success story where they led to affordability? For that matter, if this model is such a good representation of reality, why has there never been a Sam Zell who has traded on it and made money? One would think that this would be enough to raise red flags, but such is the power of a paradigm, to blind people to its shortcomings.

Finally, we can extend this model to explain one other puzzling feature of the housing debate. Laypeople usually find YIMBYs (that is, pro-housing activists) to be impossibly strident and arrogant. Scott Alexander said that “YIMBYism is one of the most tribal, most emotional, most closed-minded movements I have ever seen this side of a college campus”.

This is just a natural byproduct of paradigm conflict. Most YIMBYs subscribe to some version of the Econ 101 paradigm, and that paradigm holds that in a competitive market like housing, high prices are a symptom of scarcity — prices are what we use to ration scarce goods. It is almost tautological to say that supply restrictions lead to high prices, and that the only way to lower prices (without resorting to another form of rationing) is to remove those restrictions. They see the housing crisis as frustrating and unnecessary because it is so clearly man-made and because there is one (and only one) simple solution. But when they explain this paradigm, they are belittled and condescended to by the very people they see as causing the crisis in the first place!

Naturally, their opponents see the world very differently. They cannot understand how anyone could be so certain about a subject that is so complicated, and they resent that a bunch of children are accusing them of being the source of the problem. Furthermore, they know based on experience that the Econ 101 paradigm is overly simplistic and useless.

As we have already seen, when it comes to paradigm conflict, there is no resolving the two viewpoints. Because they are operating from two different frameworks, and two separate sets of facts, they each think the other side is ignorant, and they are always talking past each other. The YIMBYs can insist all they want that the shortage of duck eggs is because the “ducks” are actually rabbits, and their opponents will counter that they know a duck when they see it, and the duck egg shortage is obviously caused by greedy farmers, and nothing will sway either side.

Ignaz Semmelweis had a problem.

The year was 1846, and the 28-year-old Semmelweis had just been appointed First Assistant in Division I of the obstetrics department of Vienna General Hospital. This was a promising step in his career: Vienna General Hospital was one of the largest and most respected medical institutions in all of Europe, and a third of all births in Vienna occurred there.

However, Semmelweis was joining an obstetrics clinic in crisis. In recent years, it had been ravaged by outbreaks of what was then known as childbed fever, a fatal disease afflicting mothers who had recently given birth. New mothers who fell victim to childbed fever would come down with chills and a fever, and would almost inevitably pass away within a week; their newborns would frequently die as well soon thereafter, with the same symptoms. According to Theodore Obenchain in Genius Belabored (an excellent book, from which much of this section is drawn), childbed fever was the second leading cause of death for women in that era.

Childbed fever had been known since antiquity, but was rare before the 18th century. At that point, more people started living in big cities, where they were more likely to give birth in hospitals rather than at home. For reasons unknown at the time, childbed fever was often (but not always) much more common in hospitals.

At Vienna General Hospital, childbed fever had been very uncommon between 1784 (when the maternity ward was founded) and 1822, affecting less than 1% of mothers admitted. Between 1823 and 1840, childbed fever became much more widespread, with the mortality rate hovering around 5%.

In 1840, the growing maternity ward was split into two clinics: Division I was staffed by the (male) doctors, and Division II was staffed by the (female) midwives. At that point, the mortality rates of the two clinics diverged: the reported mortality rate in Division I averaged 10% (and in reality was higher: Division I would sometimes move terminally ill patients into the main hospital and off their books), while the mortality rate in Division II averaged only 3%.

Division I thus acquired a terrible reputation both within the hospital and in Vienna at large: women would sometimes give birth on the street (or would do so at home and pretend that they had given birth on the street) before checking into hospital for free medical care afterward, simply to avoid the risk of giving birth there. Some of the patients who were admitted would beg to be allowed to give birth in Division II. As Semmelweis described it:

The disrespect displayed by the employees toward the personnel of [Division I] made me so miserable that life seemed worthless. Everything was in question; everything seemed inexplicable; everything was doubtful. Only the large number of deaths was an unquestionable reality.

Semmelweis was absolutely determined to discover the cause of childbed fever. In his quest, he had a couple of unique advantages. First, Vienna General Hospital was very good about keeping detailed records, and Semmelweis had a two year break before assuming his position during which he took a course in logic and statistics and he studied and charted the available data that pertained to childbed fever.

