2. Subtract

Marcus Aurelius wrote four words that I ignored for about a decade: if you seek tranquility, do less.

I nodded at the line on Instagram. I kept the 47 open tabs. What I didn’t understand until recently is that “do less” is not a vibe. It’s an instruction with a specific operation attached. Subtract.

Elon Musk turned the same instruction into five steps. He calls it his “algorithm” and runs it on every part and process at Tesla and SpaceX, as Walter Isaacson describes in his 2023 biography. In strict order: question every requirement, delete what you can, simplify what remains, accelerate the cycle, automate last.

The order is the whole point. The most common mistake is to skip ahead. People simplify a step that shouldn’t exist. They optimize a process that should have been deleted. They automate a workflow that was broken to begin with. Musk’s rule of thumb is that if you don’t end up adding back at least 10% of what you deleted, you didn’t delete enough.

Here is the loop for one person.

1. Question every requirement. Pick one workflow you repeat weekly. Write down every requirement in it. For each one, find the human who added it and ask why. Many “requirements” are not actually requirements. They are old assumptions nobody bothered to delete. If the answer comes back as “best practices” or “legal” or “safety,” find the actual person standing behind those words.

2. Delete what you can. Subtracting is harder than adding because everything you subtract has a defender. The app is only twelve dollars a month. The meeting has been on the calendar for a year. The client pays. Each individual item has a good reason to stay. If you respect every good reason, nothing ever goes. Cut anyway. Run a week without the thing. You can always add it back.

3. Simplify what remains. Only after you have deleted. If you simplify before you delete, you are polishing a mistake.

4. Accelerate the cycle. Now you can speed things up. Ship today, iterate tomorrow.

5. Automate last. Automating a broken process freezes the broken process. Run it by hand until it’s reliable. Then automate.

I run this loop on my own life about once a quarter. Not on parts of it. On all of it. Professionally and personally. I sit down and ask what I can kill. Not what I can optimize. Not what I can delegate. What I can kill outright. The answer is usually more than I’m comfortable with.

The word decision comes from the Latin decidere. To cut off. Every real decision is a small execution. You pick one thing and you kill the rest. People who can’t decide have not run out of information. They have run out of willingness to kill options.

The first time I ran the loop properly was 2024, after a stretch when I had every AI tool, every framework, and every option open and couldn’t finish anything. Unlimited options don’t feel like freedom. They feel like drowning. I shrank the game. One stack. One format. One week. One page. The work came back.

If you do exactly one thing from this chapter, do this: pick a workflow you repeat. Cut one step. Run it for seven days. If it breaks, add back the minimum fix. If it doesn’t break, you have learned something about your life that’s worth more than the step was.

What are you still refusing to cut?