Small Changes Can Be Reviewed And Small Changes Can Be Reverted

Size is not a stylistic preference. It determines whether the safety mechanisms work at all.

Two of the strongest safety mechanisms available to a working team depend entirely on the size of a change. Review depends on it because attention is finite: a reader can hold a modest change in their head, follow its consequences, and form a real opinion. Reverting depends on it because a small change is a clean unit of undo. If something goes wrong at three in the morning, the ability to remove one contained piece of work and restore known behaviour is worth more than any amount of analysis. Both mechanisms degrade sharply as size increases, and they degrade quietly, without anyone announcing that the protection has stopped working.

Keeping changes small is mostly a sequencing skill rather than a discipline of restraint. A piece of work that seems inherently large can usually be staged: prepare the ground in one change that alters no behaviour, introduce the new path alongside the old in another, move traffic across in a third, remove the old path in a fourth. Each stage is separately reviewable and separately reversible, and at no point does the system exist in a half converted state that only one person understands. This takes longer to plan and less time to recover from, which is the trade most teams should be making and few make by default.

The objection is that splitting work adds ceremony and slows delivery. Sometimes it does, and there are situations where a larger change genuinely is the honest unit, particularly when a partial state would be worse than either end. But that case should be argued rather than assumed. The common pattern is not a genuinely indivisible change. It is a change that grew because unrelated tidying attached itself along the way, until the reviewer faces a mixture of substance and noise and can no longer tell which is which. That mixture is the real cost, and it is avoidable.