On the Tesla Model 3 line, a robot laid fiberglass mats on top of the battery pack. The robots kept jamming, so the engineers worked hard to fix them, making the machines faster and more reliable. The line still choked. They optimised the robot’s path. They accelerated the cycle. They automated more of it. And the battery pack kept holding up the entire car.
Eventually someone asked the question no one had asked. What are the mats actually for? Fire safety, came the answer. Checked, it turned out they were for noise and vibration. So they ran a test: a microphone in a car with the mats, and one without. Nobody could hear a difference. The mats did nothing detectable, so they deleted them, and deleting them removed an entire step that had required two million dollars of robotics.1 Engineers had spent months perfecting something that should not have existed.
Why is this the rule that matters most now?
Because the same mistake is about to happen to you, and AI is the robot. You have a pile of low-value work, and the most natural thing in the world is to point your shiny new tools at it and start doing it faster. Resist that with everything you have. It is the single most expensive mistake available. Before you make anything more efficient, ask what it is actually for. If the honest answer is “nothing,” you do not automate it. You kill it.
What is the principle underneath?
Peter Drucker drew the line in 1963: efficiency is doing things right, effectiveness is doing the right things, and enormous energy goes into doing the wrong things with great efficiency.2 Tape his sharper version to your monitor: there is nothing so useless as doing efficiently that which should not be done at all.3
Elon Musk turned the same insight into a procedure he repeated “to an annoying degree,” and the order is the entire point: question every requirement, delete the part or process, simplify, accelerate, and only then automate.4 Automation sits dead last for a reason. As Musk put it, the most common error of a smart engineer is to optimise a thing that should not exist.5
Why does AI make this more dangerous, not less?
For most of history, doing pointless work faster was hard, which put a natural brake on it. AI removes the brake. The temptation Musk had to resist with two million dollars of robots, you can now indulge in thirty seconds with a chatbot. The path of least resistance now runs directly toward optimising things that should not exist.
Picture the ordinary version. You have a weekly status report that takes forty minutes and that almost no one reads. The modern reflex is to have AI generate it in four and feel clever for saving thirty-six. But look at what you did. You made the pointless report permanent, because now it costs almost nothing to produce. You made it longer, because the machine fills space easily. And you pushed the cost downstream onto whoever now skims a bloated document for the one line that matters. Automating it was the worst option. Deleting it was free.
What should you do?
Run Musk’s question over every recurring task in your audit. What is this for, and who actually asked for it? Anything that survives, automate. Anything that does not, delete without guilt. Eliminate, then automate, never the other way around. The order is not a preference. It is the difference between a lighter week and a faster treadmill.
Frequently asked questions
Why should you eliminate work before automating it?
Because automating a task is the most expensive way to discover you should have deleted it. Doing the wrong thing faster is still wrong, and AI makes producing pointless output nearly free, so the temptation to optimise things that should not exist is greater than ever.
What is the eliminate-then-automate principle?
It is the rule that you must question and delete unnecessary work first, and only automate what genuinely survives that scrutiny. It echoes Drucker’s warning that there is nothing so useless as doing efficiently what should not be done at all.
What happens if you automate a pointless task?
You make it permanent, because it now costs almost nothing to produce, and often longer, because machines fill space easily, pushing the cost downstream onto whoever has to deal with the output.
About the author
Tom Goodwin is the author of Don’t Work Harder, a book about taking the time AI gives back as time rather than more work. He is a co-founder of GAMEPLAN and writes on productivity, technology, and the economics of the working week.
Footnotes
Footnotes
-
The fiberglass-mat episode is recounted in Walter Isaacson, Elon Musk (Simon & Schuster, 2023). Deleting the part removed a step that had required roughly two million dollars of robotics. ↩
-
Peter Drucker, “Managing for Business Effectiveness,” Harvard Business Review (May 1963). ↩
-
The line is widely attributed to Drucker and reflects the argument of his 1963 essay. ↩
-
Walter Isaacson, Elon Musk (2023): “the algorithm”, question requirements, delete, simplify, accelerate, automate. ↩
-
Musk, quoted in Isaacson, Elon Musk (2023): “the most common error of a smart engineer is to optimise a thing that should not exist.” ↩