Learning by Doing

I spent a few days last week on the campus of the University of Michigan, my alma mater. Always a fun, and humbling thing to do. The students there had a lot of good questions, and we had some good conversations.

I get asked a lot about how to build a successful career, and I never really quite like the answers I can give, because so much my own career was lucky, or privileged. I will often give advice in the direction of doing over thinking, which I believe in a lot. So, for example, if you want to be a coder, well, coders code, so write a lot. If you want to build products, you have a lot to learn from the market, so make it easy to do that - get your tools sharp and make it easy to conduct experiments quickly, things like that.

So I was happy when I was in one of these conversations with someone, and she pointed out some actual research in this direction. An experiment where two groups were asked to make pots - one group told to make as many as possible, and one group told to make a single pot but as well as possible. Guess which team made the best pot? Yep, the one that practiced a bunch and made a lot of them. (The story is here).

There’s a lot to unpack here. Why did the first team do so well? One important part of it was that they were given permission to experiment and fail. The instructions contained nothing at all about quality, just about making pots. This meant they were just focused on the process and learning by doing and didn’t have any internal editors second guessing them.

You can do this for yourself too! I often tell people “just follow your nose where it leads”. Again, I was lucky that this worked as well as it did, for sure, but it’s a really good technique for letting go of that inner editor and just building things. In fact, one of my favorite parts of any new, disruptive area is that no one is an expert yet, so it’s by definition OK to just try things and see if they work.

But of course, any area is always new to you when you first encounter it. If you can get into the mindset that the Japanese call “beginner’s mind” and let yourself just work without judgement, you will often find that you get better results more quickly (this is why that’s a concept - beginners often to surprisingly well because they don’t have to unlearn, and they aren’t second-guessing).

We are in a moment like this. We might not know whether or how generative AI is intelligent or useful, but there is no doubt that it’s complicated enough for there to be a debate. No one has the roadmap. You will likely get farther by trying to build lots of things, doing lots of “weird” experiments, than you will by sitting in a corner thinking things out until the time is up.

Make pots, lots of them. That’s the best way to make one perfect one.