This is the defence almost everyone skips, and it is the one that pays off the deepest mechanism in the whole problem. Remember the Jevons trap: when a task becomes cheaper to do, we do not pocket the saving, we do more of the task, and the efficiency vanishes into extra output. This is the exact way AI-saved time turns into AI-inflated workload.
There is only one defence against it, and it is deliberately, almost rebelliously counterintuitive: cap your own output.
What is an output cap?
It is a decision, made before you begin, about what “enough” looks like for a given piece of work. Then you produce that, and you stop. You bank the saved time as time, rather than letting it flow automatically into a longer report, a fancier deck, or a fourth option nobody asked for.
The crucial word is “before.” If you decide what enough looks like only after you have started, the work will already have expanded to fill the time, and you will mistake the expansion for necessity. The cap has to be set while you can still see the task clearly, from the outside, before you are inside it polishing.
Why is this so hard?
Because every instinct and every incentive pushes the other way. More feels safer. A longer report looks more thorough. A fancier deck looks more committed. Stopping at “good enough” feels like laziness, especially in a culture that reads visible effort as virtue. So we keep going, adding passes and options and detail, and the saved time disappears into work that no one asked for and few will notice.
The honest truth is that most of that extra effort is invisible to its recipients. The third revision of the slide, the fourth option in the proposal, the extra paragraph of context, these rarely change the decision the work was meant to inform. They consume your reclaimed hour and add almost nothing, which is the worst trade available.
What is the principle underneath?
The economist Herbert Simon called it satisficing: choosing an option that is good enough against a clear threshold, rather than exhaustively optimising.1 Given the cost of the search itself, satisficing is often the rational choice, not the lazy one. Cal Newport draws the related line between quality, the best result within sensible constraints, then shipped, and perfectionism, the fear that prevents completion and dresses itself up as standards.2 An output cap is how you choose quality over perfectionism on purpose.
How do you set a cap?
Define the deliverable concretely before you start: the length, the number of options, the depth of analysis, the time you will spend. Make the threshold a real line, not a vague aspiration. Then, when you hit it, ship, even if a part of you itches to keep going. The itch is the Jevons trap asking for its tribute. Refusing it is the whole point.
The cap is what converts a speed-up into reclaimed time rather than extra work. Without it, AI just helps you produce a more elaborate version of the same thing in the same hours. With it, you produce what the task actually requires, in less time, and you keep the difference. Same tool, same task, opposite result, and the only variable is whether you decided in advance where to stop. Enough, chosen deliberately, is the most underrated productivity skill there is.
Frequently asked questions
What does capping your output mean?
It means deciding before you start what enough looks like for a piece of work, producing that, and stopping, rather than letting saved time flow automatically into a longer report or a fancier deck. It is the direct defence against efficiency turning into more volume.
Why should you cap your output?
Because of the Jevons trap: when a task gets cheaper to do, we do more of it and the saving vanishes into extra output. An output cap banks the saved time as time instead of letting it become a heavier workload.
What is satisficing?
Satisficing, a term from Herbert Simon, means choosing an option that is good enough against a clear threshold rather than exhaustively optimising. It is often the rational choice once you account for the cost of the search itself.
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
-
Herbert Simon’s concept of satisficing: choosing an option that is good enough against a clear threshold rather than exhaustively optimising, often the rational response given the cost of search. ↩
-
Cal Newport, Slow Productivity (2024): the distinction between quality (the best result within constraints, then shipped) and perfectionism (fear that prevents completion). ↩