What did "smart" promise, and who promised it?

In the first essay, I described how the same loop came to mind in the solutions I saw at the fair: notice, decide, respond; at different time scales.

In the second essay I asked whether that loop could run a city, and I concluded that it can help the parts of a city but cannot run the whole city, which is a problem of organized complexity.

The question I will now try to answer is this: who promised that this loop could run a city, when, and how — and did they manage it?

Three words

In November 2008, IBM CEO Sam Palmisano gave a speech in New York. He said the world was becoming three things: instrumented, interconnected, and intelligent.¹ He added that this was not a metaphor, that intelligence would be built into the way the world literally works.¹

Reading those three words, I stopped. My loop was “notice, decide, respond.” Palmisano’s words covered the first two: the instrumented world was “notice,” the intelligent world was “decide”; and “interconnected,” in between, was the network I had seen at the fair. But the third step, “respond,” was not in the promise. In the second essay, I saw that the limit lies exactly there.

Within a few years the promise had turned into a marketing language. Writing in 2013 about the “canonical smart city,” Adam Greenfield notes that he drew every detail directly from the promotional copy of IBM, Cisco, Siemens, and companies that build cities from scratch.²

The control room

Rio de Janeiro showcased the promise. The operations center set up with IBM in December 2010 gathered data from many of the city’s agencies onto a single wall of screens.³ About a year later, at a conference, the mayor Eduardo Paes said that he could govern his city with technology from there, from California — that he knew everything.³

I remember the control center in the second essay: the screen had seen a bottleneck but had not seen the care home on the corner. “Knowing everything” was exactly that screen’s promise.

Cities from scratch

The purest form of the promise was to build the city from data, from the very beginning.

Songdo, in South Korea, was promoted as the cure for Seoul’s ills; in 2018 a newspaper described it as “overdue, overpriced and underpopulated,” and a resident called it a “ghost town.”⁴ Masdar, in Abu Dhabi, began in 2006 as a zero-carbon city for 50,000 people; ten years later, less than five percent of the plan had been built.⁵ The number of people living and working there today is given as around 15,000, and there is no longer a completion date; Masdar now presents itself less as a city than as a green business and research zone that hosts the headquarters of the International Renewable Energy Agency (IRENA) and aims at net zero by 2050.⁶ Google’s sister company Sidewalk Labs proposed building a neighborhood “from the internet up” in Toronto; in May 2020 it withdrew from the project, citing economic uncertainty. Opponents put it differently: “We have a consent problem.”⁷

Each of these projects stumbled for a different reason: some over money, some over politics, some over the privacy debate. I cannot say they all came from the same mistake; but together they make it worth questioning the assumption that a city designed with technology will also work socially as expected.

The flaw in the promise

According to Greenfield, the smart city claims — even in principle — an objectivity, a wholeness, and a perfect knowledge that are nowhere reachable.⁸ Put in the language of the second essay, there was a contradiction inside the promise: it promised to solve a problem of organized complexity with the tools of simple problems.

But to be honest, the sellers were hiding nothing. What was sold was out in the open: measure, connect, optimize. The real question is why cities bought this sales promise as a form of government.

2019: did the language change?

In 2019, at the fair, one person was summarizing what was said in the sessions by drawing it and hanging the drawings in a common area just outside the rooms. I took photographs of them; looking again today, I see three different promises standing side by side at the same fair.⁹

On the session poster of the Siemens executive Roland Busch, the old promise stands upright: “Measuring,” “Go to all the devices & connect them,” “Act · Decide · Prevent · Predict.” The same poster also includes an honest admission: “If planning is wrong… technology does not deploy or work.” The fair’s own announcement says, “Make the invisible visible.” And on the panel posters there is a new language: “Citizens are not data carriers,” “Fundamental trust is the real currency,” “With humility.”⁹

I do not know for certain where this new language came from; but by 2019 the context had changed. Europe’s data-protection law has been in force since 2018.¹⁰ The Toronto project was under great dispute.⁷ Barcelona, the fair’s host, was trying another road too: the city’s 2015–2019 digital plan aimed to go “beyond the smart city,” saying that technology ought to be at the service of people and not people at the service of technology.¹¹ But Barcelona had not given up measuring; both the sensor network and the management dashboard were in the plan. The plan’s aim was not to stop measuring, but to win back for the city control over the data that digital technologies produce.

Even so, the old promise had seeped into the new language. On the poster of a human-centered panel it read, “Algorithms (no biases).”⁹ Yet that making some things visible can make other things invisible was hardly voiced on the posters. I think the promise changed its language, but not entirely its essence.

Smart enough

Ben Green calls the belief beneath this promise “tech goggles”: the belief that technology offers neutral and optimal solutions to social problems and is the primary engine of social change. His proposal is the “smart enough city”: a city that uses technology as a powerful tool but does not treat it as an end in itself.¹²

This is my limit too: the system supports the decision; handing the decision over is possible only in narrow matters, such as signal timings, and only with great care. But “support” is not safe on its own either. Green’s experiments show that this arrangement, in which people decide on the algorithm’s recommendation, has serious limits.¹³ It is not enough for a human to be in the loop; that human must have the time and the reasons to think. I trust the car’s brakes, but I keep my foot near the pedal.

Who promised what?

The technology companies made the promise most loudly: a city that is instrumented, interconnected, and intelligent.

Some mayors bought it as a wall of screens. Investors tried to build it from scratch. The promise was that the loop I saw in Barcelona could run the city.

Did they manage it? Cities built from scratch couldn’t deliver on the promise. What actually worked is a separate question; I will come back to it at the end of the series.

Three questions remain open for me, one for each step of the loop; I ask them of myself. I know what I notice; but do I know what I fail to notice? How much are my decisions shaped by my ability to carry them out? And, most important of all: when there is more than one option, what values will I use to rank them?

The newest form of the promise today is the “digital twin”: a living copy of the city, a claim to perfect the “notice” step. That is the next essay’s question: What is a digital twin — a map, a model, or a mirror?