Thesis Drift

How to get out of hell

Why should an individual investor bother with picking individual stocks? There is usually no good financial reason to do so: even full-time professionals struggle to beat the S&P, and the few skilled investors who can beat the S&P often don’t have enough money to make it meaningful.

One good answer is that investing is a fun intellectual puzzle; if you are wired that way, stock picking can provide a form of entertainment that compares favorably to doing the crossword or watching Netflix, both in terms of utility and in terms of cost (negative if you are good!).

Another answer is that investing can be intellectually enriching: it requires you to come up with models of industry economics and company performance and then actually bet your own money on the predictions generated. It is possible to learn the theory of statistics, economics, and psychology from textbooks and professors, but there is something about repeatedly wagering one’s own money that unlocks important practical dimensions of those subjects, bridging the gap from theory to application.

Investing can teach important lessons about separating skill from luck and signal from noise and the role of emotions in decision-making, lessons that are broadly applicable to the rest of our day-to-day lives. Here we will examine a particularly useful investing concept: the idea of thesis drift.

Thesis drift refers to the tendency of investors to invent new rationales to continue owning a security after their original thesis has been falsified, instead of cutting their losses and moving on.

For example, let’s say you come up with a theory that the steel industry is ripe for imminent consolidation, and you buy stock in Anacott Steel because you think it is a likely takeover target. However, months pass with no sign of any deal activity in the space, and Anacott stock is down.

At this point, you should just admit that you were mistaken, and take your losses and move on. But that is rarely what happens.

Instead, you come up with a different intellectual justification to be long Anacott Steel – perhaps now you have come to the realization that Anacott’s new management team is underrated by the market, and Anacott’s stock will rise as a result.

A year later, Anacott has shown no sign of outperformance, and the stock is still stagnant. Undaunted, you have another brainstorm: steel industry margins are going to rise because of greater capital discipline. And if a year later, that thesis doesn’t play out…well, you get the idea. You just move from thesis to thesis in an effort to justify your original position, losing money at every step of the way.

The problem with thesis drift isn’t necessarily any of these new theories themselves; any of them could be correct. The problem is that when your first theory was falsified, you got a clear, unambiguous signal that there was something wrong with your overall model of the steel industry. You don’t know what was wrong with your model, you just know that your model produced a really bad prediction.

At that point, you should simply take your money out of a situation you clearly don’t understand, and move it to a situation you do understand. Why do you need to speculate on steel stocks? You could invest in one of a hundred other industries, or you could just throw your hands up and put the money in an index fund. Also, you are clearly spiraling. You’ve stopped actually thinking, and instead you are just spinning up narratives that justify your original course of action.

Thesis drift is entirely psychological in nature. It is natural to want to avoid admitting to yourself that you were wrong or that you didn’t know what you were doing, as it conflicts with your self-image as someone who is smart or has good instincts. If you have made public pronouncements about your position, changing your mind threatens your external image as well.

If you don’t recognize thesis drift, you can get stuck in hell. Hell, in this context, is when you get stuck doing the same thing over and over and expect a different result. All you have to do to stay out of hell is to respond to the clear, unambiguous feedback you got when your initial thesis was falsified. That feedback is your lifeline! Once you sever that feedback loop, it’s easy to enter a tailspin, because there is nothing left that can pull you out of it. You just keep losing money on the same failed position with no end in sight. And what’s worse, when you are stuck in hell, you look like an insane person, doing the same thing over and over again while expecting different results.

People expect investing to be a game of intellect, where you produce clever theories that you can turn into money. In reality, it is more of a psychological battle to keep your brain from sabotaging your portfolio. The concept of thesis drift is an example of “name it to tame it”: by anticipating this psychological pattern in advance, you can recognize and neutralize it when it surfaces. You know that all investors will be wrong sometimes, you know that there is no shame in admitting that you were wrong, and you know that it is exceptionally dangerous to stick with a position that stemmed from a broken thesis. It’s actually a clever trick: you don’t need to embrace the Anacott bear case, you don’t need to tinker with your model of the steel industry, you just walk away and don’t waste any time or money on it.

The good news is that in the world of investing, it is expected that even the best investors will make mistakes, and it is considered a sign of discipline when a money manager sells out of a losing position. This is why there are sayings like “Losers average losers” and “You don’t have to make back money the same way you lost it”. To a layperson, it can be jarring to see Bill Ackman suddenly blow out of a $1 billion Netflix stake after a bad earnings report or to see Warren Buffett liquidate a $7 billion airline position post-Covid, but an experienced investor recognizes it as rational and healthy behavior.

