Elon Musk - Book Notes

Categories: Study 2024-04-14

I recently finished reading “Elon Musk” during my morning reading time. I highly recommend it to fellow engineers. Although Musk sometimes acts strangely on X, the book reveals him as a true engineer, a genius skilled at combining technology, business, and cost control, but also a person with many flaws. I’m jotting down some notes for my own reference. After reading this book, as a Tesla shareholder, I am willing to continue holding.

The Five-Step Algorithm

1. Question every requirement. Every requirement should come with the name of the person who made it. Never accept a requirement from a department, like “the legal department” or “the security department.” You need to know the name of the person who made the requirement. Then you should question it, no matter how smart that person is. Smart people’s requirements are the most dangerous because people are less likely to question them. This should be done continuously, even if the requirement comes from me, Musk himself. After questioning, everyone should improve the requirement to make it less stupid.
2. Delete as many parts and processes of the requirement as you can. You may have to add them back later. In fact, if you end up adding back less than 10% of what you deleted, you haven’t deleted enough.
3. Simplify and optimize. This should come after step 2, because a common mistake is to simplify and optimize a part or process that shouldn’t exist in the first place.
4. Accelerate the cycle time. Every process can be accelerated, but only after following the first three steps. At the Tesla factory, I mistakenly spent a lot of energy accelerating production processes, only to realize later that some of them should have been eliminated.
5. Automate. A big mistake made at the Nevada and Fremont factories was that I tried to automate every step from the beginning. We should have first questioned all requirements, deleted unnecessary parts and processes, screened out and dealt with problems, and then pushed for automation.

Some corollaries sometimes derived from this algorithm

1. All technical managers must have hands-on experience. For example, managers of software teams must spend at least 20% of their time programming. Managers of the solar roof business must spend time on the roof doing installations themselves. Otherwise, they are all talk and no action, like cavalry captains who can’t ride horses, or generals who can’t wield swords.
2. “You’re good, I’m good, everyone’s good” is dangerous. People will stop questioning their colleagues’ work. People have a natural tendency not to want to kick a good colleague off the boat, and this dangerous tendency must be avoided.
3. It’s okay to make mistakes, but it’s not okay to refuse to admit it.
4. Never ask your team to do something you are not willing to do yourself.
5. Whenever there is a problem to be solved, don’t just talk to the relevant person in charge you directly manage. To conduct in-depth research, you must communicate across levels, go and talk directly to your subordinates’ subordinates.
6. Hire for attitude. Skills can be taught, but it’s too much effort to change a person’s work attitude, it’s like giving them a “brain transplant”. A crazy sense of urgency is the operating principle of our company.
7. The only rules to follow are those that can be derived from the laws of physics. Everything else is just a recommendation.

Strong Cost Awareness

1. For decades, cost-plus contracts have made the aerospace industry indifferent to cost control. A valve on a rocket costs 30 times more than a similar valve used in a car, so Musk constantly pushed his team to source parts from companies outside the aerospace industry. A latch used on the space station by NASA costs $1,500. A SpaceX engineer modified a latch used on a bathroom stall door to create a locking mechanism that cost only $30. An engineer walked into Musk’s cubicle and told him that the air cooling system for the Falcon 9 payload fairing would cost more than $3 million. He yelled at Gwen Shotwell in the next cubicle: How much does a home air conditioning system cost? She replied about $6,000. The SpaceX team then bought some commercial air conditioning equipment, modified the pumps, and used it on top of the rocket.
2. Ever since Musk’s trip to Russia, where he calculated the cost of building his own rockets, he has popularized a concept internally called the “idiot index,” which is the ratio of a component’s total cost to the cost of its raw materials. If the idiot index is high, for example, a component costs $1,000, while the aluminum to make it costs only $100, then it is likely that the design is too complex or the manufacturing process is too inefficient.
3. One night, lightning struck a test stand, damaging a tank’s pressurization system, causing a bulge and crack in a layer of the tank. A normal aerospace company would have replaced the tank, which would have taken months. “Don’t replace it, just fix it,” Musk said. “Go up there with a hammer, hammer out the bulge, weld the crack, and get on with it.” Buseza thought this was crazy, but he had learned to just listen to his boss. So they went to the test stand and hammered out the bulge. Musk jumped on a plane, flew for three hours, and personally supervised the whole process. “When he showed up, we started testing the fueled tank, and there were no problems,” Buseza said. “Elon believes in paving a road when you encounter a mountain and building a bridge when you encounter a river. There is always a way to fix any situation. This has benefited us a lot, and the whole process has been very interesting.” This also saved SpaceX months of time to test the rocket prototype.
4. We should ask everyone here to see if they can reduce the cost of the parts they are responsible for by 80%,” Musk suggested. “If they can’t, and others can, we should consider replacing them.

