The harshest line in Musk’s five-step method is not about automation. It is about deletion: “If you never have to add back something you removed, you are not deleting enough.” The five steps themselves — question the requirement, delete, simplify, accelerate, automate — are worth nothing; every engineering-management textbook carries them. The valuable part is the order. Not one step may be reversed.
Question first, then delete. Most requirements handed to you are, in Musk’s own words, stupid — whoever wrote them. What is harder to guard against is the requirement written by a smart person: people relax their scepticism and treat it as scripture. One round of questioning costs no money. It buys insurance.
Then delete. Every step kept “just in case” could just as well not exist, by the same just-in-case logic. Removing a tenth and adding some back is the normal score; never adding anything back is the sign of a soft hand. This is the only quantified standard in the whole method, and the easiest to skip — deletion produces no visible credit.
Only then comes simplifying and accelerating. Musk’s own metaphor is digging: if what you are digging is a hole, you do not want to dig faster. You want to stop. Acceleration amplifies direction — profit if it is right, loss if it is wrong.
Automation goes last. Musk admits the detour personally: automated, sped up, simplified, and only at the end realised the whole thing should have been deleted. Automation is cement. Whatever step you pour it on becomes permanent — and a wrong process running on machines is one nobody dares to touch again.
Why must the order hold? Because every earlier step saves money for the next one. Clever engineers keep jumping straight to optimisation, and the root cause sits in schooling: students are trained to answer questions. Questioning whether the question is stupid costs marks.
First objection: if it ain’t broke, don’t fix it; redundancy is a safety cushion. — A cushion is bought with cost, not free by default. The pile-up of “just in case” is the single largest source of process bloat, and the five-step method turns it from a slogan into an auditable action. Second objection: is acceleration not just another word for efficiency? — On the wrong direction, acceleration’s only achievement is a bigger mistake. The method exists to stop the bleeding from exactly this kind of diligence.
The boundary sits on both sides. The sequence fixes processes, not people, and not fires. In creative work where the requirement is still fog, step-one questioning can slide into infinite delay — build a rough prototype first, question it afterwards. At a burning site, run fast first; whether the ditch you draw water from is the right one can wait. Five steps are the ledger of ordinary days, not the emergency plan. A sense of order looks like fastidiousness in peacetime. After one mess, it looks like saving money.
