Stephen Ervin, lecturer in Landscape Architecture and Assistant Dean for Information Technology at the Graduate School of Design at Harvard University gave the closing talk at the recent GeoDesign Summit. Ervin examined both the near and far future of geodesign in that talk (viewable online here). V1 editor Matt Ball had a conversation with Ervin that covers the current practice and potential future for more responsive design and infrastructure.
V1: I understand that you’ve just returned from China, where they are doing some large-scale city building. Are you seeing evidence of geodesign at the city planning level?
Ervin: I co-teach with a Chinese colleague a studio in the Landscape Architecture, Planning and Urban Design and Planning program here at Harvard. For the past three years we’ve taken a dozen students to China for a week and confronted them with the problem of the expansion of Beijing. For the last 30 years it has been expanding mostly into its flat southeast agricultural plain. Anybody who thinks about the problems of food, security, and sustainable practices realized that this is an unsustainable plan.
This year my colleague, Yu Kongjian, a landscape architect practicing in China and an academic at Peking University, has come up with the “foothills” proposal. Instead of building in the flat lands, he suggests: build up into the gently sloping foothills that sit between Beijing and the steep mountains to its north and northwest.
In this year’s class we’re looking at what’s involved in building a new city of 100 to 300,000 people in this hilly arid region. It’s a huge problem, but it’s an important one as Beijing is expected to double its population in the next 20 to 30 years.
I can’t say that we’re actively using the best geospatial tools and practices in this project. Our students are as likely to be walking around with sketchbooks and digital cameras as with well-equipped tablets with ArcGIS or equivalent. It’s a process of evolution as we figure out what are the right tools, and how to use them in the right way. What’s so exciting about geodesign right now is that we’re all still working to figure that out.
V1: One comment that you made in your presentation at the GeoDesign Summit really caught my attention, when you asserted that we’re doing geodesign for the 10 billion people that will be on our planet soon, and not necessarily for trees or the environment.
Ervin: It’s easy to get lost when you’re designing a dam, a subway station, a park or new city, to think that the contour lines and tree symbols that you’re pushing around on paper or on-screen are what you’re working with and for, but it’s not, it’s for the people.
V1: You made another point in terms of designing for human health, and in Asia what you see in terms of environmental degradation is starker than elsewhere.
Ervin: Human health is equally important everywhere in the world, and it’s more threatened in some places. It’s certainly the case that there is air pollution and water pollution in China, and in and around Beijing it is quite severe, and that’s a serious problem.
While design can help combat these problems, it’s hard to get a handle on air pollution. It’s a multi-faceted thing, but we can have a tertiary effect on particulate and other kinds of noxious emissions with the adoption of various green technologies. We can do calculations on second and third order considerations for air and water pollution, and work toward ways to combat that.
Designers can‘t easily suggest what to do to solve pollution, but in designing a city for 100,000 people, among the many things that we’re thinking about is how to support a clean environment and a sustainable set of processes across the board. These processes include both hydrological and industrial processes, and others.
V1: It’s interesting how policy can drive the process, such as in the United States where the Sustainable Communities program combines funding from the Dept. of Housing and Urban Development, the Environmental Protection Agency, and the Dept. of Transportation in a multidisciplinary way. Are there encouraging signs that a more holistic planning process is taking place more globally.
Ervin: I don’t know that I have such a broad view to understand if it’s happening everywhere. What little I know about the European situation, I understand that it’s a more integrated planning process where, for example, transportation takes into account issues related to solid waste, air pollution, and all kinds of things. I have an impression that these kinds of discussions aren’t happening in China very well, and I think that’s still a maturing process there. Certainly the need for interdisciplinary approaches exists everywhere, because everything is connected to everything — that old ecological truism.
V1: Is it a toolset that will help bring greater collaboration or will mandates force the approach?
Ervin: I’m a big believer in tools and toolsets; mandates are a little more difficult, but I guess they are a tool in the policy toolbelt. Sometimes it seems that we need mandates, such as “thou shall not employ children in horrible working conditions”. If we need that mandate, then we better have it.
V1: One of the other areas of your presentation that interested me was the incorporation of simulation, with design that is informed through tools that are connected to process through inputs. This extends design through animations that might predict outcomes. Is that a focus of your work?
Ervin: I got a Masters of Landscape Architecture just as the micro computer was being invented. I was in school from 1978 through 1981, and we saw both the Apple and the PC appear. During the same time, I was being taught the traditional methods of hand sketching, and while we were conceiving of systems as dynamic, all we could do was try to draw their effects. We could have a spatial mathematical representation of process, but we didn’t have good tools for visualizing that.
I became aware of what computers could bring to this. In the first place it brings numeric precision (if that’s what you want) and accuracy (if that’s what you have) to an otherwise “make it look good by eye” enterprise. Computers enabled our ability to actually visualize how tall a retaining wall was going to be, and measure what it would look like in the distance, rather than eyeball it. We had pretty good ways of eyeballing rather well- dimensioned drawings, but the complexities of horizontal road alignment and earthworks calculations, and hydrological network analysis were all pretty loosely coupled — or uncoupled — to the drawings that we were making.
We were able to make some connections by moving into an early computer-aided design environment, where we were able to take some of the early system simulation tools, and think about how they might begin to inform what we were doing. The feedback loop where design is informed by tightly coupled simulation and analysis is something that I’ve engaged in over time.
