Interview with Adam Greenfield on designing for urban computing

Adam Greenfield is one of the most interesting thinkers on many of themes that we regularly address at The Mobile City. He is head of design direction for service and user-interface design at Nokia, and is currently working on a book called “The City Is Here For You To Use: Urban form and experience in the age of ubiquitous computing”. When he was in Amsterdam recently to talk at the Picnic Conference, I looked him up for a short talk on the design of mobile services.

MdW: In the Situated Technologies Pamphlets 1: Urban Computing and its Discontents you wrote that the design of mobile services or ubiquitous computing often comes down to very banal things: it’s about everyday life acts as being able to find a good breakfast place when you need it. Yet at the same time – as Paul Dourish has argued – every act of design is also a philosophical act. The starting point is always an –often implicit- idea of how users are acting in the world.

AG: Yes, that’s very true. Every act of design proceeds from an implicit “object model,” which is to say that there’s always some idea bound up in it of what this person “the user” is and how that person behaves. I think it’s vital to make these models explicit, to drag them out into the light, and to validate them against everything we know about how people are actually using the city.

People, in other words, are not abstractions. People have particular attributes. My own might include things like my height, my weight, my eye color, etcetera, but also my physical location. And physical locations themselves have attributes we might specify: temperature, humidity, carrying capacity, current population, and so on. What I would ideally like to see is a coordinated effort across people working in planning, architecture, user experience and other interested disciplines toward setting up a shared and open metadata framework in which we might usefully describe people, places and things as computational objects.

In some ways, I’ll grant that this sounds an awful lot like those old dreams of the semantic web, dreams in which I don’t place a good deal of credence. But I do think we need some kind of shared descriptive framework if our intention is to build services of enduring utility. Any such effort, again, necessarily involves a philosophical commitment, and a political one. But that doesn’t mean it shouldn’t be done.

Of course it will be hard to set up any such shared framework. So far we haven’t been good at managing the fault lines between the implicit, roughly shared understandings that we muddle through with in everyday life and the razor-sharp explicit delineations made by technical systems. For instance, consider the matter of neighborhoods. The idea of neighborhood is very subjective – different people will have very different notions of the exact contours of a neighborhood, who legitimately gets to claim affiliation with that neighborhood, or even what it should be called. A real-estate agent is likely to have a different conception from a fireman or a recent immigrant. So I was surprised when I uploaded some pictures to my map on Flickr, and saw that it recommended labeling them as having been taken in “Tenderloin, New York.” Now I’ve lived in New York most of my life, I’ve ready Luc Sante’s great history of the place, but I don’t recall ever having heard that particular name for the district before. And yet this was somehow the name that Flickr had arrived at for pictures taken at such-and-such a lat/long.

This turns out to be an important challenge when is a city is being represented digitally. There might well be some slippage between the subjective, everyday notion of a city, the juridical domain that constitutes the formal municipality, and the geolocative framework we use to access data about that place. You can be psychologically in Amsterdam, but legally somewhere else and technically in one of the two, or in none at all.

MdW: If we extend Dourish’s philosophical notion of design, could we then say that designing ubicomp or locative media services should also imply a philosophical notion of what a city is and how it should function ideally? When you take the idea of a city as a community of citizens as a starting point, you’ll end up with a different set of locative applications then when one starts off with the idea of a city as an aggregate of services offered at consumers. Now, in the Situated Technologies Pamphlet you gave some examples of ubiquitous computing in architecture. Most of them are in the category of adaptive architecture: a building responds to input from a variety of sensors. Do you see any examples or discussions that go beyond this approach? That start from a more philosophical idea of what a city is or should be?

AG: Well, my own personal interest in architecture is first phenomenological, and maybe immediately after that social. In other words, how does it feel to inhabit this particular region of built space, and what kinds of social possibilities might it give rise to? But in the real world, this is almost never the way things work. For institutions, at least, space is an envelope to be managed, not anything remotely like a platform for conviviality. It’s an envelope of material, physical and economic constraints that must be optimized against in any planned scenario of use. So at present, most of the digital interventions we see in architecture take the form of BIM, building information management systems.

And maybe, you know, that’s for the better. Certainly the social-networking systems that we have been seeing so far are very clumsy. So while personally, I am reasonably comfortable with interventions aimed at, say, environmental management, I’m a whole lot less comfortable with the idea of the building management trying to “accelerate serendipity.” I’d rather have them manage humidity dynamically, than even try to manage the social stuff inside the building. That should be left to the people using the space.

