The Atlantic’s City Lab blog posted an interview with Josef Konvitz and his comments on modeling and forecasting stuck out to me —
Great projects have dynamic effects, precipitating changes throughout urban regions that are all but impossible to model in advance because no one can anticipate their impact. Projects are often sold on the basis of the number of construction jobs or new housing starts that will follow. This narrow approach to cost-benefit analysis would have led the Victorians to conclude that a major sewer system for London was too expensive. By the same token, bridges and tunnels linking New York and New Jersey would never have been built a century ago. The test of good infrastructure is whether it makes best use of the density, size, and complexity of cities.