Path-Dependence and Bottlenecks

Dates and Keyboards

Have you ever wondered why clocks are seemingly base 12? Or why we use the Imperial system over the metric system? Or why we know degrees instead of radians?

These are harmless, but here are some things that are not. The Gregorian calendar is bad. The QWERTY keyboard (which I’m using right now) was designed to prevent the jamming of keys, and otherwise makes no sense. Better keyboard designs exist. We don’t use them.

As a fan of efficiency, I have to moan at our alphabet. For some reason we preserved the redundant letters Q, X, and K, but not the Þ. (Þ as in Þe, the original spelling of Ye, a word actually pronounced the same as The). Don’t even get me started about English spelling.

Going back to the QWERTY keyboard, if it’s inefficient I say, then why do we still use it?

The reason new computer users are taught QWERTY is it’s the most common default keyboard. The reason most keyboards have the QWERTY layout is because most people know it. And most people know it because that’s what’s taught.

In short, it’s cyclical. To transition to DVORAK, you would have to re-train everyone, and also transition every keyboard, all at the same time. This is a coordination problem to the scale of switching a whole country to driving on a different side of the road. An incremental approach does not work.

Screen Shot 2019-03-29 at 9.07.40 PM.png


Screen Shot 2019-03-29 at 9.08.10 PM.png

The more you delay the transition, the more costly it will be. But it never feels like the right time, because of the cyclical effect.

Compounding everything

You have probably heard of the butterfly effect. That thought experiment is used to illustrate how minor events can have profound ripple effects.

No – “ripple effect” is a poor analogy because ripples die down as they get farther from the epicenter. The better term is “compounding.”

This is sometimes called the power law. I can only recommend you read the entirety of Chapter 7 from Zero to One, by investor Peter Thiel who is intimately acquainted with the power law.

I have shown in the past that the ability of things to compound leads to “black swan events.” This is merely an affect of recursion.

The popularity of trends can fluctuate for eons, ultimately for the winning trend to be the one that happens to be what’s popular when the black swan occurs.

Screen Shot 2019-03-29 at 9.00.12 PM.png

This is important to us because we have just experienced the precipice of an enormous black swan event: the industrial revolution.

What this tells us is that we are slaves to our past. Whatever lucked out by getting popular during the advent of industrialization is now burned into our society like it was always important.

Consider the artists who have had their names etched in history, like Mozart or Shakespeare. It is easy to argue that these were some of the most talented individuals of all time, but they have another thing going in their favor: they lived on the eve of industrialization! How much of fame is luck?

A black swan like industrialization leads to a lot of inventions. If you invent something that works, you can get widespread adoption. Once something has widespread adoption, you get the QWERTY dilemma: any transition represents a coordination problem. We are “stuck.”


Consider the following thee actions. A: killing half to the world’s population. B: killing all but about twenty people. C: killing every single person. From a point of view with an extremely long time horizon, B is actually more similar to A than it is to C. The final twenty killings are more significant than billions leading up to them. Why? Because the final twenty people can repopulate.

We come to this crucial Nassim Taleb article. He argues that statisticians and behavioral psychologists model risk incorrectly, modeled as the first population minus 20 killings, rather than the final 20.

You are offered a seat at a casino and told the rules of the game. Each time you play, you pay $200, then roll a die. If you roll a 1 you loose all of your money. If you roll anything else you double your money. How many times would you play such a game? According to the expected utility model, anything short of infinity is “loss aversion” until you’re broke. This is short term small mindedness that fails to take into account the game-ending nature of failure.

A roll of a 1 is like action C from the population example. You “go bust,” i.e., loose the ability to continue playing.

A situation like “B” drives of a lot of genetic changes. While “A” doesn’t have a strong enough effect, and “C” just ends the story, “B” can operate as a stopgap on compounding growth. What few beings emerge from the bottleneck will again enjoy compounding growth. Their particular traits, selected by the bottleneck, will reign supreme.

This brings us to gradual evolution vs punctuated equilibrium. I’m more partial to punctual equilibrium, given the effect of bottlenecks.

Screen Shot 2019-03-29 at 10.27.51 PM.png

A bottleneck is like a black swan event, but one that kills most of anything it touches. A successful trend will do two things. (A) capitalize off of positive black swan events, and (B) survive bottlenecks. This is difficult business, given that the timing of both A and B is almost completely unpredictable.

But it’s not all luck. The ability to not be fooled by unpredictability is itself a skill.