Anthony M. Townsend - Smart Cities - Big Data, Civic Hackers, and the Quest for a New Utopia

Decades of research lie ahead before we can hope to create software simulacra of cities that approach the psychohistorians’ standards of society-scale prediction. And on top of the challenges that have dogged past efforts, new challenges for urban models are on the horizon. For starters, the very same apparatuses that will feed big data into future models—mobile phones, instrumented infrastructure, and digital transaction records—are changing the way cities actually function. As Batty explained to me: “That’s the other side of the coin. New communications systems at the local level are actually changing how we communicate. It’s not just a question of measuring things that we always did. It’s a question of new things emerging. There is a lot of new interaction going on ... building dynamics into the city that we’ve never had a hold on at all.” Even if IBM’s model is perfect today, tomorrow it could be out of date, as new technologies allow us to rewire behavior at the individual level. Even if we can measure the movement of every person in real time, all we’d have is topsight, the big picture. Without an understanding of why individuals are, say, changing the time of day they commute (based on real-time traffic reports beamed to their phone perhaps), we can’t accurately simulate their behavior. The models break. It’s even possible that those new behaviors are evolving so fast that even our revised assumptions will be out of date by the time they’re programmed into the simulator. Theory will lag reality, and the way cities work might actually get weirder and more complicated far faster than we can decode and model it.

Then there’s the risk that by measuring something, we change it—a kind of observer effect for social science. Typically understood, the observer effect describes

how instruments used to quantify dimensions of the physical world can actually alter the conditions they seek to size up. In experimental physics, it means that to measure the velocity of one subatomic particle, you've got to bounce another one off it like a billiard ball, thus changing the thing you're trying to measure. In electronics, a voltmeter actually becomes part of the circuit itself. This principle is so fundamental to scientific measurement that Asimov even incorporated it as a central axiom of psychohistory. “The human conglomerate,” he wrote in Foundation , must “be itself unaware of psychohistoric analysis in order that its reactions be truly random.” Will people in a sensed and modeled city behave differently, either by choice or because some plan or policy based on the model directs them to? Either way, it could break the model’s assumption and reduce its results to nonsense.

If we assume for a moment that all of these obstacles can be overcome, we are still left to ponder whether better computer models will lead to better cities. The technocratic, top-down style of city planning that gave rise to earlier models is today as archaic as their computer code. Citizens now expect to see, participate in, and even initiate plans. But complex computer models will bring back technocratic opacity, “black boxes” where, as Douglass Lee put it, “What goes in and what comes out are known exactly, but the process by which one is transformed into the other is a mystery.”90

A far bigger risk is that public officials will accept the advice of these black boxes unquestioningly. As Colin Harrison recounted, early in the Portland models development, the mayor “formed an idea in his mind of what this model was going to be able to do ... the planners thought that he was viewing this model as a kind of oracle. He could ask any planning question of the oracle, and it would tell him what the right thing to do was. The planners got very, very nervous about this, and we had to work through this to make sure that he understood that models aren’t oracles.”91 It was a surprisingly responsible response by IBM.

Gelernter saw this as perhaps the greatest risk of mirror worlds—that we would mistake them not as reflections or representations, but as reality. Toward the end of the Mirror Worlds' epilogue, his alter ego Ed rants: “I can in fact believe that a Mirror World would suck life from the thing its modeling into itself, like a roaring fire sucking up oxygen.

The external reality becomes just a little bit... not superfluous; second-hand.... Couldn’t it happen that, instead of the Mirror World tracking the real world, a subtle shift takes place and the real world starts tracking the Mirror World instead?”

Computer simulations seduce precisely because they replace the complexity of the real world. The video game SimCity is addictive because of the simplicity of its underlying model—players quickly figure out how to win by exploiting its predictable dynamics (in fact, the design of early versions was directly borrowed from Urban Dynamics. Following trends in research, SimCity 2013s GlassBox simulation engine now uses a sophisticated agent-based model).93 But even the best mathematical models of real-world phenomena are always approximations. Newton’s laws made sense for centuries until physicists began looking at the very small scale of matter inside the atom. There a weird new physics reigned and a new model, quantum mechanics, had to be developed to give a better (but still not perfect) approximation of reality.

When I first learned of IBM’s work to bring back urban dynamics in Portland, I set out to unmask a villain. What I found was a company perhaps ignorant of a long-buried past, yet willing to listen to experts and learn from its missteps. IBM now knows the political limits of system models of cities. But I wonder if the company has absorbed the more fundamental lesson on their practical limits. Cybernetics redux in Portland was premised on the notion that bigger data, bigger computers, and bigger models were the remedy to Forrester’s shortcomings. It’s a familiar, but hollow refrain. As Lee wrote in 1973, “Despite the many-fold increases in computer speed and storage capacity ...” in the 1960s, “there are some researchers who are convinced that it has been the hardware limitations that have obstructed progress and that advances in modeling are now possible because of larger computer capacity. There is no basis for this belief; bigger computers simply permit bigger mistakes.”94

A Tale of Two Models

IBM’s Banavar is sanguine about the centralization of power in Rio’s Intelligent Operations Center. “For better or worse,” he reflects, “we have given a lot of power to our municipal governments.” There is clearly a case to be made that the urgency of urban problems, especially those faced by mayors in the developing world, justifies arming them with powerful new software and richly detailed information. “I strongly believe we should give them the right tools and the right data to be better managers,” Banavar says.93

But if we share Gelernter’s concerns, we should worry that the mirror world Rio’s mayor Eduardo Paes has created in cahoots with IBM will tip the balance of power decidedly in his favor. For now, Paes claims to act in the people’s interest. “Every day since I joined the city government,” he expounded in the promotional film produced for the Rio Operations Center, “I have dreamed of having this space for the people ... for people to know that they are being cared for.” Paes doesn’t hide his paternalistic philosophy of governance; neither is it completely out of place in Brazil. But as IBM exports this new technology and management playbook to the rest of the world, can the ideology from which it was spawned be left behind? And what happens when progressives leave power, and the new tool is turned by autocrats against the people instead?

“Brazil is not for beginners,” the country’s most famous songwriter, Tom Jobim, the bossa nova genius who gave the world “The Girl From Ipanema,” once said. One wonders at the wisdom of picking such a complicated place to launch a high-profile showcase for IBM’s smart city ambitions. The fashioning of Brazil’s cities has been a story of chaos, dissent, and grassroots improvisation—a century-long struggle to deal with the cruel social and economic legacy of slavery.