Memetic mistakes

I have a friend who likes to read and is interested in history. She is kind, helpful and organized. What’s more likely – is she a Doctor or a Librarian? Most people associate her characteristics with being a Librarian, but they’re wrong. There are 5-20 times more Doctors than Librarians, so she’s far more likely to be a particularly bookish Doctor. In choosing the label of Librarian, based on the ‘memes’ of that profession, we’re committing what behavioral economist Daniel Kahneman calls ‘base rate neglect’. We saw this happen at a certain stage of the pandemic, with claims that most of the people dying of COVID were vaccinated. In a population where the vast majority of people were vaccinated, that’s completely to be expected. Most of us are biased in this way, weighing the information right in front of us rather than making an estimation of relative proportions.

Of course we have to make these kinds of generalizations just to get through our day. If we stopped and waited for enough information to make every decision, it’d be like navigating by measuring every square inch of a trail, instead of just using a map. “A map is not the territory” as Korzybski says, but “if correct, it has a similar structure to the territory, which accounts for its usefulness”. Most of the time our generalizations don’t fail us: our guesses are good enough to get by. When memes get divorced from reality, we risk context collapse and a rapid decline in usefulness. If all of your information is arrived at by analogy (i.e. borrowing someone’s map), and you never do your own calculations from first principles, you risk getting it spectacularly wrong. Physicist Richard Feynman wrote of Cargo Cults – islanders saw airplanes deliver supplies during the war, so they “arranged to make things like runways, to put fires along the sides of the runways, to make a wooden hut for a man to sit in, with two wooden pieces on his head like headphones and bars of bamboo sticking out like antennas” – they’re doing all the right things, but no airplanes land.

Famous investor Charlie Munger calls this ‘chauffeur’ knowledge, after Max Planck’s chauffeur, who was able to stand in for Planck having memorized his lectures, until he was asked a question. You can often get remarkably far by imitating the observable attributes of something, but without understanding the data-generating mechanism underneath, you’ll quickly go off track. This is one of the reasons why copycat companies almost never beat the original: they might get the first version more or less right, but if they don’t understand why it works, they won’t correctly extrapolate on where to go next. Eventually they get outpaced by the real innovators, and they may not even understand why. Do the work to gain a first-principles understanding, and you’ll spot opportunities where incumbents are using an outdated map. To paraphrase Robert Frost, “take the road less traveled, and it’ll make all the difference”.

That said, copying from successful examples is a necessary first step in learning something new, according to the Dreyfus model. You have to learn what it feels like to do great work, by first going through the motions. Only then can you spot patterns of what works (or doesn’t), and formulate rules to follow. Once you learn the rules you can start to break them, and invent new ones as you begin to achieve mastery. The trick is to not stop at copying. As Arno Rafael Minkkinen says, “Stay on the f*cking bus”. Your route might be adjacent to others at first, but eventually your path will diverge as you find yourself. Eventually even your earlier derivative work will retroactively be held up as signs of your inevitable success.

About the Book 📕

← Back to Essays 📝