Reference class forecasting

Before estimating, look up how long similar past projects actually took — not how they felt.

Why it works

The planning fallacy is driven by an "inside view": you imagine the specific steps of this project and anchor on the plan. Reference class forecasting forces an "outside view" by asking: among projects like this one, what was the distribution of actual completion times? The outside view corrects for the systematic optimism bias because it encodes the full population of outcomes, including the long tail of delays that the inside view ignores. Kahneman credits the outside view as the single most reliable debiasing move for planning.

How to do it

1. Name a reference class: comparable past projects in type, scope, and team.
2. Pull the actual completion times from that class — not your memory of them, but records.
3. Anchor your estimate on the median or 75th percentile of that distribution, not on your current plan.
4. Adjust modestly for genuine unique features, then stop adjusting.

Evidence

Kahneman and Lovallo (1993) formalized the inside/outside view distinction and its effect on planning. Flyvbjerg and colleagues documented reference class forecasting in large infrastructure projects and found it consistently reduced cost and schedule overruns. (observational)

Most evidence is from large projects or lab studies; field evidence on personal tasks is thinner, though the mechanism is the same.

Sources

• Kahneman & Lovallo (1993), Timid choices and bold forecasts: A cognitive perspective on risk taking, Management Science
• Flyvbjerg (2008), Curbing optimism bias and strategic misrepresentation in planning, European Planning Studies

Common mistake

Treating the reference class as a starting point to argue away from ("but my situation is different") — the whole point is that every inside view feels unique.