Paying for education is a trap

The best are becoming much more productive than the average. Soon, if you’re not one of the best at what you do, you won’t be able to contribute to the economy. In such a world, most people who pay for an education won’t be able to recuperate their investment because most people won’t be one of the best at what they do.

Of course, there’s no limit to the kinds of work we can do. In theory every individual can be the best at something unique to them, that the world values. But in practice this won’t be possible anytime soon.

Better technology changes our work in a way that leads to greater variance in ability, regardless of quality of education. Better technology also increases compounding: small differences in ability lead to much larger differences in productivity. Thus, when technology improves, the gap in productivity between the best and the rest grows in tandem. We’re rapidly moving towards a “winners-take-most” world.

It will be difficult to tell who will become good enough and who won’t, no matter how good the education is. Investing capital and mentorship in individuals in exchange for equity in their success is the only viable business model for education in a “winners-take-most” world.

When a skill becomes more complex, we find that there’s greater variance in ability even in those with the highest quality training. An activity’s complexity is proportional to the number of possibilities. A more complex activity has more possible paths you can take, and thus more decisions, and more choices.

Startup founders and creators will always have a large variance in ability because what they do is very complex. Simple skills have little variance. The best factory worker is only moderately more productive than the average. Simple skills can be predictably trained for: an able person who spends a few months practicing will get good enough.

Even if the best entrepreneur or artist coached students 24/7 for a decade, we would still see great variance in ability in their students. Some would be far better than others because the skill itself is complex.

For any skill, if we could plot the ability of professionals on a graph, we’d notice a power law distribution. The more complex the skill, the more extreme the power law. The gap in ability between the best creator and the next best is much larger than the gap between the best factory worker and the next best. By many orders of magnitude.

When technology improves, the skills the world values become more complex, leading to greater variance in ability.

Better technology makes our skills more complex by eliminating the need for simpler skills, and by making complex skills even more complex.

Before computers, we spent a lot of time crunching numbers. As technology improved, we could spend more time figuring out what to create and how - which are much more complex skills than arithmetic.

Better technology makes what was once impossible possible, increasing the possibilities of what you can create. Building software got much more complex when the world adopted smartphones because there were many more possibilities in what you could build and who you could build for.

Better technology also increases the rate at which we make decisions, leading to more decisions and greater compounding. Small differences in ability lead to larger differences in productivity.

The gap in productivity between the best software engineers and the average has gone up significantly after GPT-4 because the complexity of engineering has greatly increased. There are more possible ways to engineer systems and the best ways are now much better than the average than they were earlier. The gap between the best and average software engineers is much higher now than a few years ago. When technology improves, the best go even further ahead. We can observe this trend in any kind of knowledge work.

It’s tempting to believe that even when the gap in productivity between the best and the rest grows, the average can still create value. They might, but what they will earn will be far less than what they did when the best weren’t as good. Eventually no one who’s average will be able to make a living.

When the best are thousands or even hundreds of thousands of times better than the average, you can’t justify hiring someone who’s average. Hiring 1000 average people to replace someone who’s great doesn’t work because the communication overhead of managing such a large team makes them far less productive than the individual great they’re trying to compete with. This wasn’t the case when the best were only, say, 10x as productive as the average. You could still hire average.

We’re seeing early signs of this trend: with a great leap in AI, Tech companies laid off many average employees, while simultaneously adding job listings seeking best - with compensations higher than ever before. We will see more rapid layoffs of average and a stronger demand for the best as the gap between the best and the average rises rapidly.

Any institution that charges money for education will crumble under pressure from both ends:

1. More graduates will fail to get hired because you have to be closer to being one of the best in order to be valued.
2. The cost of a quality teacher increases because fewer individuals will have the required skill to teach. Those who can will be more expensive than ever before because they will increasingly need to be one of the best at what they do for their teaching to be good enough.

Most bootcamps, paid online courses, and colleges will not survive. Those that do will not scale beyond a niche audience that doesn’t need a return on investment. The top colleges will linger around for a while because we are slow to update our craving for outdated prestige.

“Free” education that someone else pays for (taxpayers) also won’t work because they won’t have the success rate required to sustain them.

Leverage grows at different rates. We will see paying for education fail for software engineering much sooner than for manufacturing related skills. But, training for a simpler skill in a less rapidly changing domain is dangerous because software and AI will eat the world sooner or later. There’s nowhere to hide. It’s better to embrace the challenge than to shy away from it - especially if you’re young.

We will need to create the legal, cultural, technological infrastructure to enable investing in individuals similar to how we invest in companies (with some key differences that I’ll dive into later).

This is the only business model for education that works as the best become much better than the average leading to a “winners-take-most” world. The best knowledge will be available for free because it will come from those who are one of the best, who will have other means of wealth that far surpasses what they’d get from selling knowledge [1]. The best will give personalized feedback to those with potential, in exchange for the opportunity to gain equity (ownership) in their success. The best will also be incentivized to help train AI to scale personalized feedback because it will increase the pool of people they can invest in.

Paying for education is a trap. It doesn’t have the same value as it did in a far less complex world in which you could predictably train yourself to be good enough.

We are at the start of a big change in how the world will be educated, starting with a fundamental shift in the fundamental business model that powers education.

Notes

[1] As the best become much better than the average, we will increasingly find that those who charge money for their knowledge aren’t one of the best.