Why visual ambiguity is one of the greatest tools for design
George Stiny is Professor of Design and Computation at the Massachusetts Institute of Technology in Cambridge, Massachusetts. He joined the Department of Architecture in 1996 after sixteen years on the faculty of the University of California, Los Angeles, and currently heads the PhD program in Design and Computation at MIT. Educated at MIT and at UCLA, where he received a PhD in Engineering, Stiny has also taught at the University of Sydney, the Royal College of Art (London), and the Open University. His work on shape and shape grammars is widely known for both its theoretical insights linking seeing and calculating, and its striking applications in design practice, education, and scholarship. Stiny has recently completed a book on design and calculating — Shape: Talking about Seeing and Doing (The MIT Press, 2006; and HERE) and is the author of Pictorial and Formal Aspects of Shape and Shape Grammars (Birkhäuser, 1975), and (with James Gips) of Algorithmic Aesthetics: Computer Models for Criticism and Design in the Arts (University of California Press, 1978; and at www.algorithmicaesthetics.org).
In this conversation, George Stiny elaborates on what he values most in design: seeing. The ability to talk about what we see is indispensable to design. However talking, for Stiny is not a straightforward process; it incorporates interpretation, manipulation, and even omission of the things we see. The binary nature of computing and its symbolic structure proves inadequate to deal with Stiny’s shapes, thus he prefers talking about calculating as opposed to computing. The discussion covers broad ground, aiming to develop well-structured insights about the role of calculating, computing, and computers in design. However, the conversation also turns to talk about Leonardo’s sponges, Duchamp’s Fountain, and the renowned MIT Design and Computation Group. The conversation took place on May 21st, 2012, at Massachusetts Institute of Technology, Department of Architecture, Cambridge, MA.
Onur Yüce Gün: What is/what should be the relationship between design and computing?
George Stiny: I like to think about the relationship between design and calculating, as opposed to computing. The reason I like “calculating” is because it is much more down to earth. It sounds very ordinary, so it’s probably a surprise that it’s capable of doing anything that has to do with art and design.
Onur: How is computing different than calculating then, especially in its relation to design?
George: I’d like to save the term “computing” for the actual use of computers, which is different than the kinds of questions I usually try to ask and answer. As far as I am concerned, the relationship between design and calculating is equality. I say that for two reasons: one, I think when you design, you are actually calculating in a visual sort of way, whether you know or not, and the real central issue, at least for most of my work, is to try to figure out how calculating includes design. I think the formula runs amuck when people make design look like calculating in the usual sense. When you do that, it diminishes design, and as a result, calculating, too.
Onur: So calculating is not something that is auxiliary to design, as opposed to common belief…
George: If in fact calculating can be extended to include design, and I think it can, then that runs the formula
design = calculating
in the other direction. Then the question becomes: “What do you learn about design by calculating?” You can pick particular kinds of styles or designs; you can investigate how rules change. You can come back later and apply different rules to change designs. It becomes very dynamic, a very open-ended kind of process. But this all depends on making calculating generous enough to include art and design. That really can’t be emphasized enough, because it’s not making design conform to calculating; it’s quite the opposite: as a result, calculating becomes more than it usually is.
Onur: So, as I understand, you are expanding the meaning and potential of calculating when you use the formula “design = calculating.” We see this equality quite often in your book “Shape” (www.stinyshape.org) — and you have a chapter entitled “What Makes It Visual?” (Stiny, 61) Could you elaborate on this question?
George: First of all, the title of the chapter “What Makes It (Calculating) Visual?” speaks to the central issue of the whole enterprise. If you show that design can be carried out in visual sort of way, you answer the question that I just posed about the relationship between design and calculating. In order to see that there is a difference between visual calculating and ordinary calculating, you have to first look at ordinary calculating. The classic model there is the Turing machine, or maybe grammar or syntax in linguistics. And really what this has to do with is symbol manipulation. Symbols behave like little stones or discrete components; you can shuffle them around, put colors on them, and so on. However when they combine, they keep their original identity, so in combination they’re independent. A symbol is a symbol is a symbol — always!
It’s seems to me that Turing machines are really just one model for what calculating is all about. Since they came out of logic, they deal with a system of rules with true and false, or zeros and ones — discrete kinds of entities. It wouldn’t surprise me if Turing were to welcome an alternative kind of calculating that comes out of the visual, artistic enterprise, that requires something quite different than zeros and ones — the discrete mathematics that he used for his calculating machine.