Second, the two-clinic setup was perfect for isolating the cause of childbed fever. Patients were admitted to Division I or Division II solely based on the day they arrived (e.g. Division I for Monday arrivals, Division II for Tuesday arrivals, etc.), making it possible to rule out characteristics of the patients themselves (since they were randomly selected) or temporary epidemics. The two clinics were housed next to each other in identical buildings, making it possible to all but rule out environmental theories, e.g. that childbed fever could be due to bad air or defects in one of the buildings. In fact, a prior investigation had gone so far as to have the two clinics swap buildings, and still the high mortality rate for Division I persisted.

Semmelweis went about systematically testing and eliminating different hypotheses. One theory was that the bells of the priests who came to perform last rites were frightening the patients and making them more susceptible to illness, so he made the priests silence their ringing, with no effect. He noticed that the midwives had patients lie on their sides during delivery, so he tried that in his clinic, but that produced no improvement either.

Semmelweis turned to the data to eliminate other theories. One popular idea was that male doctors induced a sense of shame in patients, but since the disease also afflicted newborns, he concluded that this story must be incorrect. Yet another theory centered on overcrowding, but the data revealed no correlation there. Semmelweis was stumped, and the stress of the situation was building.

In October 1846, only four months into his appointment, Semmelweis was suddenly forced out of his job: his predecessor had returned and wanted his old position back. Semmelweis took the opportunity to spend the next few months studying English, with the intent of going to Dublin, to get a different perspective on fighting childbed fever.

In February 1847, before he could leave, Semmelweis learned that his old position was re-opening: his predecessor had found a more senior role at a different hospital. Instead of going to Dublin, he joined his friends on a short vacation in Venice, to clear his head before re-taking his old job. When he returned to his job eighteen days later, he was saddened to learn that while he was away, a colleague had passed away in a tragic accident: a student had accidentally cut him on the finger during an autopsy, and he fell ill with a fever, rapidly deteriorated, and died.

Out of curiosity, Semmelweis read the autopsy report. While he was reading it, it struck him that the abscesses and other symptoms described were identical to what the symptoms he had observed when performing autopsies of victims of childbed fever. He realized that the disease that killed his colleague and the disease that was killing new mothers and their infants were one and the same – a kind of pyemia.

That’s when it all clicked for Semmelweis. He understood that his colleague had been killed by “cadaverous particles” introduced to his bloodstream by a wound. He also knew that the doctors of Division I sometimes examined patients after performing autopsies, while the midwives of Division II almost never performed autopsies at all. He concluded that the differential in infection rates was because he and his colleagues were killing new mothers by introducing something from corpses into their bloodstream during examinations.

For Semmelweis, this theory explained some well-documented anomalous patterns of infection that no one had ever been able to make sense of before. Patients who underwent more examinations before delivery were at a much higher risk of infection, whereas patients who never underwent any examinations or gave birth on the street or at home were at almost no risk at all — of course this would be true if the examinations themselves were the cause. Cases of childbed fever would be discovered in random rows within the clinic — this must happen because doctors would examine patients in order, and a doctor carrying cadaverous particles would infect an entire row of patients before he got done.

This theory also explained the historical trends of childbed fever at Vienna General Hospital. From 1784 until 1822, the maternity ward was led by a man who preached minimal intervention, and childbed fever was almost unknown. In 1823, the hospital mandated the use of cadavers in teaching and fired the old head of obstetrics, and childbed fever became much more common. Semmelweis even made a note of a marked decline in childbed fever after he lost his job the preceding October: his replacement was much lazier, and rarely did autopsies on his own, nor did he encourage his students to do them either. When Semmelweis returned in March, deaths surged again.

This discovery also pointed to a possible solution. Doctors already washed their hands with soap and water, but that was insufficient to do away with the corpse-y smell that lingered after performing an autopsy (suffice to say this was before doctors used gloves). Semmelweis’s idea was that the smell was evidence that dangerous cadaverous particles remained. He figured that he could kill or neutralize the cadaverous particles by having doctors thoroughly scrub their hands with a chlorine solution, taking care to brush under their fingernails, until the smell was entirely gone.