Modern examples of such feedback loops are forest fire simulations that show how fires might move up a hillside under certain conditions or streamflow animations or traffic simulations. These are all areas that are hard to do statically or in a drawing, but it’s possible and getting easier with computation, and that’s why computation or computer-aided design is so powerful.
V1: You had touched on the design continuum, where simulations that view what is actually happening are on one end, and the initial conceptualization and diagramming process are on the other end.
Ervin: My Ph.D. thesis was about diagrams, what are they and how might we make them, and what would it mean to have a set of computer tools to enable that. Back in the bad old days, twenty years ago, there was discussion that “computers can’t possibly be helpful in the conceptualization and visualization stage of design, because they are too precise with hard lines, and they aren’t intuitive, and don’t enable a flow of ideas, etc.” The coupling of eye to brain to hand couldn’t possibly be supported by computation. That still remains a barrier to the software industry, and that’s why there’s a fascination with interfaces, gesturing, and tablets. We are trying to figure out a way of getting something as fluid as pencil on paper.
What’s fascinating to me about diagrams is that they’re not just about shape or form, they’re about other ideas that are embedded — connections, topology, system structure, and more high-level spatial relationships. I find all of that very fascinating, and the lack of tools in that area in terms of software still is very frustrating.
V1: With form-based design there is some means to tackle that, changing parameters and then quickly changing things as in the procedural realm.
Ervin: Procedural and parametric design is quite a fascination these days in architecture and environmental design in general. We begin to describe forms mathematically, and then by changing formulas we can change the forms, and perhaps hooking parameters to some other output such as a simulation, so that you literally have a responsive building or drawing. Such as: the more people we drop into the drawing the size of the building responds to get larger. Parametric is one forms of the simulation that I talked about.
That we might have cities or buildings or houses or parks that are responsive over a wide range of inputs seems inevitable. It means that we will design systems that can change and thereby have what we might call ‘behavior’, rather than just static forms.
What if roads could straighten out as traffic sped up, and get more curvy as traffic slowed down? Wouldn’t that be interesting? This is all science fiction that I’m talking about, but part of what I was asked to do was to talk about the future, whether that was ten years or 100 years away.
It’s certainly clear that sensors are becoming a part of the world that we live in. Robotics is all about devices that can sense on the one hand and actuate on the other. We already have all kinds of things like automatic door openers that are effectively sensor-based robotic doors, and we certainly see all kinds of moisture sensors and lighting sensors that turn things on and off. Knowing the way that information technology is exponentially changing, there will be all kinds of much more interesting sensors with which to design and build our environments in the coming years.
V1: What about the idea that sensors are better data collectors than humans, and that the combination of sensors and systems can be applied to tune systems without human input.
Ervin: One thing that is great about sensors is that they can hear, and see, and sense things that are invisible to people. That’s a huge advantage, and is person-less because they have a different range of perception. On that topic, of course what sensors and artificial-intelligence still aren’t as good at as people are, is pattern recognition.
As good as they may get, they still aren’t as good as humans at pattern recognition, and a deep question is whether robots will ever be as good as humans at activities like pattern recognition. It’s pretty fascinating that some cameras can now figure out where the faces are and focus on them, but there are aspects of pattern recognition that machines still struggle with, such as handwriting analysis. When machines completely fail, we call on a person and the person can usually figure it out.
I’m not scared or concerned about a world where all sorts of robots are doing things without human oversight or intervention, although I probably would insist on a big red button with an override somewhere so that human judgment can still prevail. I do read science fiction after all.
V1: A last thread is the role of the designer to look at the many possible futures, with the hierarchy of what is possible and the designer’s role to determine the best future.
Ervin: ‘Best’ is a highly subjective analysis. A lot of designers throw around ‘optimum’, but most don’t know what that means mathematically, and most designs aren’t optimized in any sense. The idea is that in any moment in time there are an infinite number of alternate futures unfolding ahead of us, and part of our job is to narrow that range to the ‘best’ (or at least preferred) one. Ideally, we come to the most desired future with whatever processes we have on hand, some of which are social (how do large numbers of people choose their future?), some are technological (what handles do we have to change future impacts?).
“What can we actually change today that might have a desirable impact tomorrow?” is a complicated question, and continually changing as our technologies change and our understanding of the linkages of systems change. Sometimes there’s a naive discussion of “the design process” – I say ‘naïve’ , because there are actually many design processes. People who know about how design happens concede that there is a part (or many) that we don’t understand; and there’s not any one ‘design process’!
Carl Steinitz addressed this at the GeoDesign Summit when he talked about an experiment on ways of designing: giving the same problem to different teams of approximately the same sets of talents, but forcing them to each use a different method. The idea is to understand the applicability of methods under the same constraints, which is a study still in its infancy. We don’t really have firm results, though it seems anecdotally that certain kinds of problems are more amenable to certain kinds of design approaches.
This is an important question in geodesign, because there’s the sense that if we get the software and interfaces right, and the sensors and algorithms right, then we’ll have it. That’s not true, because then there’s the whole question of design processes and how we engage people. That goes to questions of visualization, political organization, personal and communal motivation, and a whole host of complicated questions. This just reminds us that the enterprise of geodesign is huge and multifaceted.