This is just a generalization. It’s a very natural generalization, and it’s one that opens up calculating to include art and design — ambiguity, things changing, no vocabulary or symbols to begin with. Calculating in the visual sense doesn’t depend on vocabulary. It doesn’t have primitives, atoms, symbols, or units. These evolve during the course of calculating; they aren’t there at the start. By calculating visually, you allow anything you see to enter the process of calculating, and that’s quite different from what happens in a discrete process, such as a Turing machine.
Onur: Then, is design an entirely visual act?
George: When we do things visually, what we do now may not be what we see next. Here I might add a little figure: if I take two squares and add them together, I might get three squares or two L’s (Figure 1), or I might get four triangles or a bunch of pentagons, and big K’s and little k’s (Figure 2). I might get any number of different kinds of things that don’t preserve the atomic, unitary, symbolic properties the squares. Now if I represent this in rules that say square-one goes to square-one plus square-two, I can’t see that these squares go together to do something quite different from squares. That’s what I think is required to make calculating visual, and that requires generalizing the idea of identity that preserves symbols, and extending it to something I call embedding. And the interesting thing is that identity is a special case of embedding, so by doing visual calculating you automatically get symbolic calculating for free.
Now having said that, there is the deeper question: “Is design — is art — entirely a visual act?” Well, certainly design and art have a lot to do with making, and if you look at school curricula or schoolteachers talking about art and design, they almost never mention seeing, but always mention making. There’s a lot of enthusiasm for making in art and architecture. People talk about making all the time. At MIT the model is “Minds and Hands” (“Mens et manus”), which is thinking and making. But without seeing, it’s hard to understand how there can be any art at all. There are many cases in art and design that have nothing to do with making — that are described in terms of seeing. Duchamp’s Fountain is a nice example. He didn’t make anything, and it’s one of the most famous works of art in the twentieth century! Certainly you can show that making requires seeing. But just looking makes new art, too — it becomes observational.
Onur: So, there are minds, hands — and then eyes, maybe?
George: You can’t really avoid seeing. By focusing on seeing you include everything that happens in art; but if you focus only on making, you end up excluding many things. What happens to the object after it’s made? Artists may look at their artwork and change their minds about what they see. And this may change what they’ve done. So from my point of view, the observational component is the key, with everything else following along. Without seeing, art and design “disappears.”
So making calculating visual is the first thing to do — the visual aspect of calculating is at the very heart of art and design.
Onur: Then, how do all these ideas affect the ubiquity of computers and computing? How do we move from calculating with shapes and rules in your visual sense to computers and computing? What should the role of the computer be in a designer’s life, and is this the case for the actual use of computers in contemporary design practice?
George: When I started doing this stuff many years ago, designers didn’t want to have anything to do with computers. So there’s been a major change in the way designers think about computation, as opposed to calculating, and how they approach computers. Certainly today, it’s hard to imagine a serious, large-scale design practice that hasn’t invested heavily in using computers. However, I think this investment is mostly used for representation in computer aided design. I think it’s more of an archival kind of enterprise in which you are really trying to provide representations to evaluate designs and to track them throughout their lifetimes.
Practices utilize computers in ways they find useful to further practice. Does that mean they are doing calculating in the sense that I describe? The answer is ‘no’. I think what happens when people are designing and calculating visually is that there is a kind of free flow of ideas — use of rules or schemas. It’s a very dynamic, interactive process between the designer, the artist, and their work. One becomes engaged with whatever he or she is doing: an architectural drawing, a painting, a sculpture. And the action is calculating — seeing and doing.
Making calculating something that a computer can use today requires a certain kind of translation. And that translation is really a description of the object that someone is reacting to. That description can (and will) change when you look at the object again and again, trying to correct it all the time. So I think the computer doesn’t really have much to do with design in the sense that I am talking about. The real fear is that people will give up on design in this robust, general, visual, artistic sense. The kinds of things you can do on a computer make life easier and straightforward and I think that could be a serious loss for architectural design.
So my conclusion would be this: people should use computers in practice whenever way they want to. But don’t forget that there’s something else involved that has to do with how you see and talk about designs. Calculating is much more generous and flexible than the kinds of things you can do using computers today.
