It would be hard to argue that designers intended the linear parking lots that many of our city freeways have become. They simply did not understand the nature of induced demand: build it and they will come -- at which point you'll learn that you didn't build it big enough. This is cyclic, congestion-abetting design, however unintended.
But the unintended is not always a bad thing. Take the construction of I-93 through Boston's central business district. There's no doubt it rendered the city in ways it's still suffering from, but as the elevated roadway curved northwest at Atlantic Avenue, it effectively created a wall between the North End, Boston's oldest residential community, and the rest of the city. In important physical and psychological ways, the North End was preserved by this severing of the formerly continuous urban fabric. (While many in "urban renewal" circles saw the North End as blight to be removed, the publication of Jane Jacobs' "Death and Life of Great American Cities" is credited with preventing it from being razed as Boston's West End had been shortly before.)
The North End's Little Italy neighborhood is arguably the most authentic in the United States. The scale of interaction is utterly human, its streets and buildings echoes of usage rather than overlaid standardization. One need only look to Boston's "Disneyfied" Faneuil Hall marketplace to see what commercial development of a historic area could have meant for the North End; the urban freeway as an unintended factor in historic preservation.
But the Big Dig buried I-93 and the barrier to pedestrian movement to the North End is mostly gone. And so is the barrier, perceived or real, to the forces of gentrification and large-scale development that characterize so many historic districts today. The consequences of this remain to be seen.
One might point out that high-performance computing today gives us the ability to model complex system dynamics, simulating possible outcomes of given decisions. But these are early days -- and even if a truly comprehensive, responsive city simulation could chart the probability curves for every possible outcome, we'll always have to confront decisions that have implications for a future further out than we can currently predict. (Early supporters of the automobile in cities rested much of their argument on a very obvious environmental benefit: cars didn't leave piles of poop in the street.)
So while technology does not seem to offer short-term hope for predicting the unintended in complex systems, the invisible architecture of data and services does afford more nimble rearrangement and redeployment than do viaducts of concrete or steel girders, at least in theory.
Cities, of course, are a model of resilience. Buildings are adapted, the street finds a way, patterns of usage can be reshaped -- albeit with the pace of a medium made of brick and mortar. In this way, but with far more speed and flexibility, the architecture of information and services in cities can be built not to prevent unintended outcomes but to enable them. If we agree that inadvertent outcomes are unavoidable in sufficiently complex systems such as cities, then the opportunity we're presented with in matters of digital design -- the lesson, so to say -- is that extensibility should be a primary goal. Our new urban information architecture must, above all, encourage adaptation.