Skip to content
How to ThinkIn the Age of AI

For persistent problems, study the exceptions — when did…

When a problem persists across multiple attempts, identify exceptions where the problem was absent or reduced rather than analyzing why the problem occurs, as exception conditions are more directly actionable than problem mechanisms.

When This Fires

  • A problem recurs despite multiple attempts to fix it
  • You've analyzed a problem thoroughly but can't find an actionable intervention
  • Negative analysis ("why does this happen?") has produced insight but not change
  • Any persistent difficulty where you can identify at least one exception

Common Failure Mode

Dismissing exceptions as irrelevant: "That time didn't count because I was on vacation" or "The conditions were unusual." The unusual conditions are the data. If the problem disappears under specific conditions, those conditions are your intervention target — even if (especially if) they seem unrelated to the problem as you understand it.

The Protocol

When a problem persists across multiple attempts: (1) Stop analyzing the problem. (2) Ask: "When was this problem absent or significantly reduced?" List 3-5 specific instances. (3) For each exception, document the conditions: what was different about the environment, your state, the task framing, the people involved? (4) Identify conditions shared across multiple exceptions. (5) Engineer those conditions into your next attempt. The exception conditions are your intervention — they're more actionable and more likely to work than any theory about the problem's cause.