Path Dependence—and Disruption—of Our Life’s Work
Path dependence is a powerful force in life. John Sterman’s excellent book Business Dynamics defines path dependence as, “a pattern of behavior in which small, random events early in the history of a system determine the ultimate end state, even when all end states are equally likely at the beginning.”1
What does this mean? Well, as a practical example, where you were born can have a large impact on the choices available to you. This study shows that’s even true at the neighborhood level within the same city.
Other early decisions with large impact include whether you go to college, your choice of major if you do attend, and economic conditions when you graduate.
But other factors, even subtler, also can determine a life’s course. A chance comment from a fourth-grade teacher that you’re good at history? That you’re not good at math? A sign-up for an art class because the accounting class you were supposed to take was full? An embarrassing moment delivering a speech in front of classmates?
In general, the earlier on your path an event occurs, the more influence that event may have, primarily because it may occur before you’re “locked in” to that path. Before you commit and lock in, any slight breeze has the potential to shift your course and set you on a different path.
After you commit, a hurricane-force gale can still blow you off course, but it would really have to try.
How about a visual way to think about path dependence?
Sure. A simple model can illustrate path dependence. It’s called a Polya process, and in its linear form, it goes something like this:
Imagine a jar containing one white stone and one black stone. You have a machine that spits out either a white or black stone into the jar when you press a button. The first time you press the button, you have a 50/50 chance of receiving either a white or black stone, since you already have 50% white stones in the jar and 50% black stones in the jar (one of each).
But after that, your odds change. If you receive a white stone on the first button-press, you now have two white stones and one black stone. With a Polya process, past events affect future probabilities, so the second time you press the button, you have a 66% chance of receiving a white stone and a 33% chance of receiving a black stone.
If you receive a black stone on that second button-press2 (with a 33% chance that this happens, remember), you’ll have two white stones and two black stones, so the next time you press the button, you once again have a 50/50 chance of receiving white or black. Here’s how that looks:
But if you instead receive a white stone on that second button-press (with a 66% chance that this happens), then you’ll have three white stones and one black stone, so the next time you press the button, you’ll have a 75% chance of receiving a white stone and a 25% chance of receiving a black stone. That sequence of events looks like this instead:
The takeaway: As time goes by and more stones are added to your jar, the proportion is likely to stabilize at some level, since it gets harder and harder for any single stone to shift the ultimate outcome. As Sterman puts it in Business Dynamics, “The more stones, the less likely there will be any movement away from the current proportion: the system locks in to whatever balance emerges from its early history.”3
But real-world systems aren’t usually linear.
True. In the Polya process described above, the system can stabilize at any proportion, whether 70/30 or 80/20 or 50/50.
In a nonlinear system, it’s different. There may be only a few stable equilibrium points, instead of every possible outcome being roughly equally likely.
For example, let’s say we start with one black stone and one white stone. We have a 50/50 chance of receiving either black or white next. But after that, the dynamics change in a nonlinear system.
If we receive a black stone on our first button-press, we now have two black stones and one white stone in our jar. If we were using a linear Polya model, we would have a 66% chance of receiving a black stone next and a 33% chance of receiving a white stone. And we would continue this process until we arrived at one of many possible stable equilibria.
But in our sample nonlinear model, let’s say our chance of receiving a black stone next is now greater than 66% (more than the proportion would indicate), and our chance of receiving a white stone next is less than 33% (less than the proportion would indicate). In that case, we are likely to converge to a situation where our jar is almost entirely full of black stones.
Conversely, if we received a white stone first, we now have two white stones and one black stone in our jar. Our chance of receiving a white stone next is now greater than 66% (more than the proportion would indicate), and our chance of receiving a black stone next is less than 33% (less than the proportion would indicate), so we are likely to converge to a situation where our jar is almost entirely full of white stones.
Nonlinear systems can favor extremes in this way, and lock-in can be rapid and difficult to overcome. In this nonlinear system example, you’re highly likely to end up with either a large pile of white stones or a large pile of black stones. The odds of maintaining a 50/50 ratio are vanishingly small (you’d have to draw white, then black, then white, then black, in an alternating pattern each time with no exceptions).
This nonlinear model can operate with any number of options to start—green stones, blue stones, red stones, orange stones. All other conditions held constant, the outcome is still likely to be convergence to a single path.
How about some real-world situational examples?