Thesis drift has clear applications in our personal and professional lives: most of us find it hard to change jobs or careers or get out of bad relationships when confronted with mounting evidence that we made the wrong choice, and we come up with all sorts of rationalizations. We also see it in organizations of all kinds (for-profit, non-profit, and governmental) that stick with failed strategies with no end in sight. But we will not dwell on these applications here.

Instead, let’s look at thesis drift as it applies to public policy. Via Marc Rubenstein, Rusty Guinn of Epsilon Theory has a wonderful visualization of thesis drift as it pertains to tariffs, reproduced below:

The original rationale behind broad tariffs was that they would strengthen and protect domestic manufacturing, but when that thesis was contradicted, the messaging pivoted to “tariffs can replace the income tax”. When it became clear that wasn’t true, the message became “tariffs give us leverage to force neighbors to secure their borders / lower their tariffs / pay more for defense.” From the outside, it’s clear that these different narratives are contradictory (e.g. a crucial source of revenue or pillar of industrial policy can’t be bargained away) and that tariff supporters are engaging in thesis drift to rationalize their original policy.

It would be tempting to write this off as a quirk of Trumpism, but this is actually an issue with all flawed populist policies that manage to gain traction. Populist policies gain support because, to borrow the words of Dan Williams, they feel like they should work. People support tariffs because it feels like tariffs should make us richer, no matter what the experts say.

The problem with populist policies is that they only feel like they should work. In reality, they are based on flawed models of the world, and as a result they produce undesirable outcomes that were predicted by experts but come as a surprise to their supporters. This is where thesis drift comes into play: instead of abandoning the policy and admitting that their model was wrong, populists simply pivot to a new narrative to justify their original policy.

Thesis drift is especially dangerous because of the confluence of two separate modern intellectual trends. The first trend is the shift toward populism. We tell people that they should “think for themselves” and be skeptical of elites and experts, but we never warn them that most of modern society is built on our experts’ understanding of counterintuitive but (relatively) accurate models of reality. For example, pretty much everything we know about the physical sciences runs contrary to the intuition of anyone that lived before the Scientific Revolution, and the only reason that we easily grasp that objects with different weights fall at the same speed or that the earth revolves around the sun is because we are taught these concepts before we try to figure them out on our own.

When people go forth and think for themselves, they expect to uncover original hidden truths, but instead they rediscover centuries-old folk theories like mercantilism – ideas that originally gained popularity because they feel like they should be true before being debunked through careful study and bitter experience. To compound the problem, when people compare notes and find they have independently come up with the same conclusion, they imagine that they have stumbled on an important idea, instead of it being yet another case of two dumb bitches telling each other “exactlyyyy”.

Populist economic ideas have always been a part of public policy, in one form or another. We’ve had tariffs, price controls, supply restrictions, demand subsidies, “supply-side” deficit spending, and austerity, just to name a few. We are getting more populist health policy as well: the anti-vax movement is well known, but now we even have a health secretary that doesn’t believe in germ theory, instead opting for a version of 19th-century miasma theory.

The most dedicated adherents of populist ideas will make it a major part of their identity, at which point there is no turning back. They involve themselves in activism, they make speeches, they take up positions at think tanks and write papers. They see their advocacy as an important part of their life’s work, a contribution to society that gives them purpose and meaning. This makes them especially prone to thesis drift, as they cannot accept the blow to their status and self-image that would result from repudiating their original ideas. This accounts for the stubborn minority that dig in and defend failed populist policies, even after they produce disastrous outcomes.

The second trend is the idea that we should be maximally charitable toward populist ideas as a good form of epistemic humility or intellectual hygiene. The idea is that any popular theory generated through lived experience has merit and should be taken seriously, and we don’t want to risk throwing away a good theory simply because people can’t articulate why they think it works.

One problem with this approach is that it relies on naive realism, the idea that people see the world as it is. In fact, our first-hand experience captures only a narrow, distorted slice of reality, and we use mental shortcuts that lead us to misinterpret causal relationships and evidence. Experts have developed more rigorous statistical techniques to tease out hidden parts of reality and create better predictive models of the world. It is not impossible for an intelligent layperson to find an accurate idea that the experts have overlooked, but it is a remote probability relative to the much higher odds of simply rediscovering bloodletting, flat earth theory, or another similar debunked idea.

