True First Principles

The cheap version of first-principles thinking produces a posture. The real version produces a number.

That's the test I keep coming back to. If you've actually done the work, at the end you can hand someone an arithmetic answer that disagrees with the consensus. If all you have is "we rethought it from scratch" and a slightly different vibe, you reasoned by analogy in a costume.

Musk's rocket is the canonical example because the number is so specific. A rocket costs sixty-five million dollars. The aluminum, copper, carbon fiber, and titanium inside it cost about two percent of that. Everything between two percent and a hundred percent is convention, supply chain, and people who have been getting paid for thirty years to do it the same way. The first-principles answer is the two percent. Once you know that number, the rest of the company designs itself.

Most attempts at the practice never produce a number. They produce a deck. The deck has the words "from first principles" on slide three and a conclusion on slide twelve that any senior person in the industry would have written without the deck. There's nothing wrong with the conclusion. It just isn't first-principles. It's confirmation in fancier clothes.

The reason real first-principles thinking is rare isn't intellectual. It's that the process is socially expensive in a way the cheap version isn't. You have to write down every assumption sitting in your head about the problem and then, for each one, ask who first told you it was true. The list is always longer and dumber than you expected. Half of it came from a teacher you don't respect anymore. A quarter came from a tweet. The rest is industry folklore that hardened into physics in your head sometime in your twenties.

Then you have to delete the ones that don't survive contact with arithmetic. Every deletion is somebody's life work. They will defend it. You will spend a quarter being told you are naive, reinventing the wheel, or about to learn something the rest of the industry already learned the hard way. You might be. There is no way to know in advance.

The shape of a real first-principles answer is that it sits outside the Overton window of its industry. Reusable rockets. A battery factory the size of a small country. A drug priced at cost. If your answer would have been acceptable at any board meeting in your sector last year, you didn't get to first principles. You got to first conventions and then stopped.

I keep my own version small because the full thing is too expensive to do casually. Once a quarter, I pick one belief I act on without questioning, write it on an index card, and ask where it came from. Most survive. Some turn out to be a cargo-cult version of advice somebody gave me in 2018 about a problem I no longer have. The rest are the interesting ones.

The discipline isn't the thinking. It's keeping a number at the end. If I can't write the answer as a quantity, a cost, a ratio, or a deadline that disagrees with what everyone else in the room would have written, I haven't finished. I've just rearranged my opinions.