Onur: I am not quite sure if that flexibility is always acknowledged by designers. Computing or calculating in design is immediately associated with algorithms, and their complex geometric outputs in contemporary architectural practice and education. Maybe that is why computing and computers are embraced for their potential to help harness variety in design. However, simulations and performance analyses fall short in determining the proper designs from amongst a virtually unlimited number of possibilities. Is there a way to benchmark countless versions of a design to pick the delightful few?
George: There are number of aspects to your question. In the three categories of the Vitruvian canon — firmness, commodity, and delight — certainly delight is at the center of architecture and art. It’s the one thing architects or artists provide that’s independent of any other discipline or profession. The issue of variety is another thing; the question is how many things you produce might be delightful? I think probably there are fewer things than most people would realize.
So my sense is this — in the first place you have to get delight right. Evaluating things in terms of performance is somehow troublesome. Think of silly buildings designed with computers that simulate how heat moves around spaces — and yet in the final picture, you see somebody in a room with a little space heater trying to keep warm.
That kind of performance evaluation and simulation in which physics is involved is really hard. It’s just that the mathematics for it is not fully implemented. It’s not up to the task yet, but it might become so one day. Will that conclude the design problem? Well, probably not, because there’s still delight, and that’s something constantly evolving, not just in respect to the designer but also in respect to the people who look at designs and use them, interact with them. Because they’re all seeing, they’re all in this important dynamic process. There’s no end in architecture, there’s no end in design, there’s no end to looking. That’s what’s exciting.
Onur: This reminds me the debate about the huge market formulated around building energy analysis and simulation tools. Their accuracy is still at stake, as you highlighted. How valuable are these tools or their byproducts?
George: Well, there’s something valuable in physical and performance analysis; certainly, it’s very important. In the past, we had vernacular designs and their physical and environmental properties were improved over time. They converged to a certain kind of design and certain things were kept or changed after this benchmarking process happened over time. I think design still works in that way to some extent. How accurate and complete are simulation results for a huge building? I don’t know. Nonetheless, what we do now is perfectly fine. The tools are definitely better than nothing.
But analysis is really a research issue. Building technology is a very serious kind of enterprise, and I think it goes hand in hand with ideas of calculating and delight. I’ve been outlining this approach in different ways; the physics I use is negligible. So there are options in which category of the Vitruvian cannon you’re trying to apply.
Onur: What about parametric design tools? They generally tend to hide computational complexity, yet they still enable users to generate geometric as well as visual complexity. Are we really able to see what we are dealing with, when we are working with such tools?
George: I am not a big fan of parametric design, mainly because I think it simply scratches the surface of what happens in design. The trouble with parametric design is that it forces you to divide something into components and treat them as symbolic objects. Then you vary those components to produce something people might like. It may produce something, but you may not like it because of the components that are involved, or because you see it in a different way that’s independent from what you’ve produced. And that makes parametric design, very much like Leonardo’s sponge filled with colors, that he throws against a wall. There’s a splash — then the designer comes in and looks at the splash. This design part needn’t count in throwing the sponge. And that’s the aspect of parametric design that’s missing. I think people often times miss that, in a way this is just sponge throwing! You still need the designer to see what’s there, and this may not match what’s been combined and varied.
Onur: Then the designers are not necessarily able to see what they produce with parametric software, is that so?
George: Right, parametric design is a symbolic enterprise, and the visual kick we get out of it has very little to do with the things buried inside the machine. In that sense, the machine is hiding what it is you’re excited about when you look at a design. It’s the sponge!
Another problem with parametric designs is that they mostly look the same. You get to the point where you don’t want to look anymore: variations are the ones you expect, colors are the ones you expect, layouts are the ones you expect. So there’s no kind of creative spark to them, and yet for a designer who re-engages the material visually, I can see how that becomes a kick, because they’re getting results they don’t ordinarily get. It’s simple, easy — press a couple of buttons or write some code, and you can get this wonderful array of things. The problem is that they look the same. And they look the same because they’re combinatorial; they’re putting components together without seeing how components interact. Change what’s there. That’s what’s visual. That’s where the kick is. That’s what’s missing.
