- Scott H Young - https://www.scotthyoung.com/blog -

How Good Is Your Theory?

If you’ve ever played a game of chance, like poker [1], you know one of the most frustrating experiences can be when you make the “right” decision, and still lose.

In a casual game with some friends of mine, one player wasn’t taking the game very seriously. When his cards were dealt in the first hand, without looking at them, he decided to go all-in.

Another friend who was playing more seriously noticed that he was dealt two kings, a very good starting hand. Since it was the first hand, everyone had the same amount of chips, he also had to go all-in to match the bet. Worrisome, because if he lost, he’d be out of the game before he even got a chance to play. But, at the same time, it was the correct move since he should win with such good cards.

The friend who wasn’t playing seriously and hadn’t looked at his cards? He had two aces! He won the hand even though the odds were stacked against him.

This kind of situation can be frustrating. Yet, nobody looking at the game would have seen my friend—who went all in without checking his cards—and used that as a model for being a good poker player. This fluke win didn’t rule out that he was using a bad strategy.

Bad Strategy or Just Unlucky?

In games like poker, we can easily distinguish those who won because they had the right strategy, and those who won because they were lucky. Statistically, those two things tend to coincide, but there’s plenty of weird moments, like my friends’ game, which offer a counter-example.

One reason we can distinguish a good strategy from a bad one that happened to be lucky is that we have an extremely good theory for how winning should work in poker. Laws of probability mean that, provided there’s no cheating going on, we can estimate the probabilities of certain plays winning, given certain decisions, almost exactly.

The theory for poker is so good, that it’s easy to see the difference between bad-decisions-which-won and good-decisions-which-lost. My friend made the right choice, he just happened to be unlucky.

Most of the luck in life, however, doesn’t have nearly so good a theory. Success in your career, business, investing or relationships, doesn’t simply reduce to laws of combinatorial mathematics.

What should you do in those cases, where working out the correct decision is much harder?

In those cases, you can simply copy. Look at people who won and do what they did. Yes, this will cause you to copy some bad strategies—like my friend who bet everything without looking at his cards—and you’ll make mistakes. But you’ll also make mistakes if you end up following a theory which turns out to be completely wrong.

Copy or Theorize?

I’d like to put these two different approaches to making decisions on a spectrum. At one end you have blind copying. There’s no attempt to understand what is being copied, it is just trying to match as much as possible, in the hopes that you pick up something that works.

At the other hand, you have pure theory. You work out all the theory exactly until you figure out what is the “correct” approach and apply it. Ignore experience and examples which are counter to this as being just luck.

Most decisions in life are somewhere between those extremes. You use your reasoning to form some model of how the environment works, and try to plan out some strategy according to it. But, at the same time, you can’t be too certain your theory is correct, so you also copy strategies you’ve seen win in the past.

The downside of copying is superstition. You may end up copying elements which are superficially successful, but don’t matter at all.

Cargo cults represent the extreme of copying failure. These were islands who, during wartime, were used as air bases. The islanders, seeing the control towers and military officials bringing in supplies, started to copy them. They make fake control towers, fake landing pads, fake headphones, all to attract the cargo that they saw before. Unfortunately, none of these things work at all because the islanders had the wrong theory.

The downside of excessive theorizing is having a model of reality which turns out to be false. You follow the “correct” path but reality doesn’t work that way.

In my book club review of Seeing Like a State [2], I discuss how James C. Scott shows how excessive rationalization was behind many of the most catastrophic failures of the past century. Many modernization efforts failed completely because the theories just weren’t very good. Model cities, modernized agriculture and political systems which looked good on paper but failed in practice.

How to Make Good Decisions With Mediocre Theories and Examples

Copying, despite being seeing as a less sophisticated strategy, is actually not a bad starting point. This seems to be how human beings learn mostly anyways, and it embodies a lot of our cultural success. Find winners and do what they do, may seem simple, but it largely works.

That said, copying isn’t always possible, and sometimes you’ll waste a lot of energy copying the wrong things. In those cases, you want to adjust how much you copy based on how good a theory you have to offset it.

The more you learn, the better your theories will be. The more you understand probability, the better your poker game. The more you understand economics and finance, the better your investment decisions. The more you understand psychology and people, the better your relationship decisions.

However, it’s important here to note that, even if you learn everything the world has to teach you, many times your theories still won’t be all that good.

In those cases, you should deviate from the copying strategy only partially. Over-doing it—taking the theories you learn as being perfectly accurate—may make your strategy even worse if you switch completely from copying and try to create your own strategy from scratch.

Some Examples of Balancing Copying and Theorizing

Learning is a good example of an imperfect theory. There’s a lot we know about how people learn, and I have my own ideas both from informal experience and reading scientific research, that give me some sense of how to learn best.

However, I always try to use, as a starting point, how people successfully learn the thing I want to learn. If I want to learn languages, I start by looking at what people who have learned languages fluently actually did, not some abstract theory of memory or language acquisition.

Then, when looking at the strategy, I might make changes to that based on my theory of how learning works. When designing my Year Without English [3] project, for instance, I rejected a common idea that immersion needs to be 100% complete (meaning no asking questions in English to clarify grammar, nobody else can teach you using English, etc.). I felt confident enough to reject that as being probably a holdover from the fact that young children who learn languages well often do it that way, but that as an adult it made understanding harder.

I think this balance between copying and reasoning is often missed because people have theories for almost all areas of life, but there’s often not a clear sense of how good they are (nor, what the alternative is, if they aren’t very good.) Copying is often seen as being a stupid strategy, but I think it can often be quite intelligent when your theory of how something works is bad.