In any process that repeats, small errors accumulate. A measurement rounded to the nearest millimetre introduces an error that matters not at all for a single cut, a little for ten cuts, and enormously for a thousand. The carpenter who ignores this will find, somewhere deep into a large project, that the pieces no longer fit. The pieces were never wrong individually; the wrongness emerged from the compounding.

This is the structure of a great many historical disasters. No single decision was catastrophic; each was a small rounding error, a minor distortion, a slight deviation from the ideal path. The catastrophe was the accumulation — the point at which the compounded error exceeded the system’s tolerance for deviation and something broke.

What makes this hard to see in advance is that the early stages of accumulation look indistinguishable from robustness. The system absorbs error and continues functioning. The accumulated error is invisible precisely because the system has not yet failed; and the longer it continues functioning, the more confident its operators become that the error is not accumulating or does not matter. The failure, when it comes, appears sudden. It was only the visibility that was sudden.

The correct response to this pattern is not paranoia but auditing. Periodic, honest accounting of where the system actually stands versus where it is supposed to stand — not relying on the absence of failure as evidence of health. Systems fail on the day they fail; they become fragile in the long quiet period before.