I can show you a lot of calculating that demonstrates the poverty of parametric variation. For example, if you rotate the three triangles in this figure, you end up with two triangles (Figure 3). There’s a huge discontinuity that parametric design doesn’t pick up, and that discontinuity has a lot to do with what visual calculating is all about. Three isn’t two!
Onur: Although the tools are ubiquitous both in practice and academia it is hard to argue that the underlying logic (of parametric systems) is studied enough for their proper utilization. Inexperienced designers may either get stuck with whatever form the parametric program produces or suppress their own design intentions dealing with the mechanics of a system they don’t fully comprehend. How could they find their way out?
George: If you want to know how these systems work, they’re certainly worth exploring. But you have to realize that once you find out how they work, things might not be as interesting as you thought. I think the scary part of this is that people use these things and never bother to find out how they work. As a result they get stuck, as you point out, with the kinds of designs that are merely “available”. Their own visual intuitions are no longer important.
The other problem is this: suppose that you have a parametric model of something and you vary things, and the boss comes in and says “Hey look at this cool thing up here, why don’t we change that!” But that’s not a part of the parametric model, so in order to incorporate the change, you have to go back to the beginning, over and over again, which can take a lot of time and cost a lot of money, and you tell your boss that. Your boss either fires you for not doing the model in the right way — which is his loss — or he says “I guess you better not change it because it costs too much money”, in which case it’s the design’s — and again his — loss. The tool, the parametric system takes over design. The loss of design activity, the loss of visual interaction, is the real thing to fear.
Onur: You have been one of the key individuals in the establishment of the renowned Design and Computation Group at MIT. This group has influenced many programs worldwide. Could you briefly talk about the spirit of the program, and compare and contrast it to others?
George: When I started the computation program in architecture at MIT about 15 or 16 years ago, the one thing I wanted to do was to make sure that calculating, as opposed to computation, was the way to investigate design, not simply an application of computer tools in design. I think the main thing that has kept the program fresh and viable, and active and exciting — that has provided a way for people to learn about design — is to actually view calculating as a good way of talking about design in the way we’re used to. In fact from what I’ve said, design is calculating. The real emphasis is on design, not computer tools.
I think many programs elsewhere in the world emphasize the computer application aspect of design, and as a result, at least from my point of view, these programs don’t have the exciting viability of the program we have at MIT. If there were no computers whatsoever, if we didn’t have even one, no one ever invented one, but we did have my notion of visual calculating, we would still have the computation group at MIT. If there were no computers, and computer software and programs, many other schools wouldn’t have a calculating component as a part of their architecture or design programs.
Onur: This might be the clearest way to explain it!
George: The key thing that makes the program work at MIT, and distinguishes it from many other programs is that it looks at calculating as a serious intellectual, academic issue that has to do with design, independent of existing software, computer programs, or anything you can buy off the shelf. Somewhere else, you would need to get that computer program, and then need to learn to use it to design. That’s not the emphasis in the program at MIT; that’s not what we’re trying to do. Calculating as a serious, intellectual enterprise that’s independent of the latest computer software, latest gadgets, 3D printing machines, etc. The subjects we teach have to do with thinking about architecture and design, and if there’s a computer tool involved for a supporting role, that’s good. We do run the whole spectrum from the theoretical kind of enterprise — which is the one that I’m most interested in — to practical concerns that have to do with visualization, parametric design, rapid prototyping, all of the rest, and fill in everything in between these two end points of the spectrum.
So I guess the short answer to your question is that the program thinks of architectural design as architectural design, and doesn’t worry about the computation part, other than thinking about architecture and design as a kind of calculating — which is different than computer tools.
Onur: This should give a lot of clues about the program to curious candidates who are considering applying.
George, thank you very much for sparing time for this inspiring conversation. Do you have any final remarks for our readers?
George: I think I would like to return to what I said initially. One thing that I really like to emphasize is the formula “design = calculating”, and to emphasize that thinking about design in terms of calculating really enriches calculating by expanding it — so that calculating does things that aren’t found in any of the computer tools that I know of today. We think about things with our eyes. And it’s that central role of seeing in design that I find most fascinating. If you ask me to think about anything that has to do what makes us creative and what makes us human, it’s design and calculating. And I think that this is the most important aspect in the whole enterprise.
There are all of these problems about what it is that you’re doing when you create and design things, when you engage your eyes. I think all of these could be thought about in terms of calculating. And how you do that is what really interests me.