The Songdo international business district, 65km west of Seoul in South Korea.
This week London hosts a jamboree of computer geeks, politicians and urban planners from around the world.
At the Urban Age conference, they will discuss the latest whiz idea in high-tech, the ''smart city''. Doing more than programming traffic, the smart city's computers will calculate where offices and shops can be laid out most efficiently, where people should sleep and how all the parts of urban life should be fitted together.
Science fiction? Smart cities are being built in the Middle East and in South Korea; they have become a model for developers in China and for redevelopment in Europe. Thanks to the digital revolution, at last life in cities can be brought under control. But is this a good thing?
You don't have to be a romantic to doubt it. In the 1930s the American urbanist Lewis Mumford foresaw the disaster entailed by ''scientific planning'' of transport, embodied in the super-efficient highway, choking the city. The Swiss architecture critic Sigfried Giedion worried that after World War II efficient building technologies would produce a soulless landscape of glass, steel, and concrete boxes. Yesterday's smart city, today's nightmare.
The debate about good engineering has changed now because digital technology has shifted the technological focus to information processing; this can occur in hand-held computers linked to ''clouds'' or in command centres. The danger now is that this information-rich city may do nothing to help people think for themselves or communicate well with one another.
Imagine you are a master planner facing a blank computer screen and that you can design a city from scratch, free to incorporate every bit of high technology into your design. You might come up with Masdar in the United Arab Emirates or Songdo in South Korea.
These are two versions of the stupefying smart city: Masdar the more famous, or infamous; Songdo the more fascinating in a perverse way. Masdar is a half-built city rising out of the desert, whose planning - overseen by the master architect Norman Foster - comprehensively lays out the activities of the city, the technology monitoring and regulating the function from a central command centre. The city is conceived in ''Fordist'' terms - that is, each activity has an appropriate place and time.
Urbanites become consumers of choices laid out for them by prior calculations of where to shop, or to get a doctor, most efficiently. There's no stimulation through trial and error; people learn their city passively. ''User-friendly'' in Masdar means choosing menu options rather than creating the menu. Creating your own new menu entails, as it were, being in the wrong place at the wrong time.
In mid-20th-century Boston, for instance, its new ''brain industries'' developed in places where the planners never imagined they could grow. Masdar - like London's new ''ideas quarter'' around Old Street, east London - on the contrary assumes a clairvoyant sense of what should grow where. The smart city is over-zoned, defying the fact that real development in cities is often haphazard, or in between the cracks of what's allowed. Songdo represents the stupefying smart city in its architectural aspect - massive, clean, efficient housing blocks rising up in the shadow of South Korea's western mountains, like an inflated 1960s British housing estate - but now heat, security, parking and deliveries are all controlled by a central Songdo ''brain''. The massive units of housing are not conceived as structures with any individuality in themselves, nor is the ensemble of these faceless buildings meant to create a sense of place.
Uniform architecture need not inevitably produce a dead environment, if there is some flexibility on the ground. In New York, for instance, along parts of Third Avenue, monotonous residential towers are subdivided on street level into small, irregular shops and cafes; they give a good sense of neighbourhood. But in Songdo, lacking that principle of diversity within the block, there is nothing to be learnt from walking the streets.
A more intelligent attempt to create a smart city comes from work currently under way in Rio de Janeiro. Rio has a long history of devastating flash floods, made worse socially by widespread poverty and violent crime. In the past people survived thanks to the complex tissues of local life; the new information technologies are now helping them, in a very different way to Masdar and Songdo. Led by IBM, with help from Cisco and other subcontractors, the technologies have been applied to forecasting physical disasters, to co-ordinating responses to traffic crises and to organising police work on crime. The principle here is co-ordination rather than, as in Masdar and Songdo, prescription.
But isn't this comparison unfair? Wouldn't people in the favelas prefer, if they had a choice, the pre-organised, already planned place in which to live? After all, everything works in Songdo. A great deal of research during the past decade, in cities as different as Mumbai and Chicago, suggests that once basic services are in place people don't value efficiency above all; they want quality of life. A hand-held GPS device won't, for instance, provide a sense of community.
What's more, the prospect of an orderly city has not been a lure for voluntary migration, neither to European cities in the past nor today to the sprawling cities of South America and Asia. If they have a choice, people want a more open, indeterminate city in which to make their way; this is how they can come to take ownership over their lives.
There's nothing wicked about the smart city confab London is hosting this week. Technology is a great tool, when it's used responsively, as in Rio. But a city is not a machine. As in Masdar and Songdo, this version of the city can deaden and stupefy the people who live in its all-efficient embrace. We want cities that work well enough but are open to the shifts, uncertainties and mess that are real life.
Richard Sennett is a professor of sociology at the London School of Economics.