The City That Already Knows You'll Take the Stairs

In 2012, researchers placed a sign near an escalator in a Swedish subway station. The sign didn't tell people to exercise. It just pointed out that the stairs were faster. Stair use went up 70 percent. Nobody's behavior changed because they read a public health report. It changed because someone redesigned the decision.

That small experiment is about to play out at urban scale, and most people haven't noticed it happening.

Cities are in the middle of two simultaneous transformations that are usually discussed in separate rooms: the push toward smarter, more responsive urban infrastructure, and a maturing understanding of how human beings actually make decisions. Behavioral economics and urban design have been parallel tracks for decades. They're starting to converge, and the results are going to be stranger and more consequential than either field's proponents typically admit.

The gap between what cities build and what people do

Urban planners have spent the last decade fixating on "smart cities," mostly meaning sensors, data, AI optimization of traffic flows and energy grids. The technology is real and the efficiency gains are measurable. But the premise underneath most smart city thinking is still roughly the same premise that traditional economics runs on: people are rational actors who respond to information and incentives in predictable ways. Give them a better bus schedule, and they'll take the bus.

Except they won't, not reliably, not enough. Decades of behavioral research have established that humans respond to how choices are framed, to defaults, to what their neighbors appear to be doing, to friction and proximity, far more than to information alone. Loss aversion is a stronger motivator than equivalent gain. The default option is chosen at dramatically higher rates than alternatives, regardless of quality. The easier path is taken, consistently, even when people know a better path exists.

None of this is a criticism of people. It's just how cognition works under the conditions of ordinary life, when you're tired and distracted and optimizing across too many variables at once.

What choice architecture looks like at city scale

So what happens when you start designing urban environments with this in mind, not just for individual decisions in a cafeteria or a hospital, but for the accumulated daily choices of millions of people navigating a city?

Some of this is already being done, mostly quietly. Cities experimenting with "sponge city" infrastructure, integrating wetlands and permeable surfaces to manage stormwater, are also discovering that residents use those green corridors more than conventional parks. The design nudges behavior without announcing itself. Bike-sharing programs placed within 300 meters of transit stops see two to three times the usage of those placed further away. The distance isn't prohibitive; the slight friction is enough to shift the decision.

AI is now starting to feed into this at a systems level. Fleet telematics companies are analyzing billions of data points to optimize city transportation flows. Energy grid management increasingly relies on real-time behavioral prediction, not just physical modeling. The question cities haven't fully grappled with yet is whether that predictive capacity should stay in the optimization layer, or whether it should start shaping the choice environment itself.

The ethics question that keeps getting deferred

This is where the discomfort lives, and it deserves more than a footnote.

Behavioral economics has always had a critics' corner that raises a legitimate objection: who decides what the "right" behavior is, and who gave policymakers the authority to engineer people toward it? The nudge literature often sidesteps this by emphasizing that nudges preserve choice, nobody is forced to take the stairs. But the reframing doesn't resolve the concern. A default is a choice, including the choice to set a particular default for a particular population.

At the scale of city infrastructure, designed by a mixture of municipal governments, private developers, and AI systems trained on optimization targets, the question of whose preferences are being encoded gets genuinely murky. Smart grids that shift household energy use during peak demand are efficient. They're also making a decision on behalf of residents who may not know a decision is being made.

This isn't an argument against any of it. Cities that adapt to climate pressures, manage resources intelligently, and make sustainable choices the path of least resistance are better cities, on metrics that matter. But the behavioral economics literature that's now informing these designs also contains its own warning: the framing of a choice shapes the outcome, and whoever controls the framing holds real power.

The subway stairs experiment worked because the goal was benign and transparent and the person walking by could see exactly what was happening. The city-scale version of that experiment is less legible, and the goals are set by more complicated processes.

That's worth keeping in view as the architecture gets smarter.