Optimizing the Thing That Shouldn't Exist

From the Everyday Astronaut Starbase tour with Tim Dodd (July 30, 2021), where Musk lays out his “5-step algorithm” for engineering and operations. Step 1 in that algorithm is “Make your requirements less dumb” (question whether each requirement should exist at all), and this is the line he uses to motivate it. Walter Isaacson documents the algorithm in Elon Musk (Simon & Schuster, 2023), tying it to Tesla’s 2018 Model 3 production hell at the Fremont factory. The line circulates across secondary sources in slightly varied forms (“Possibly the most common…”, “The most common mistake…”, and Isaacson’s paraphrase “A common mistake is to simplify and optimize a part or a process that should not exist”); the wording above is the form most commonly reproduced from the video.

The point of the line is that the skill gradient runs the wrong way. A smart engineer is, by training, good at the optimization step. Faced with a system that does X poorly, the trained reflex is to make X faster, cheaper, or simpler. The reflex fires before the prior question gets asked: should the system be doing X at all? Talent makes the trap worse, not better. A more efficient engineer optimizes the unnecessary thing more thoroughly.

The closest established framing is Peter Drucker’s distinction between efficiency (“doing things right”) and effectiveness (“doing the right things”). Russell Ackoff put it more pointedly: doing the wrong thing right is worse than doing the right thing wrong, because the latter at least produces useful feedback. In the Lean and Toyota Production System vocabulary, the same failure mode is “overprocessing,” one of the seven canonical wastes (muda): work added to a product or process that the customer would not pay for, often because nobody asked.