Why does error detection fail?

The most common reason error detection fails: Conflating the feeling that something is wrong with the detection of what is wrong. Vague dissatisfaction is not error detection. It is an unprocessed signal that something in the system has deviated from expectation, but without specificity about what deviated, where it deviated, and by how much. People who treat 'this doesn't feel right' as a complete detection step skip directly to correction — changing things semi-randomly until the bad feeling subsides. This produces the illusion of fixing while leaving the actual error untouched. Genuine detection produces a specific, testable claim: this variable, at this point, deviated from this standard by this amount.

The fix: Choose one recurring output in your life — a report you write, a meeting you run, a decision you make weekly, a conversation type you repeat. For the next three instances of that output, add a 5-minute detection pass immediately after completion. Do not try to fix anything yet. Instead, write down three things: (1) What specific errors, if any, did you notice? (2) At what point in the process did the error likely originate? (3) How confident are you that your detection is catching the actual errors rather than surface symptoms? After three rounds, review your detection log. You now have data about your error detection capacity — its coverage, its blind spots, and its reliability.

The underlying principle is straightforward: You cannot fix what you cannot detect — invest in error detection mechanisms.