Crazy Management Style

1. Typically, in such unsettling moments of success, Musk would deliberately create drama. He would launch a whirlwind campaign, immediately go into action, and announce a deadline that was usually not only unrealistic but also unnecessary. Autonomy Day, Starship rocket stacking, solar roof installation, Tesla production hell, each time he sounded the alarm, forcing everyone to conduct a “fire drill.” Kimball said, “Usually, he would walk into one of his companies, stir things up, and turn it into a crisis.” But this time Musk didn’t do that. Instead, he decided to buy Twitter without much thought.
2. However, driving change is not so gentle. Since the Falcon 1 rocket, Musk’s methods have always been the same: maintain rapid iteration, take risks, act rudely, accept phased failures, and try again.
3. From some perspectives, Musk is like Steve Jobs, a capable but unsociable “foreman” whose reality distortion field can drive his subordinates crazy, but also push them to do things they thought were impossible. They can be disrespectful to both colleagues and competitors. But Tim Cook, who has been in charge of Apple since 2011, is different. He is calm, highly disciplined, and gentle. Although he is tough when necessary, he always avoids unnecessary conflicts. Jobs and Musk seem to be easily attracted to dramatic conflicts, while Cook has an instinct to defuse them—his moral compass is very stable.
4. I calculated that he would have to manage six companies—Tesla, SpaceX and its Starlink division, Twitter, the Boring Company, Neuralink, and X.AI. This is three times the number of companies Steve Jobs helmed at his peak (Apple and Pixar).
5. Whenever he decides to hire or promote someone, Musk holds the view that a person’s attitude is more important than their resume or the skills they possess. His definition of a “good attitude” is a person who desires to work like crazy and work hard. So Musk hired Denchev on the spot.
6. Machine learning systems usually need a goal or metric to guide them during self-training. Musk likes to decide which metrics are most important by giving orders. So he gave them a standard: the mileage a Tesla fully autonomous vehicle can travel without human intervention. “I want the first slide of every meeting to show the latest data on the mileage driven before each intervention.” Musk said. “If we are training AI, what should we optimize for? The answer is to increase the mileage between interventions.” he told them. It’s like playing a game, you can see your score every day. “A video game without a score is boring, so watching the mileage per intervention increase every day is very motivating.”

Attitude Towards Products

1. This is a very typical Musk-style approach. The designers who draw the styling blueprints for a new car should work hand-in-hand with the engineers who can determine the car’s manufacturing process. von Holzhausen said, “In other companies I’ve worked for, some designers are in a state of ‘washing their hands of it’. They have a design idea, and then they send it to the engineers, who may be working in another building, or even in another country.” But Musk would have engineers and designers in the same room. von Holzhausen said, “The goal of doing this is that we want to cultivate a group of designers who think like engineers, and a group of engineers who think like designers.”
2. Musk has a core idea, which is not to separate engineering and product design. In fact, product design should be driven by engineers. Like Tesla and SpaceX, Twitter should be engineer-oriented at all organizational levels.
3. Musk adopted an iterative design approach: rapidly prototyping rockets and engines, testing them, blowing them up, modifying them, and trying again until he finally made something that worked. Move fast, blow up prototypes, and repeat the process. You don’t need to perfectly avoid problems, the key is how quickly you can find and solve them.

The Story of Acquiring Twitter

1. Musk has a keen intuition for engineering problems, but his brain is not so good at handling interpersonal emotional issues, and this was the problem with his acquisition of Twitter. He thought it was a technology company, but in fact it belongs to the advertising and media industry, an industry whose logic is based on emotional communication and interpersonal relationships.
2. Musk told them that in the future, Twitter’s team would be led by engineers like them, not by designers and product managers. This was a subtle shift that reflected his values: Twitter’s essence should be a software engineering company, and its leaders must be people familiar with code; it is not a media and consumer product company, and should not be led by people who are proficient in interpersonal relationships and have insight into human desires.
3. When Musk is angry, he often throws out very specific questions to stump people. Facing the Twitter executives, he threw a barrage of questions. How many lines of code did their software engineers write on average per day? His self-driving team at Tesla has 200 software engineers, why does Twitter have 2,500? Twitter spends $1 billion a year on servers. What are the features that consume the most of this computing time and storage space? How are these features prioritized? He found that they had a hard time giving straight answers. At Tesla, some people lost their jobs for not knowing these details.