In May 1847, Semmelweis implemented this new hand-washing treatment in Division I, and it worked almost exactly as theorized. Childbed fever all but vanished from Division I, with mortality plummeting from 11.4% in 1846 to 1.3% in 1848. Division I was now actually safer than Division II. Based on additional observation, Semmelweis refined his theory to state that childbed fever was spread solely by contact with decaying animal-organic matter of any kind, and that hands, instruments, and sheets should be disinfected so that no contaminants could be introduced to the bloodstream of patients.

His peers at the hospital were naturally quite impressed, and insisted on helping him spread the word about his wonderful discovery around all of Europe. Letters were written, speeches were made, and the response was…crickets. Or even worse than crickets: ridicule. Semmelweis’s theory was almost universally rejected as mistaken and totally implausible. Even Semmelweis’s own boss argued that the true cause of this miracle was a new ventilation system that had recently been installed. More commonly, they alleged that Semmelweis had fallen victim to a simple logical fallacy, confusing correlation with causation. Epidemics of childbed fever were known to come and go, and this wave just so happened to go away at the same time he implemented his handwashing regimen, or so the thinking went. Very few obstetricians were moved to implement chlorine handwashing. Semmelweis himself was pushed out of Vienna and picked up an obstetrics position at a hospital in his native Budapest, where, undaunted, he again wiped out childbed fever with his handwashing regimen.

Why did everyone ignore Semmelweis? In popular culture, it is assumed that 19th century doctors were dumb, arrogant, or possibly even misogynistic. There is little evidence for any of these theories: doctors of that era were intelligent and educated and they cared for their patients. Another hypothesis is that doctors refused to consider any theory that implicated doctors, but even back then doctors were well aware that they could spread disease and harm patients, and Semmelweis readily accepted that he personally had caused many deaths. One prominent obstetrician who successfully implemented Semmelweis’s treatment became so depressed thinking about the women he had killed in his career that he committed suicide.

Some modern scholars see the issue as having to do with politics, egos, or professional incentives. Undoubtedly these were contributing factors, but it hardly seems likely that doctors would have opposed Semmelweis if they truly believed in his theory, given the stakes. They were still heavily incentivized to limit patient deaths under their watch.

Others point the finger at Semmelweis himself, who was initially averse to writing, leaving his audience to glean his theory secondhand. Semmelweis was also hardly the most tactful ambassador for his theory, especially after it was publicly rejected, making him bitter and resentful. When he finally published his Etiology in 1861, he included a 200+ page section devoted to individually naming and eviscerating his many haters, logically demolishing their flawed beliefs and hammering them with personal insults. (He derided them as efficient murderers who were training even more murderers, and he told his biggest rival that he would be “memorialized as a medical Nero” and that he “[would] not escape God’s justice”.) That being said, none of Semmelweis’s more diplomatic allies had much success winning new converts either.

The most persuasive theory comes from Donald Gillies, a philosophy professor who in 2005 invoked our old friend Thomas Kuhn. His idea is that we have to put ourselves in the shoes of an early 19th century doctor, and understand what paradigms they used to think about disease. They knew about contagious diseases like smallpox, but childbed fever didn’t seem to work like that. They had an idea that disease was related to foul conditions and bad air, hence the theories about ventilation and building defects. They also had a notion that disease was often caused by epidemics that are subject to random spikes or come and go with the seasons, thus the fatalism in the face of sudden, prolonged outbreaks. Finally, they universally agreed that diseases were too complex to possibly only have only one possible cause – every disease almost certainly had many different causes.

These frameworks seem primitive today, but are understandable in a world with no knowledge of viruses or bacteria. They at least have some explanatory power – bad conditions do spread disease and illnesses like the flu are seasonal. The problem was that doctors of the time would see what they expected to see, and they would keep returning to these theories that Semmelweis had already disproven.

But what about the overwhelming evidence? Why didn’t that sway them? Semmelweis himself initially assumed that the evidence was so conclusive that he would have to exert little effort for his theory to be widely adopted across the Continent.

When we read the story of Semmelweis today, we immediately grasp what he was doing. We know all about causal inference and statistics and deductive logic, and we are primed to approach problems the same way Semmelweis did. We see the resemblance to a randomized control trial, which wasn’t commonly used in medicine until the 1940s. We can examine his 19th century study with 21st century methods and see that it holds up in terms of study design and statistical significance.

