Reducing what needs judgment is the only move that scales
The only remedy that gets bigger as the problem does is making less of the work need somebody to judge it.
One move gets bigger as the problem does, and it is the only one. Make less of the work need somebody to judge it.
| the move | what it trades away | where it works best |
|---|---|---|
| narrower tasks — ask for less at once | some capability, and some convenience | anywhere the output is currently large enough that judging it means reading all of it |
| outputs checkable without understanding them | expressiveness. A structured answer is easier to check and says less | anything where the result can be tested rather than read |
| failures that surface themselves | engineering effort, up front | anywhere a wrong answer currently looks like a right one |
| producing less on purpose | throughput, visibly, and it is the least popular row here | where the marginal work was not worth judging and therefore is not worth having |
Each of these is a trade and the volume says so
None of them is free, and presenting them as free is how the plan gets adopted and then resented. Narrower tasks mean more round trips. Checkable outputs are less expressive. Self-surfacing failures cost engineering time that could have gone to features. Producing less is producing less.
What makes them different from part three's remedy is not that they are cheap. It is that they scale: each one reduces the amount of judgment needed per unit of work, so the benefit grows with production instead of having to keep pace with it.
The second row is where most teams have the largest gain available. A great deal of output is currently free-form because nothing forced it not to be, and structuring it converts judgment into a check. That is not a small engineering project and it is the one with the clearest payback.