College

The best builders tend to have both breadth and depth. They have seen enough of the world to notice unusual problems, and enough concentration to understand one of them properly.

College is unusually good for the first part. In computer science terms, it encourages something closer to breadth-first search. You encounter ideas and fields before knowing which of them will matter later. A popular example is Steve Jobs taking calligraphy in college, which later influenced the typography of the Macintosh.

Interesting ideas often come from seeing a problem through the language of another field. Diversity of experience gives you more of those languages, and for a similar reason, it is useful to be bored enough for connections between them to form.

As artificial intelligence becomes better at execution, the ability to come up with useful ideas becomes more important. But execution does not become less important. Startups are inherently competitive, and if the baseline quality of execution rises for everyone, the standard required to win rises with it.

Building a company is therefore much closer to depth-first search. Once there is something worth pursuing, you have to understand it more deeply and execute better than other capable teams pursuing the same opportunity.

That is why I think builders should stay in college. It gives you a few years to widen the set of things you understand before the work naturally becomes narrower and more consuming.

There is also an asymmetry in how valuable those years are. Starting work a year earlier adds relatively little to a career that may last forty or fifty years. But a year in your late teens or early twenties is a much larger fraction of the period when you are old enough to have real autonomy and young enough to do almost anything. Trading too much of that period for a marginally longer working life does not seem like a particularly good trade.