19th century doctors had no familiarity with any of this. Remember, other hospitals didn’t even keep clean records — no one outside of Vienna could have constructed a useful study even if they had understood logic and statistics (which they most certainly did not). The only reason Semmelweis knew anything about it was that he had two years off to learn it from teachers who were at the forefront of the subject at that time, and even that was very little by modern standards. Of course, 19th century doctors didn’t recognize their own ignorance — they knew just enough about logic to be dangerous, observing Semmelweis’s theory superficially resembled a common logical fallacy that they learned about in school and dismissing it from there.

They didn’t even regard the pathological evidence in the same way that Semmelweis did. Semmelweis looked at the description of abscesses and tissue and immediately drew a connection between his colleague and the victims of childbed fever. Other doctors thought that the parallel was absurd: Semmelweis’s male colleague obviously didn’t have anything wrong with his uterus, so how could anyone be so dense as to think that there was any similarity?

To a 19th century doctor, childbed fever resembled any ordinary epidemic: it came and went for reasons no one could possibly explain, and even exhibited pronounced seasonality when it was present. Semmelweis correctly demonstrated that his theory easily explained the observed seasonality: in the summer, there were fewer students doing examinations, and due to the heat, autopsies were conducted in the evenings, after they saw patients, rather than in the mornings before rounds, as was done in the winter. But to most 19th century doctors, this sounded like a roundabout justification — if it looked like a seasonal epidemic, and quacked like a seasonal epidemic, then it must surely be a seasonal epidemic.

Semmelweis’s idea that childbed fever only had one cause (decaying animal-organic matter) was so foreign to their conception of disease that even his supporters refused to accept that part of his theory. His detractors thought it was obvious to anyone with medical experience that a disease must have any number of causes, and he was engaging in a nonsensical rhetorical trick by redefining a disease in such a way that it only had one cause. But Semmelweis scholar K. Codell Carter explains that the way disease had been traditionally defined, as fuzzy groups of symptoms, made it very hard to pin down effective treatments, because the effectiveness of a treatment would vary depending on what disease the patient actually had — e.g. we know today that bacteria and viruses can cause similar symptoms but antibiotics will only work on bacteria. Semmelweis’s modern paradigm was a necessary leap for medicine to progress, but no one in the 1840s was ready for that yet.

Semmelweis’s theory was poorly understood in its time because it didn’t fit with any existing paradigm. Some doctors incorrectly believed Semmelweis was saying that cadaveric material was the only cause of childbed fever (rather than any animal-organic matter) and concluded that he was mistaken, because they knew childbed fever occurred in settings with no autopsies.

One school of British doctors, the contagionists, actually already long believed that doctors were responsible for spreading childbed fever, having observed that cases of the disease contracted during home births followed individual doctors from delivery to delivery all over the map, rather than concentrating in one location like ordinary epidemics. The contagionists thought that childbed fever followed doctors in some sort of miasma, and their prescribed treatment was for affected doctors to bathe, burn their clothes, and close their practices for a while. When they heard of Semmelweis’s theory, they were angry and even accused him of plagiarism. Semmelweis had to explain to them that because childbed fever was only spread by decaying animal-organic matter, burning clothing and closing down shop was useless and unnecessary, and that they just needed to disinfect their hands and instruments.

Following Kuhn’s documented pattern with new paradigms, the first doctors to embrace Semmelweis’s theory were the junior doctors and students who saw first-hand the power of his hand-washing regimen in Vienna. They understood the power of the cure in a visceral way that could not be communicated in a speech or a publication, and they were young enough to be open to a new model. Semmelweis’s most ardent supporters were of course the young doctors who had just taught him modern statistics and logic and pathology: they were fully ready for such a discovery, even if they didn’t comprehend the bit about diseases only having one cause.

We also see here once again the pattern of vitriolic debate, where the adherents of the new paradigm become angry and frustrated that the old school cannot see something that to them is so obvious, that the solution is so simple and sitting in plain sight. By contrast, those that were wedded to the old paradigm experienced Semmelweis as arrogant and irritating, lecturing as confidently and condescendingly as he did about something he clearly misunderstood.

