Elon's decision making: an anecdote compilation
I recently finished reading Walter Isaacson's Elon Musk. I liked the book's little anecdotes describing the way Elon is in his high and lows. I didn't come out of reading the book thinking that one has to be yelling at people and being obnoxious to get things done, though certainly being willing to have difficult conversations is something one needs.
Rather the main things I took from the book that make Elon Elon are two things:
-
Questioning life's default settings. If there's an option that others usually do, question why that option. You can usually buy a part. How much would it cost to make it? What if it's 30x cheaper. Someone might say they'll take 6 months to do something. Why not a day? (Founders Fund asks a similar question: How long until you can achieve goal X? and then Why not in half the time?)
- The algorithm: question requirements, remove parts, simplify and optimize, accelerate cycle time, automate
- By default some ideas are considered crazy. Too risky, hard, or complex to pull off. Elon will seriously consider them.
-
Being all in; knowing what's going on. Elon is all-in into his work. Other than playing videogames he has no hobbies. He spends his time learning about aluminium-lithium alloys, stainless steel, rocket propellents, the price of turbopumps, what are his companies up to, and so forth. In a set of slides I wrote a while back I had that "Management is 80% awareness". The same is demanded of others; he wants employees to know the weights and prices of the parts they are designing
The two together: If one has enough awareness (both of the issue at hand but also historical precedent and similar problems), when faced with a new problem it's easier to come up with novel ways to solve them (almost like as if one were a lookup table). If faced with a meeting discussing materials for a rocket, if you don't know the properties of stainless steel you may not try to vouch for using something other than an alluminium alloy.
Readers of Nintil will remember that I love concrete examples to illustrate points like this. Hard to even begin to talk about these points otherwise! Here I am trying to compile anecdotes from the book an elsewhere on unusual decisions that Elon has made either strategic or tactical (As opposed to instances of him say yelling at someone) and what the outcome was. If you have more, get in touch at [email protected] and send them my way :)
| Source | Anecdote | A success? |
|---|---|---|
| Elon Musk (Isaacson) | Elon had the belief that because humans can drive basically just with vision, cars should be able to. Many engineers insisted in adding other systems like radar and LIDAR but Elon argued that would add complexity and cost. While FSD is not here yet, it looks like Elon's approach is winning and that indeed with sufficient data and compute one can make cars drive themselves with just cameras | Yes, though Elon's insistence may have delayed the company's FDS plans |
| Elon Musk (Isaacson) | Put people's names into requirements. There may be a requirement like "Part X must be able to do XYZ" that other have to play with. But it can be forgotten why a requirement is in there and how serious it is or how context dependent. By linking requirements to individuals, one can ask the actual human being that generated a requirement why, and if it can be discarded or not in a particular case. | Probably useful but the consequences of this are not discussed much in the book |
| Elon Musk (Isaacson) | Elon wanted to move servers from a twitter data center in Sacramento to another in Portland. He was told it would take 6-9 months for a variety of reasons. He went to the data center, got together some people and U-Hauls and got the servers moved within weeks instead. | In a way yes and in a way no: the servers did get moved faster than the "default setting". However, Elon later judged that moving the servers (to close that datacenter) had been the wrong move |
| Elon Musk (Isaacson) | After PayPal, he wanted to send a greenhouse to mars. He thinks of buying old Soviet rockets to that end. When he couldn't do that, he calculated the cost of the raw materials involved in a rocket and saw that the real price of a launch was 50x cheaper after making a couple of spreadsheets (price divided by raw material cost is the "idiot index") | Led to SpaceX |
| Elon Musk (Isaacson) | In early SpaceX, when looking for parts they were quoted prices for parts Elon thought were insane and they ended up doing them inhouse or buying off the shelf components: valves (250k->a fraction of the cost), actuator (120k->5k). Later, for the Falcon 9, the HVAC system(3million->6k) | Yes |
| Elon Musk (Isaacson) | Tom Mueller (head of propulsion) gave Elon an estimate for how long it would take for a version of the Merlin engine. He then asked him to halve the schedule because "It takes so fucking long" and then asked him to halve it again. | No; the engine was developed in the timeline that Mueller had originally estimated |
| Elon Musk (Isaacson) | In early SpaceX, one of the fuel tanks developed a fissure. Instead of throwing it away and making another one (delaying progress by months) and against the advice of some of the engineers, they jut fixed it and kept using it. It worked | Yes |
| Elon Musk (Isaacson) | Same as above, but for a coating inside of the rocket engines. Tom Mueller told Elon his fix was insane and wouldn't work. It didn't work | No |
| Elon Musk (Isaacson) | Originally SpaceX was going to launch from Vanderberg Air Force Base but that meant working with US Air Force regulations and have some extra delays. Elon chose to launch from Kwaj, a tiny island in the middle of the Pacific instead. | No; per Elon years later the hellish logistics ended up being worse than dealing with the Air Force so Vanderberg would have probably been a better choice |
| Elon Musk (Isaacson) | To cut costs, the Falcon 1 rocket used cheap aluminum B-nuts that ended up getting corroded and causing the rocket's first flight to end up in an explosion | No; rocket failed to launch |
| Elon Musk (Isaacson) | At Tesla, Elon questioned the existence of a part that was holding back the line; he was told the part was there to reduce noise; he called the team in charge of that; they said it was there to reduce risk of fire. He asked them to record noise inside the car with/without the part. There was no difference and the part was removed. | Yes |
| Elon Musk (Isaacson) | In early Tesla Elon decided to go all-in on automation. It ended up being slower so he reverted back to human workers for many of the stations in the assembly line | No; he ended up having to remove lots of robots |
| Elon Musk (Isaacson) | During the "production hell" period at Tesla, Elon found out it was possible to set up tents outside the factory to set up additional assembly lines. They got built and started rolling out cars in 3 weeks. The tents remain there to this date. | Yes; more cars got built |
| Elon Musk (Isaacson) | Elon decided to make the underbody of new cars in a single part using a press (the same ways legos are made). All companies making such presses refused except one. The Gigapress was born and now other automakers like Toyota are adopting this. | Yes |
| Tom Mueller interview | Early in SpaceX history, when designing the Merlin engine, Elon pushed for removing a series of valves against the advice of Tom Mueller who said it'd be hard. Ultimately, it was possible to build the engines the way Elon wanted. | Yes, Merlin engines are cheap, reliable, and powerful. Tom Mueller says Elon made the right call. |
Changelog
- 2023-10-28: Addded one more anecdote. Thanks for Adam Comella!