The Rule of Three

Definition

The Rule of Three is a completeness check on your own thinking: "If you can't think of three things that might go wrong with your plans, then there's something wrong with your thinking." The number three isn't magic — it's just large enough to force you past the one obvious risk everyone names and the second one that comes easily, into territory you hadn't actually searched. Failing to reach three is itself the diagnostic signal that you've stopped looking too soon.

In the Book

Weinberg introduces the rule after losing a bet to his colleague Judy over a "sure thing," having failed to notice an obvious missing case in his own reasoning. He generalizes it from a software-design practice: teams checking their designs for "what's missing" by demanding three candidate failure modes before declaring a design sound. The book then supplies concrete techniques for getting past the first two easy answers to a genuine third: arguing by analogy to an unrelated system (a technical-training program's problems became clear only via the analogy to animal training, revealing an unnoticed punishment structure), moving an attribute to an extreme (imagining zero employee turnover surfaced turnover's hidden benefit — the steady influx of new ideas), and looking at system boundaries, where "cracks" between components hide the problems each side assumes the other is handling. Weinberg also uses the rule as a review tool with clients directly — teaching it in classes and applying it to check whether a software rollout might fail in ways nobody had listed.

Why It Matters

Most flawed plans aren't sunk by the risk nobody could have imagined — they're sunk by the risk that was findable but wasn't looked for, because the first plausible answer felt sufficient. The Rule of Three converts "have you thought about what could go wrong" (which any confident planner will answer yes to) into a falsifiable test: name three, specifically. It travels well beyond consulting — into premortems, security reviews, editorial fact-checking, or any setting where the failure to generate a third alternative is more diagnostic than the three alternatives themselves.