Planning Fallacy Pattern (Extended Buehler)

Roger Buehler's extended research on the planning fallacy shows it's not just optimism about time—it's a systematic failure to use base rates for comparable past projects, combined with a focus on the specifics of the current plan rather than the statistical reality of similar plans. The extended pattern: (1) a new project is evaluated based on its unique features ('this time we have a better team, clearer requirements'), (2) the 'inside view' generates an optimistic estimate, (3) even when base rates are known ('similar projects took 2x the estimate'), they're dismissed because the current project feels different, (4) the unique features that justify optimism turn out to matter less than the generic factors that drive all projects (coordination costs, scope creep, unknown unknowns), (5) the project follows the base rate. The Sydney Opera House (estimated 7 years and $7 million, actual 16 years and $102 million), the Big Dig, and virtually every software project ever demonstrate the pattern. Flyvbjerg's research on major infrastructure projects shows an average cost overrun of 28% with remarkably stable distribution across decades and countries.

When to use it

Before committing to timelines, budgets, or resource plans. When optimistic about a new project despite history of underestimation. When estimates feel based on 'if everything goes well' assumptions. During any project planning phase.

How it can help

Provides specific, proven de-biasing techniques that go beyond 'add a buffer.' (1) Reference class forecasting (Flyvbjerg/Kahneman): identify a reference class of similar past projects, use their statistical distribution as your estimate, then adjust only for factors with demonstrated predictive power. (2) Murphyjitsu: for each timeline assumption, ask 'If I learned this step took 3x longer, would I be surprised?' If not, plan for 3x. (3) Break the project into sub-tasks and estimate each—total of sub-estimates is usually larger (and more accurate) than the holistic estimate. (4) Have someone who is not emotionally invested in the project provide the estimate. (5) Record every estimate and actual, building a personal 'estimation calibration curve.'