A few years ago I read a short book on bridges. I gained a very basic understanding of the history of bridges, different types that exist today, aesthetic and utilitarian principles of bridge-building; then, I annoyed the hell out of my friends for a few months by pointing out details about bridges in the field of view. I no longer do this. Broadly speaking I think I am a Fox, rather than a Hedgehog1.
Hedgehog: There can’t be that many insights that change the world. Deeply study one or two of them, and apply them everywhere.
Fox: There is always more to learn. You have all your life completely ignored almost all of the millions of features in the world around you. Learn everything you can about the world and collect lots of different mental models each of which will call your attention to different things about the world.
Evidence suggests that Foxes make better predictions about the world, but I think that foxes are dumb too, without a meta-rule. Which features matter now? Which mental models should I apply? If you only have one rule—even if that rule is “accumulate lots of different types of knowledge, then apply the first three ideas that come to mind in a particular situation”–doesn’t that just make you a higher-level hedgehog? Is there a better way to do it?
Every TED talk, for example, is supposed to contain an insight that changes the way you see the world. But I have trouble believing that watching a ton of TED talks is actually making you smarter or, really, better at anything. Maybe most TED talks aren’t good enough, but that only raises more questions: how do you know how to learn about a new topic? How do you even pick topics that matter? And how do you know when you have learned “enough” about a topic?
If you just trust your intuition my guess is that you will pick out a few fields that you find really interesting (for arbitrary and path-dependent reasons) and then learn a lot about those. But then you get stuck in wheel ruts with diminishing marginal returns, but you keep investing mental resources by calling on the same few fields. For example, a few of the fields where I spend a lot of mental effort are2):
- Behavioral economics
- Presenting information clearly
- Quantitative data analysis
- Organizational psychology
So I see a sign next to an elevator which has the floor plan and instructs: “In case of fire, do not use elevator. Use stairs.” I think “hmm, people don’t read signs and aren’t great at reading maps, especially under times of high stress like when there’ s a fire. (Field 1) And the written instructions aren’t great either, why don’t they just say “If there’s a fire”? Worse, the map is super confusing–I don’t know how to read a floor plan at all, and there should just be a set of directions saying “Make two lefts to get to the stairs.” (Field 2.) I wonder how many lives would be saved if you changed to better signs, and how you could measure it, and how the costs and benefits would add up. (Field 3.) Finally, why is it designed in this particular way? Is it because of a regulation? Or is it because the corporations that put up the signs were cutting corners and it was a little tougher for one person to do it in a better way, or because little details like this are really hard for organizations to get right? (Field 4)”
It’s a simple sign–but actually a very rich source of interesting things to think about, given the few fields that I really care about. I must be missing out on so many interesting thoughts just because I am not well rounded enough? Why then did I decide to stop learning about bridge building and architecture and electrical engineering? One possibility is that I had actually gotten all the low-hanging fruit. After all, these fields might just be really boring and shallow. But this seems unlikely. Instead, it felt like I had gotten all the low-hanging fruit, because I wasn’t spending any time with people who knew more than me about these topics. So I knew enough to impress everyone I talked to, and I had no idea whether I was applying the concepts correctly or in interesting ways. Worse, I had no idea of what I was missing out on–how much was actually left to learn. So maybe you could start with what you know best and branch out slowly.
But this can’t be the right rule either, because ideas are more powerful when they are really fresh and new to you. Someone who knows everything about effective TV advertising will probably not gain that much from learning about effective ads on Hulu. The fields are close together so our TV expert had probably already encountered the Hulu insights in a different form, or seen their corollaries, or they’re already working from the same presumptions. So are they really learning that much?
I don’t have answers to all these questions, but I’ll start with a possible meta rule that seems pretty good: “Learn about lots of different things that are distant from what you already know but pick in advance which you will spend a lot of time on. Talk with people who know more than you, both while you are learning it and when you have more thoughts about it. Think a lot about which rules to apply in which contexts.” So I’ll start.
Readers, what is a field that you know a lot about that is worthwhile for me to pick up as a hobby?
The citation is to an essay by Isaiah Berlin, but I learned about the distinction in the work of Philip Tetlock ↩
Now, I think these are some of the best fields to care about. In part this is a cognitive bias: “I care about it, so it must be important”. However I think they are actually important because they’re “higher level” in some sense–if you want to think clearly, these are good ones to care about. But of course I can imagine analogous self-serving arguments for many other interest. I care a lot about cooking, and I justify it by saying “cooking is an instinctive and primal and important source of pleasure and social bonding”. But if you have a unique justification for every topic you like, then you haven’t gotten anywhere for defining a general helpful rule. You’ve just decided that you trust your intuition. Does that make you comfortable? ↩