How Big Things Get Done

In the AI Story

The book's core empirical finding has the force of diagnosis. Of the sixteen thousand projects in the database, only 0.5 percent came in on budget, on time, and delivered the promised benefits. The remaining 99.5 percent failed on at least one dimension, usually all three. This is not incompetence distributed across isolated cases; it is a structural pattern that demands structural explanation. The explanation Flyvbjerg offers — the twin engines of optimism bias and strategic misrepresentation — organizes the empirical record without reducing it to any single cause.

The book's practical advice is encoded in several memorable heuristics. Think slow, act fast: spend extensive time in planning, testing, and reference-class comparison before committing, then execute rapidly once commitments are made. Take the outside view: resist the inside view's felt specificity and calibrate against the statistical reality of comparable cases. Get your team right: the quality of the people matters more than the quality of the plan. These heuristics are Flyvbjerg's translations of his academic findings into operational guidance that practitioners can apply.

The AI-era relevance is direct. Every diagnosis the book offers for megaproject failure — optimism bias, strategic misrepresentation, uniqueness bias, the inside view's systematic distortions, the political incentives that reward inflated forecasts — applies with equal or greater force to the AI industry. The trillion-dollar valuations, the promised capabilities, the timelines for artificial general intelligence, the industrial deployment strategies: all exhibit the structural features Flyvbjerg documented in infrastructure planning, operating at compressed timescales and higher stakes. The book's heuristics transfer directly: think slow about AI adoption, take the outside view of previous AI cycles, get the team right for judgment-heavy phronetic work.

The book's reception was broad and commercially successful, reaching audiences far beyond the planning and project management communities that had absorbed the underlying research. It was selected as a New York Times notable book and translated into dozens of languages. The success demonstrated that the empirical pattern — that big things rarely get done as planned — has intuitive resonance for readers who have experienced it in their own organizations, even those who have never worked on an infrastructure megaproject.

Origin

Flyvbjerg and Gardner worked together for several years to translate three decades of technical research into accessible prose. The book was published by Currency (Penguin Random House) in 2023 and became an international bestseller.

Key Ideas

99.5 percent failure rate. The empirical record: of sixteen thousand projects, essentially none came in on budget, on time, and on benefits; the pattern is structural rather than exceptional.

Think slow, act fast. The operational heuristic: extensive planning and comparison before commitment, rapid execution after commitment.

Outside view discipline. The corrective for optimism bias is systematic comparison with the reference class of structurally similar prior cases.

Team quality over plan quality. The best predictor of successful execution is the quality of the people executing, not the quality of the planning document.

Structural pattern, structural correction. Individual projects fail through individual missteps, but the pattern of failure requires structural understanding and structural intervention.