The Planning Fallacy
Knowing you underestimate how long things take doesn't stop you from doing it. The only known fix isn't better planning. It's using someone else's past, not your own.
We finish about 60 percent of the way to our own estimates. Kahneman and Tversky's original studies on the planning fallacy found people miss their own deadlines by 30 to 50 percent, on projects they themselves scoped. The striking part isn't the miss. It's that being shown their past misses before the new estimate doesn't fix it.
The reason is how we build estimates. We imagine the specific path this time: the steps, the people, the dependencies. We don't imagine the interruptions, because we can't know them yet. So we price the plan, not the reality.
There's one fix with replicated evidence. Flyvbjerg calls it reference-class forecasting. Instead of asking how long will this take us, ask how long did five similar projects actually take. The base rate, not the narrative. Teams that estimate this way reduce overrun by 30 to 40 percent in infrastructure and product studies.
The nudge: Before approving the next roadmap, ask your team to list three past projects most similar in scope. Use the average of their actual durations as your estimate. Apologize to the plan.
