Notes on The Algorithm by Jon McNeill
Jon McNeill was the President of Tesla. He's seen Elon's operating system up close, and he's one of the few people who's distilled it into something usable for the rest of us.
The Algorithm is a five-step framework for getting any process down to its leanest form. It sounds simple. It isn't. Most of us skip steps 1 through 3 and rush straight to optimizing or automating something that shouldn't have existed in the first place.
These are my notes.
The 5 principles:
Question every requirement
Delete every possible step in a process
Simplify and optimize
Accelerate cycle time
Automate last
1. Question every requirement
Double-check every requirement.
Is that really necessary?
Was it a rule, a recommendation, or just something people followed blindly for centuries?
Is it absolutely needed, or a constraint we can break? Ask why again and again. Each constraint is an opportunity.
Any requirement that's supposed to apply to everyone must be questioned.
2. Delete every possible step in a process
Reduce the number of steps.
Review each step. Really drill into it. Anything that isn't truly needed, get rid of it. Question the legacy stuff, the status quo, anything that simply got bloated over time.
Get down to as few steps as possible. Then think really hard and delete some more.
3. Simplify and optimize
Optimize what's left. Make it simpler.
First understand the ins and outs of the process. Then map out each step. See how non-essential items can be combined or eliminated.
Note the order here. Simplification comes after questioning and deleting, not before. There's no point optimizing a step that should never have existed.
4. Accelerate cycle time
The magic of speed is that it accelerates cycle time, and faster cycle times mean higher throughput with the same fixed resources.
Think radically. Sometimes you have to create an entirely new process rather than accelerating the old one.
Speed wins. A process can only go as fast as its slowest step.
5. Automate last
Don't automate until you know exactly what you're doing. Every motion, every turn, the pathway of every component. With that knowledge, you optimize: shortening the routes parts have to travel, slicing seconds off jobs. Only when the flowchart is shorn to the simplest, most efficient form can you think about inviting in a robot or two.
It's okay (in fact, necessary) to do everything manually at first. Perfect it by hand. Only then automate.
A few generic notes that stuck with me
Seek radical solutions. Use more creative thinking.
Dig into everything we take for granted: rules, customs, accepted truths. That's where great opportunities are usually hiding.
Expand the definition of your product to the customer's entire experience, not just the thing you ship.
Inject urgency and accountability into the organization.
Elon always came well prepped for every session. That alone inspired engineers to bring their A-game and set the bar really high.
When the team sees the CEO and the President of the company working 18-hour days on a problem and sleeping at the factory, they put their all into it. It creates an atmosphere of urgent problem solving.
Eat your own dog food. Experience what it actually means to be the customer of your product.
Closing thought
The thing I keep coming back to is the order of operations. The whole framework breaks if you start at step 3. Most companies (most engineers, honestly) jump to optimization because it feels productive. But you can't optimize your way out of a requirement that shouldn't exist, or a step that should've been deleted three quarters ago.
Question. Delete. Then simplify. Then accelerate. Then, only then, automate.
Worth pinning somewhere visible.