Also, reasoning from first principles requires you to be aware of all of the "principles" that need to be taken into account.
Say you decide to build a ship. You calculate the strength needed for the backbone when the ship is sitting in water, you calculate the size of the motor to get efficient movement, you calculate the placement and size of rudders to make your ship as manoeuvrable as possible. You calculate the exact shape of the hull at the waterline, taking into account the flow of water at the desired speed. Your ship is going to be the most efficient ship in the history of the world!
But there's a problem. You forgot that occasionally your ship is going to encounter massive waves. Massive enough that half the ship can find itself out of the water as it crests the wave. And then the backbone of the ship breaks, and your wonder of efficiency becomes an artificial reef. Or you forgot to take oxidizing effects into account, and one fine day as you negotiate the port of Newcastle, your rudder's pivots catastrophically fail due to corrosion, the rudder falls off the ship, and your oil tanker runs aground creating a massive environmental disaster.
Every time Elon Musk does something that hasn't been tried before, he runs the risk of running into these types of issues - things that we just hadn't thought to think of. Maybe that hold down system that they have for launching Falcons will one day assymetrically fail, dragging the rocket over horizontally before failing totally, leaving a very large missile to shoot along the ground. Or maybe we're going to discover that all of those redundant engines that a Falcon 9 uses create a failure mode that makes one engine failing create a cascade of failures elsewhere, due to pressure imbalances, or an engine explosively damaging those around it, or whatever. Those fires that we've seen in Teslas recently? Guess what, the designers forgot to think of something.
Which is not to knock Musk's achievements - indeed, I'm a big fan. First principles are important, they're the only known antidote to cargo-culting, but when you hit out in a new direction because you think that you've found a flaw in the reasoning of previous designers, be aware that here be dragons and your idea may catastrophically fail in novel ways...
> Also, reasoning from first principles requires you to be aware of all of the "principles" that need to be taken into account.
Yes exactly. That's where most people fall down, or what certainly seems to be the meme in software, where a person's first principles approach is actually just a naive approach.
Say you decide to build a ship. You calculate the strength needed for the backbone when the ship is sitting in water, you calculate the size of the motor to get efficient movement, you calculate the placement and size of rudders to make your ship as manoeuvrable as possible. You calculate the exact shape of the hull at the waterline, taking into account the flow of water at the desired speed. Your ship is going to be the most efficient ship in the history of the world!
But there's a problem. You forgot that occasionally your ship is going to encounter massive waves. Massive enough that half the ship can find itself out of the water as it crests the wave. And then the backbone of the ship breaks, and your wonder of efficiency becomes an artificial reef. Or you forgot to take oxidizing effects into account, and one fine day as you negotiate the port of Newcastle, your rudder's pivots catastrophically fail due to corrosion, the rudder falls off the ship, and your oil tanker runs aground creating a massive environmental disaster.
Every time Elon Musk does something that hasn't been tried before, he runs the risk of running into these types of issues - things that we just hadn't thought to think of. Maybe that hold down system that they have for launching Falcons will one day assymetrically fail, dragging the rocket over horizontally before failing totally, leaving a very large missile to shoot along the ground. Or maybe we're going to discover that all of those redundant engines that a Falcon 9 uses create a failure mode that makes one engine failing create a cascade of failures elsewhere, due to pressure imbalances, or an engine explosively damaging those around it, or whatever. Those fires that we've seen in Teslas recently? Guess what, the designers forgot to think of something.
Which is not to knock Musk's achievements - indeed, I'm a big fan. First principles are important, they're the only known antidote to cargo-culting, but when you hit out in a new direction because you think that you've found a flaw in the reasoning of previous designers, be aware that here be dragons and your idea may catastrophically fail in novel ways...