The other problem with this approach is that it encourages thesis drift, which as we have seen is a dangerous trap. I hate to pick on anyone in particular, because I know the intent is good, but here is an old Scott Alexander post on “steelmanning the NIMBYs”, speculating about ways that NIMBYs could be correct about housing policy.

Now we have analyzed “NIMBY” housing policy at length here in past editions, because it is such a typical example of populist economics. A large proportion of populist economic ideas are based around a misunderstanding of what prices are: it is commonly thought that prices are set by greedy suppliers rather than emerging from market forces to ration scarce resources and provide important signals and incentives. The idea behind NIMBYism is that new housing construction causes high housing prices (because you see rising rents anywhere you see new construction), and by restricting construction, we can make housing more affordable.

In reality, this is completely backwards. Rising rents are a function of rising demand (which themselves are a function of external economic forces), rising rents are what incentivize new construction, and new construction is what pushes rents back down by taking away leverage from existing landlords. The cities that have maintained NIMBY housing policies have seen rents spiral out of control, as tenants are forced to outbid each other for (artificially) scarce units.

The problem with “steelmanning” NIMBYism or any other populist policy in this manner is that people are already inclined to find excuses to philosophize about different ways that their pet theory could be correct, even though it has clearly failed in practice. That’s what thesis drift is! Indeed, we see that NIMBYs give ever-changing, contradictory reasons for opposing housing: new buildings will lie empty and they will also create more traffic, and they will attract poor people and will also cause gentrification, and they will make housing unaffordable and will also cause home values to tank.

Experts sometimes imagine that they can convert populists by carefully demonstrating how their models fail to meet rigorous standards for logic and evidence. The problem with that approach in practice is that people who are attached to debunked models also fail to recognize the merit of rigorous standards for logic and evidence — otherwise they wouldn’t have gotten so attached to those debunked models in the first place.

We previously looked at how 19th century European doctors rejected Semmelweis’s discovery of handwashing as the way to prevent the spread of deadly childbed fever (a bacterial infection afflicting new mothers spread through dirty hands or sheets), even after he deduced the method of infection through a (pseudo-)randomized controlled trial and implemented his solution again and again with spectacular results.

One of the problems is that other doctors did not understand causal inference and proper statistical methods as well as Semmelweis did, so they attributed his spectacular results to chance and actually claimed that handwashing was an arrogant theory that had been debunked, and that childbed fever was in fact complex and multicausal. It was nearly four decades before a new generation of doctors recognized that Semmelweis was right and handwashing was actually widely adopted.

This is the danger of thesis drift when reasoning within a complex system. If you can’t accurately connect cause and effect, you get trapped in hell, attributing your bad results to spectacularly terrible luck or sabotage from those around you, finding excuse after excuse to maintain the status quo when the cure is sitting right there in plain sight.

The next best solution is to have a sort of intellectual stop-loss in place, a no-fault way to recognize and get out of a bad position, similar to the way an investor can sell out of a stock with their reputations and self-image intact. Ultimately, the outcome is all that matters: you need to avoid frittering away all your money speculating on steel, you need to wash your hands, and you need to allow housing to be built, even if you don’t understand or accept the theory behind why you need to do these things.

One way this happens in practice is that people find ways to back out of the status quo without having to admit that they were mistaken, either publicly or to themselves. They find a way to graft part or all of the policy they opposed onto their debunked framework, giving them a path to generate desirable outcomes without admitting fault.

For example, many of the doctors that ridiculed Semmelweis publicly also quietly adopted handwashing regimens at their own hospitals. They still denounced Semmelweis’s theory as simplistic and arrogant but they allowed that dirty hands might be a minor contributing cause, so why not do it if you feel like it? This frustrated Semmelweis to no end, and was less than ideal in that it prevented handwashing from spreading as widely as it should have, but it succeeded in saving lives.

We can see elements of this already happening in trade and housing policy. Reframing tariffs as a way to extract concessions gives a clear path to eventually eliminating tariffs without admitting fault. Progressive politicians are finding ways to allow more construction without abandoning their overall intellectual framework.

It’s easy to get hung up on the ideal outcome of making people understand an important model at the same deep level that you do, but the reality is that even very intelligent people either won’t have the time to truly learn about a complicated model and all of the evidence that goes into it (Paul Krugman put it well when he said that modern theories of trade, like most scientific models, relies “on a dense web of linked ideas”) or their brains will reject any model that runs contrary to their intuition. (“My intuition can’t be wrong; I went to Yale.”) The messy reality of life is that sometimes we only get the outcomes we want when we realize that we are trapped in hell, and we need to find a way out.