From where I stand, technologists should aim at creating a service framework that is subtle and inobtrusive, yet robust and open enough so that people can reach in, grab it and use it. I believe that this is where artful and effective interventions can be made. And we certainly shouldn’t leave this to the good intentions of governments or building managers.

MdW: This reminds me of something you brought up in a talk here in Amsterdam earlier this year. You mentioned that designers should take changes in city life into account. You talked about a conceptual shift from flaneurs to consumers to users. Now how then do you design for ‘users’ rather than for ‘consumers’?

AG: In the discourse of interaction design I’ve always opposed framing people as mere “users.” I’ve always found that very reductive and offensive. The irony is, that when I started to write about urbanism, when I was first thinking about how to properly characterize people interacting with the networked city, it struck me as being exactly that: actively, proactively, engaging with the environment, *using* it. Picking up functionality and *using* it. If the design of urban computing is about empowerment, then it’s also largely about how we might allow the “user” to become an active, conscious co-creator and co-participant of the systems they’re implicated by. And to do this in such a way as to lower both the information costs bound up in making a decision, and the opportunity costs of having made one. Where before, in the complexity of urban situations, we’ve constantly been compelled to make choices based on outdated or insufficient information, we can now do better.

Concrete example: think of the GPS systems that incorporate real-time traffic data. As you drive through the city and encounter successive decision points along your route, at each step of the way you’re able to make a choice based on the actual, immediate state of the system you’re immersed in. This closes the loop, vests the power in your hands, redresses information asymmetry. And that state of information – which, by the way, your own actions are a constituent of – has been represented in such a way that you can use it to make better decisions: left or right, freeway or surface streets, optimized for time or for energy consumption or for cost.

Now it’s clear that what we’re talking about here is utterly banal, something as completely unsexy as: am I going to change lanes here? It’s already well-assimilated and unremarkable. But I don’t think we should for a moment miss that it is also, and at the very same time, quite profoundly empowering. And even, dare I say, transformational.

More, the new forms of data visualization that I refer to in this example and elsewhere – even as they promise to give us a better sense of what is going on in the city – these visualizations are not unproblematic. They’re so incredibly seductive and visceral and compelling that we tend to take them at face value, but again there is always an underlying layer of politics in their assembly and representation. I always say, for example, that there are three concentric circles of data that might potentially be represented: quantities in the world that can be sensed in principle, the subset of those for which economically cheap and robust sensors exist, and the subset of those values which actually wind up being sensed and plotted on some kind of representation. When Co2 sensors are cheap, we wind up with a great many maps of carbon dioxide emissions, and so on. And so the pictures of the world that we create and consume are based on particular economic and physical constraints, and this winds up inflecting the judgments we’re able to build on them.

Or another example: who names the categories that we measure? For instance, Stamen’s Oakland Crimespotting project gives a very vivid sense of where street crime is happening in that city. But in considering it, we should always remember that it is the police department and the juridical code they enforce that define, for these purposes, what constitutes a crime. They define the taxonomy of criminal acts, and often this will not mesh with the public’s perception. For instance, the Oakland PD chooses to not break rape out as a separate category. Rape is folded up with the aggravated assault figures, and so it never appears in the visualizations. If you’d allowed a bottom-up reporting of crimes by citizens’ groups, on the other hand, or by the victims of crime, you would end up with a completely different map of the city. This is absolutely crucial for us to keep in mind as we encounter these compelling visualizations.

Michiel de Lange
Michiel de Lange (1976) is a part-time Lecturer in New Media Studies at Utrecht University, and a researcher and adviser of new media and urbanism. He is trained as a cultural anthropologist, and holds a PhD in philosophy (2010) with a dissertation about mobile media technologies and urban identities. He collaborated in a locative media art & science project (www.nomadicmilk.net). He worked for Knowledgeland, a Dutch think-tank that aims to strengthen the knowledge-based society. Here he worked on several projects at the intersection of ICTs and the city, e.g. co-organizing the Creative Capital conference. He also volunteered and worked for Cybersoek, a computer neighborhood center in Amsterdam. He is advisor e-culture at Mediafonds. Michiel is on Twitter and LinkedIn.