Right. Of course, the world is not a perfect Polya process laboratory. But imagine our hypothetical student whose fourth-grade teacher tells her she’s better at English than math. Maybe in the next year, fifth grade, she focuses a little more on English class and a little less on math class, so she gets an A in English and a B in math. Then the next year, she focuses a little more on English and a little less on math, and she wins a writing award from the local newspaper. And so on. And so on. Until she majors in English and never takes another math class after finishing her college core requirement.
Was she really better at English than math, or did she just lock in to that path? It’s hard to tell, because aptitude and practice are both necessary to excel. What’s clear is the outcome: she becomes a journalist instead of a mathematician.
Now imagine a college student who’s equally good at acting and computer science but chooses to pursue one or the other. Outliers aside (and yes, they do happen), that student’s future earnings are likely to be affected significantly by this early-career choice, set on either a high-income or a low-income path.
(Of course, because life isn’t a perfect experimental lab, the student can choose acting and later make it big in Hollywood, or he can choose acting and then later go to grad school for computer science, or he can choose computer science and continue to pursue acting as an avocation, or—possibly the most-hedged option—he can double-major and maximize future flexibility. More on that in a future article!)
The main point is that early choices are often hugely important in determining ultimate outcomes. If Warren Buffett’s initial investment choices hadn’t worked out, would he still be known as a great investor today? If Google had sold to Excite for $1 million in 1999, would we all be using Excite today, or some other different search technology?
What if we lock in to non-optimal paths?
That can happen! Path lock-in doesn’t mean the outcome will be optimal: maybe the system will stabilize at an equilibrium that no one in their right mind would have chosen if they started from first principles. For example, a 1950s decision to incentivize US businesses to offer employees health insurance locked in a path that, in our current era with lifetime employment a distant memory, seems bonkers and has been very difficult to change for the better.
Indeed, it takes a lot of disruption to shift future outcomes for an entrenched, dominant player. Once a path is set, breaking out of path dependence is hard. It often happens because of a shock that upends the status quo.
Imagine you’ve accumulated a bowl full of white stones, but it’s swept out of your hands by a 100-mile-per-hour gust. So, you snatch the empty bowl back from the wind and start almost from scratch: the bowl contains one white stone and one black stone. The next time you press the button, you have a 50/50 chance of receiving a black stone or a white stone. You receive a black stone, and then another black stone, and another. Soon enough, you’re locked in to a new and different path.
Evolution on Earth encountered this type of disruption when an asteroid crashed into the planet and shook up the dinosaurs’ dominance. The Black Death shook up the medieval feudal system. Covid shook up the 9-to-5 corporate office culture. These events are on wildly different scales, but they are all examples of path lock-in shifting in response to a significant shock.
What if I want to consciously shift paths?
It’s hard, but doable. The fact that life isn’t a closed Polya-process model with very limited options works in favor of people who seek change. Moving to another location can spur a shake-up of life path. Leaving a long-held job for a new position, career path, or sabbatical can, too. Even starting a new hobby can eventually open new doors. It may take a long time for small perturbations to build enough momentum to shift an otherwise-established path—but it can also be rewarding to try.
Right now, societies at large are locked in to some non-optimal paths, climate change being the most significant one, the rampant and ongoing rise of economic inequality being another. But societies also have overcome path dependence in the past, making strides toward greater equality and a larger middle class. Conflict resolution is, at its best, an attempt to unlock alternate, better paths for all parties.
Can we do it again? The question is whether societies—and the individuals who shape them—are willing to expend enough constructive effort to change the dynamics before our locked-in paths lead us to destructive ruin.
-<>-<>-<>-
Extra, Extra!
Three links from the depths of my bookmark archives; think of these as tangential extras for curious readers:
1. On Corpspeak - by Andre Cooper in Good Reason - why bland corporate jargon may have a purpose after all.
2. Epigenetic ‘Clocks' Predict Animals’ True Biological Age - by Ingrid Wickelgren in Quanta Magazine - telomeres weren’t the full answer to aging, but is epigenetics?
3. UK inflation will hit 18% in early 2023, says leading bank Citi - by Alex Lawson and Rowena Mason in The Guardian - why is the UK an outlier in this way? Brexit is not mentioned much but probably should be.
Sterman, John. Business Dynamics: Systems Thinking and Modeling for a Complex World, The McGraw-Hill Companies, 2000, p. 349.
Side note: I made a mistake and stated this as the third button-press on initial publication; mea culpa, sorry about that! We’re still on the second button-press. I’ll probably add some images to this article as illustrations in the next day or two. Update: I added images, later than expected due to travel, but they are there! And I think this is much clearer.
Sterman, Business Dynamics, p. 356.