First-principles thinking
What it means, how to apply it, and tons of examples of first-principles thinking in action
👋 Hey, I’m Lenny and welcome to a 🔒 subscriber-only edition 🔒 of my weekly newsletter. Each week I tackle reader questions about building product, driving growth, and accelerating your career.
P.S. Don’t miss Lennybot ✨ (an AI chatbot trained on my newsletter posts, podcast interviews, and more) and my new Swag Store (great gifts for your favorite PMs, or yourself!).
Q: I’ve heard your podcast guests occasionally mention “first-principles thinking.” What exactly is first-principles thinking, and how do I use it in my work?
First-principles thinking, or thinking from first principles, sounds a lot more complicated than it is. It’s simply a technique for approaching problems with a beginner’s mind. Instead of working within assumptions and what people around you “know” to be true, you do the hard work of figuring out what’s actually true and, thus, what’s truly possible.
There are many ways to apply the technique (which I’ll share below), but essentially you do this by asking questions, challenging people’s assumptions, digging further than other people, and going directly to the source to find out for yourself.
The concept of thinking from first principles comes from physics, is also known as ab initio, and encourages one to “start directly at the level of established science and not make assumptions based on existing models.” Aristotle phrased it as “the first basis from which a thing is known.”
Any talk of first principles must include a quote from Elon Musk, so let’s get it out of the way:
“We get through life by reasoning by analogy, which essentially means copying what other people do with slight variations. And you have to do that. Otherwise, mentally, you wouldn’t be able to get through the day. But when you want to do something new, you have to apply the first-principles approach.” —Elon Musk
Before we go further, it’s important to know that this technique takes a lot of effort. As Ozan Varol (former rocket scientist and best-selling author) explained in this interview with Farnam Street:
“The primary downside [to first-principles thinking] is it’s really difficult. […]
Reasoning by analogy, or copying what others are doing, is sort of like being a cover band where you’re playing somebody else’s music. Whereas with first-principles thinking, you go back to the fundamental raw materials of music, which are the notes, and then you build an original song from scratch. That is first-principles thinking.
It’s really difficult to do because a lot of what we do in life is informed by what we’ve done before, and also by what others are doing around us.
First-principles thinking, if you take it to an extreme, can be really inefficient, because we learn by emulating other people—[everything] from learning how to walk, learning how to talk, comes from copying others and modeling others.
Part of the difficulty comes from picking what to question, because you can’t go through life questioning every single thing you do. Picking what to question and also using knowledge in a way that will inform, not constrain, you.”
That being said, many of history’s biggest breakthroughs came from someone making the effort to think from first principles. This includes the classic examples of SpaceX and Tesla, the breakthrough that led to the transformer architecture driving today’s AI revolution, the invention of human flight, the discoveries of nuclear energy and general relativity, and so many more. It’s not an accident that history’s most important scientists come at problems in this way—with a beginner’s mind:
“I have no special talent. I am only passionately curious.” —Albert Einstein
“The greatest enemy of knowledge is not ignorance; it is the illusion of knowledge.” —Stephen Hawking
“Be less curious about people and more curious about ideas.” —Marie Curie
“Every sentence I utter must be understood not as an affirmation, but as a question.” —Niels Bohr
“To myself I am only a child playing on the beach, while vast oceans of truth lie undiscovered before me.” —Isaac Newton
“The task is not so much to see what no one has yet seen, but to think what nobody has yet thought, about that which everybody sees.” —Erwin Schrödinger
To make this even more concrete, let’s look at a bunch of examples across tech startups.
Examples, big and small
As you read through the examples, notice the way people go about thinking from first principles in practice. Essentially, they figure out what problem they want to solve, identify the levers that are keeping them from getting there, question every assumption about what’s possible within each lever, do the legwork to find out facts on the ground, and then act.
Example 1: Building the Square Cash Card—Ayo Omojola
Watch this five-minute clip and notice Ayo’s approach to problem-solving:
Notice something important that isn’t good enough
Go to the source (e.g. the factory) and talk to people to understand why
Keep asking questions until you “get to the end”
“And so what happens is you go ask somebody something, and they would give you an answer which is the thing that they believe to be true. They’re not lying and it’s not malicious, but it’s just wonk.
So you just have to keep pushing until you get to an answer. I don’t really know the right way to articulate this all the way, but basically, you can’t stop until you get to the end.” —Ayo Omojola, CPO at Carbon Health
The way that I apply this today—and I’m sure people at Carbon Health will tell you this—is I end up asking lots of questions that people think don’t matter. Because I’m like: hey, we’re trying to optimize something. And when you’re trying to optimize something for the first time, you have to look at it like 15 different ways. And then every time two things are incongruent, you have to go and figure out why. It’s just tedious work.”
Example 2: Building OpenAI—Ilya Sutskever
Watch this two-minute clip and notice that the breakthrough that led to GPT (possibly the biggest breakthrough in software history) was rooted in experimenting with an idea that “everyone” knew wasn’t going to work.
“Everybody knew that you cannot train deep networks. It cannot be done. Back propagation is too weak. You need to do some kind of pre-training of some sort and then maybe you’ll get some kind of an oomph. […] Today we take deep learning for granted. Of course a large neural network is what you need. You shove data into it and you’ll get amazing results. Everyone knows that. Every child knows that. How can it be that we did not know that? How could such an obvious thing be not known? People were really focused on machine learning models, where they could prove that there was an algorithm which can perfectly train them, but whenever you put this condition on yourself, and you require to find a simple elegant mathematical proof, you really end up restricting the power of your model.”
For more, here’s another video of Ilya sharing this same lesson.
Example 2b: The key to Stripe and OpenAI’s success—Greg Brockman
In this one-minute clip, notice how a company like Stripe, one of the most valuable tech startups today, didn’t do anything that special. They just built a much better product than was previously available, simply by “not being locked into the way that people had been doing it” before, by thinking about every single piece of what they were doing from the ground up.
“A lot of how we approached Stripe was thinking from first principles. I remember when we were pre-launched and we had some buzz going because we had some early customers, and one of my friends took me out to lunch. He was a VC, and he was like, ‘All right, look, I’ve been hearing about this Stripe thing. What’s your secret sauce?’ I was like, ‘I mean, we just make payments really good.’ And he’s like, ‘No, no, come on, you can tell me, what’s the secret sauce?’ And that really was the secret sauce. We rethought every single piece of what we were doing from the ground up, from first principles. Not locked into the way that people had been doing it. We asked how should it be? Where’s the pain and does it need to be there?
In AI, we did much the same thing. We thought about, OK, there’s this field that we’re entering and that we hire a lot of people who had been in the field, but a lot of us also hadn’t been in the field, and we came to it with beginner’s eyes. That approach of just not being beholden to all the ways people were doing it, but also becoming expert in the way that things have been done. Because if you just throw everything out, you’re also just going to be starting from scratch, in a not-helpful way.”