Book: How Big Things Get Done: The Surprising Factors Behind Every Successful Project, from Home Renovations to Space Exploration Author: Bent Flyvbjerg & Dan Gardner In one line: Big projects fail by default - the cure is to think slow, act fast: plan long, hard, and cheaply on paper, then deliver in one quick, uninterrupted window.
1 · Think slow, act fast
Spend generously on slow, tested, experimental planning; then deliver fast. A short build window starves black swans of the time they need to wreck the project. Most projects do the opposite - rush to commit, then drag on for years - and pay for it.
2 · Take the outside view
Do not trust your own inside story. Anchor every estimate on how a whole class of similar projects actually turned out. Reference-class forecasting beats optimism and wishful thinking because reality already ran the experiment thousands of times.
3 · Build with Lego
Make the big thing out of small, repeatable, tested pieces. Modularity lets you learn on a curve, scale fast, and fail small. One giant bespoke object cannot learn - and when it fails, it fails catastrophically.
Almost every big project obeys the Iron Law of Megaprojects: over budget, over time, under benefits - over and over. Flyvbjerg draws on a database of more than 16,000 projects across dozens of countries and types, and the pattern is brutally consistent: only a tiny fraction land on budget, on time, and on the benefits promised. This is not bad luck. It is the predictable output of how we habitually start big things - excited, over-committed, and in a hurry to break ground.
Two forces drive the failure. First, psychology: optimism bias and the planning fallacy make us picture the best case and ignore the base rate. Second, power and politics: strategic misrepresentation - the polite name for lying to get a project approved - means costs are deliberately low-balled and benefits inflated so the thing gets a green light. Once the first shovel is in the ground, sunk costs and momentum do the rest.
The remedy is to invert the usual order. Make planning slow, thorough, and safe to fail in, so that delivery can be fast and boring. A long, careful, well-modelled plan is precisely what buys you a short, low-risk build. The projects that beat the Iron Law - Madrid’s metro, the Empire State Building, Pixar’s films - all, in their own way, thought slow and acted fast.
Think slow, act fast. Planning is where you experiment, model, stress-test, and iterate - cheaply. Delivery is where the money burns and the risks bite, so keep that window short and uninterrupted. The Empire State Building went up in about a year because its plan was ruthlessly resolved before ground broke; the Sydney Opera House started building before the design was solved and ran roughly ten times over budget, years late.
Forecast from the outside, not the inside. Your project feels unique, so you build a bottom-up story of how this one will go - and it is almost always too rosy. Instead, put your project in a reference class of similar past projects and anchor your cost and schedule on their real, full distribution of outcomes. Ask what your project has in common with the ones that came before, not what makes it special.
Build modular - find your Lego. Ask what small, repeatable unit your big thing is made of, then mass-produce and assemble it. Solar, wind, server farms, and Madrid’s metro all scale by repeating a proven block, riding a learning curve, and failing small. Bespoke one-offs - nuclear plants, Olympic Games, custom IT - cannot learn between units and sit in the fat tail where blow-ups are unbounded.
The Iron Law
Over budget, over time, under benefits - the default outcome across 16,000+ projects. Fewer than one in a hundred hit all three targets. Knowing the law is normal, not exceptional, is the start of beating it.
Optimism and lying
Estimates die two deaths: optimism bias / the planning fallacy (we believe our own best case) and strategic misrepresentation (we low-ball on purpose to win approval). The outside view is the antidote to both because it ignores the story and looks at the record.
Reference-class forecasting
Anchor your estimate in the actual distribution of comparable past projects, then adjust. It is the single most reliable debiasing tool in the book - so effective that the UK Treasury adopted it for public projects.
Planning is a safe place to fail
On paper and in models, a mistake costs almost nothing. Pixar iterates its films through countless cheap versions before committing; Frank Gehry simulates and tests exhaustively so the build itself holds few surprises. Fail on paper, not in concrete.
Modularity and the fat tail
Repeatable units (solar panels, metro segments) cluster near the average and improve with each copy. One-off giants (nuclear, Olympics, bespoke IT) live in a fat-tailed distribution where the worst case is not just bad but ruinous. Choose the shape of your risk.
Your biggest risk is you
The deepest hazard is not the market or the weather - it is your own behavioural bias and premature commitment. Design a process that checks the planner, not just the plan.
Experience and phronesis
Deep, hands-on mastery - Aristotle’s phronesis, practical wisdom - lets an experienced team smell trouble early. Hire people who have actually done this kind of thing, and treat their gut as data.
Start with why
Lock the purpose before the form. A clear question - why are we really doing this? - keeps the project honest and prevents the wrong thing from being built beautifully.
Flyvbjerg opens with the Iron Law and the sheer scale of the failure pattern, then diagnoses the causes: the psychology of optimism and the politics of misrepresentation. From there the argument turns constructive - think slow, act fast; the case for slow, experimental planning (Gehry, Pixar); the outside view and reference-class forecasting; and the power of modularity, illustrated through solar, wind, data centres, and Madrid’s metro versus the fat-tailed disasters of nuclear power and the Olympics. It closes on people and mastery - that experience, humility, and the right team are what carry a good plan through a fast build.
Fix the what and why first. Write the real purpose in one honest line before touching design, budget, or schedule. Guard against premature lock-in on a form.
Plan slow - fail on paper. Model, prototype, simulate, and iterate cheaply until the plan is boringly reliable. Treat planning as the place to make and survive your mistakes.
Forecast from the outside. Assemble a reference class of similar finished projects and anchor your cost and time on how they actually ended - full distribution, not the sunny middle.
Find your Lego. Ask what small, repeatable, testable unit the work can be built from, so you can learn on a curve and cap the size of any failure.
Check the planner, not just the plan. Name your likely biases and any incentives to low-ball; invite an outside sceptic to pressure-test the numbers.
Then act fast. Once the plan is proven, deliver in one tight, uninterrupted window to starve disruption and black swans of time.
Hire experience. Put people with real scars from similar projects in the room, and weight their judgement.
Context The book generalises megaproject research to any ambitious undertaking - a kitchen renovation obeys the same laws as a rail line. Critics note the data skews toward large public infrastructure, so base rates for small private projects should be read with care, and that “strategic misrepresentation” can be hard to separate from honest error after the fact. The core tools - the outside view and modularity - are widely validated well beyond this book, which is part of why the advice holds up.
The Iron Law of Megaprojects: Over budget, over time, under benefits, over and over.
Think slow, act fast.
Your biggest risk is not the market or the technology - it is you, and the biases you bring to the plan.
Ask not what is unique about your project, but what it has in common with the ones that came before.
Planning is cheap and safe; building is expensive and dangerous - so make your mistakes on paper.
Find your Lego: the small, repeatable thing your big thing is made of.