The saving grace for the old school was that their paradigm allowed for diseases to be caused by multiple factors, which meant that they would sometimes quietly adopt handwashing in their own clinics, even while denigrating the theory publicly. Semmelweis’s rival and successor in Vienna, Carl Braun, argued that Semmelweis’s theory was “all humbug” and published a list of thirty possible causes of childbed fever, on which “cadaveric infection” was 28th. But in practice, he continued to enforce Semmelweis’s handwashing regimen in Vienna and even expanded it to Division II, and in so doing managed to keep mortality rates almost as low as they had been under Semmelweis. Another opponent, Eduard Lumpe, opined that childbed fever was too seasonal and handwashing too simple for Semmelweis’s theory to be accurate, but hedged by saying that obstetricians should “wait and wash”. This sort of behavior infuriated Semmelweis, but it was better than nothing.

Converts to Semmelweis’s theory were few and far between in his lifetime. He would die in an insane asylum in 1865, after cutting his finger and contracting the same infection that he had devoted his whole life to studying.

In 1857, Louis Pasteur published a paper describing the world of bacteria he had discovered under his microscope. This was germ theory, the theoretical explanation for what Semmelweis had previously deduced empirically. In 1865, Joseph Lister, a surgeon, was introduced to Pasteur’s paper and applied it to his own work, successfully disinfecting wounds with carbolic acid. Lister published his discovery and just like Semmelweis, he was initially ignored and doubted.

In 1869, a pair of French doctors, Victor Feltz and Leon Coze, showed that the deadly streptococcus bacteria was present in a patient with childbed fever. Pasteur confirmed this discovery in 1879. It was in that year that he was present at a speech where a prominent French obstetrician was lecturing about how childbed fever was caused by the miasmatic air present in a hospital. Pasteur got up and drew the streptococcus bacteria on the board and explained to the crowd that this was the true cause of childbed fever. The obstetrician retorted that even though this bacteria might be present in all victims, correlation was not causation, and that he would probably die before he saw the microbe that actually caused childbed fever.

It was only in the 1880s that the work of Lister and Pasteur and their colleagues finally paid off, and the medical community in Europe widely adopted antiseptic procedures to ward off childbed fever, almost forty years after Semmelweis originally successfully implemented and promoted the same solution. At this point, with germ theory widely accepted and childbed fever conquered, a new generation of doctors rediscovered Semmelweis and hailed him as a misunderstood hero — and they wondered how he could have been ignored for so long.

Armed with the paradigm of germ theory, doctors were able to explain and conquer many diseases that had previously been complete mysteries. But when in the late 19th and early 20th centuries, when confronted with deficiency diseases like scurvy (brought on by lack of Vitamin C) and goiter (brought on by lack of iodine), some doctors would make many of the same mistakes again, insisting that the true cause must lie within previously accepted paradigms like germ theory, and campaigning against proven treatments because it just couldn’t be true that miniscule amounts of a simple, obscure chemical could be the cure to such a complicated disease.

It is eye-opening to go back and see how often that simple solutions are ignored because they conflict with an accepted paradigm, and how quickly previously intractable problems disappeared once that simple solution was implemented. In past editions, we have talked about the struggles of Jack Welch, who insisted that massaged metrics would lead to improved results in the underlying business, rather than improving the underlying business to improve metrics. We looked at the deregulation of domestic air travel under the late Jimmy Carter, which made cross-country flights affordable within a few years. (Carter would do the same thing with trucking, rail, and telecom.)

Modern conflicts between paradigms also frequently play out just as Kuhn predicted back in 1961. If space permitted, this is where we would recount the resistance to the adoption of analytics by the old guard of baseball (as portrayed in Moneyball), complete with angry rhetoric and confusion and the eventual hiring of very young general managers to successfully change organizational approaches.

Perhaps the best approach is to view this pattern optimistically. It is almost certain that many of the most stubborn problems in our own organizations or in our own lives must have solutions that are lying in plain sight, just waiting to be adopted. It is up to us to figure out how to adjust our mental framework to be able to reach out and grab them.

Thanks for reading MD&A! This post is public so feel free to share it.

MD&A

One Funeral at a Time

Why smart people become married to stupid ideas

Discussion about this post