The iron law of projects: the Sydney Opera House

In 1957 the Danish architect Jørn Utzon won an international competition to design an opera house for Sydney with a set of drawings that were closer to sketches than plans. The roof was a series of sail-like shells, and the engineers did not yet know how they could be built. The premier of New South Wales, Joe Cahill, wanted the project started before the state election of March 1959, so that a change of government could not cancel it, and he wanted it cheap enough to approve. The estimate was seven million Australian dollars, with completion promised for 1963. Construction began in 1959 with the design unresolved.

The base was built first, to a roof design that did not yet exist. When Utzon and the engineers finally found a way to build the shells, in 1961, the podium columns that had already been built could not carry them, and they had to be blasted out and rebuilt. Costs climbed through the 1960s. In 1965 a new state government took office, withheld Utzon's payments and demanded changes. In 1966 he resigned, left Australia with his family, and never returned to see the building. Local architects finished the interiors, and the Opera House opened in October 1973, fourteen years after construction began, at a cost of 102 million dollars, roughly 1,400 per cent over the original estimate.

Bent Flyvbjerg, who has spent his career studying large projects, uses the Opera House as the clearest case of what he calls the iron law: over budget, over time, under benefits, over and over again. His database of more than 16,000 projects across more than twenty types finds that 47.9 per cent finish on budget, 8.5 per cent finish on budget and on time, and 0.5 per cent finish on budget, on time and with the benefits that were promised. The same book's counter-example is the Guggenheim Museum in Bilbao, where Frank Gehry worked through the design in detail, with computer models and many iterations, before construction started, and the building opened in 1997 on time and on budget.

The Opera House failed at the moment the date was fixed, years before anyone knew what it would take to meet it. Cahill's deadline was political, the budget was set to be approvable rather than accurate, and construction started so that stopping would be harder than continuing. Flyvbjerg finds the same pattern across his database: dates and budgets set to win approval, with the real planning pushed into the build, where every mistake is most expensive. His phrase for the alternative is think slow, act fast: spend the time on planning, where mistakes are cheap, so that building, where mistakes are expensive, can be quick.

The same shape appears far below the scale of opera houses. A launch date promised to a board before the product is specified. A hiring plan with a start date set by the budget cycle rather than by the search. A campaign date fixed by an event with the creative still undecided. In each case the date is set first and the plan is fitted to it afterwards, and the overrun arrives as a surprise to everyone except the people doing the work. The practical test is one question, asked of the people who will do the work: what has to go right for us to make this date? If they cannot answer it, the date was picked to win approval. Build the plan first, and let the plan set the date.

There is an established way to pay for that planning. American federal highway rules recommend stipends for shortlisted bidders on large design-and-build projects where preparing a proposal is costly, usually covering a third to a half of each unsuccessful bidder's cost of preparing it, and states including Texas, Minnesota and Missouri pay them. In return the client can usually use the ideas in the losing proposals. A client who wants a real date and a real budget can pay for the work that produces them.

Source: Bent Flyvbjerg and Dan Gardner, How Big Things Get Done, 2023, chapter 1 and the database figures throughout. The politics of the approval: Bent Flyvbjerg, Design by Deception: The Politics of Megaproject Approval, Harvard Design Magazine, 2005. The project database: Flyvbjerg, What You Should Know About Megaprojects and Why, Project Management Journal, 2014. The demolition of the podium columns: Flyvbjerg in conversation with Sandra Peter, Sydney Business Insights, University of Sydney. The stipends: US Federal Highway Administration, 23 CFR 636.113, Is the stipend amount eligible for Federal participation?

How Big Things Get Done

Flyvbjerg and Gardner, 2023

The book on why big projects fail and what the rare successes do differently, from the researcher with the largest database on the question. The Opera House and the Guggenheim open it, and the method, think slow and act fast, works at any scale.