How Musk deals with complexity

Elon Musk has given a personal tour of SpaceX’s Starbase at Boca Chica in Texas to a Youtuber, Tim Dodd aka the Everyday Astronaut. You can see the first part of the interview here.

Musk’s interview is in his capacity as Chief Engineer at SpaceX. Musk is also the CEO of Tesla and a founder of Paypal, and currently the world’s third wealthiest person. The interview is fascinating throughout.

What I find most interesting is seeing how Musk handles massive complexity. As the Chief Engineer at SpaceX, part of his job is getting the left and right hands talking to each other across the processes that support the design and manufacturing of a rocket.

His interview reveals some elements of his strategy: a clear objective; a small number of metrics to guide decisions; awareness of how problems emerge in complex design and manufacturing; preserving the option to experiment (and fail) cheaply because some things are only discoverable through trial and error; solve the problems that need solving first.

I can’t help but think some of these strategies for dealing with complexity in scalable systems is relevant to public policy. Officials and ministers are not building rockets or cars. But they have to make sense of practically unlimited complexity. I don’t know what strategies officials use to manage the complexity they deal with. Whatever they are, few of those strategies seem to make it into public discourse.

Musk makes the interesting observation that because every Space Shuttle flight had people on board, failure was intolerable, experimentation was costly, and innovation essential stopped as a result. Keeping the doors open to experimentation which does not hurt anybody is important. That seems like a relevant idea for public policy.

Here are some of Musk’s insights from Part 1 of his interview with Dodd.

Objective

Become a multiplanetary species

Metrics

  • $/tonne of thrust: “What is hard is how do we make a Raptor [rocket engine] where the cost per tonne of thrust is under $1,000”
  • $/tonne to low Earth orbit: “The fundamental thing that needs to be fixed is the cost per tonne to orbit”
  • $/tonne delivered to Moon, Mars

Musk talks about thrust in tonnes, rather than the more correct measure Newtons, to make it easy to compare with the mass of the rocket the engines propel. He is happy to sacrifice correctness for clarity and simplicity.

On fuel mix, the bias is in favour of oxygen over methane because oxygen is cheaper and more dense, which reduces cost per tonne.

Process

Musk has a five step process for managing complex processes.

1. Make your requirements less dumb. “It is particularly dangerous if a smart person gave you the requirements. Because you might not question them enough.”

2. Try very hard to delete the part or process. If you are not adding things back in 10% of the time you are not deleting enough. The bias is to keep things just in case. You can make that argument for almost anything.

3. Simplify or optimise. This is the third step not the first because a common error of a smart engineer is to optimise a thing that should not exist.

4. Accelerate cycle time. But only after the first three steps. Otherwise you are just “digging your own grave faster.”

5. Automate.

“I have personally made the mistake of going backwards on all five steps. Multiple times.” Musk tells an interesting story of fibreglass mats in manufacturing Tesla cars.

“Any requirement or constraint must come with a name not a department.” Otherwise some person years ago could randomly come up with a requirement that today nobody supports. The requirement will persist if it is not owned.

Other points

“All designs are wrong, it’s just a matter of how wrong.”

“Everyone is wrong some of the time.”

If you look at the various reasons for Starship’s landing failures, none of those reasons were on prior risk lists.

Musk has more than one optimisation metric, but he treats them in a sequence. “Fundamentally, the optimisation is for cost per tonne to orbit, and ultimately cost per tonne to the surface of Mars.”

A common problem in production is too much testing. The production line will be tested at each stage to identify where problems are occurring. Once the problem is diagnosed, it is common for testing to not be removed.

Production processes (manufacturing rocket parts at scale) requires 10-100 times more effort than the technology. “Manufacturing is underrated. Design is overrated.” So rocket science is the easy bit. Mind blowing.

Boosters and ships (the drone ships boosters land on) are easier to build than the launch pad. The pad includes the mechanism for catching returning first stages. The pad is going to catch them from the air.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s