The Challenger: Dear Daniel and Gilles…


Publication History:

“The Challenger. Dear Daniel and Gilles… ” Flat Out 4 (2020): 28-30

The text posted here may be different from the published version. Please only cite from copy in print


Dear Daniel and Gilles,

In the third issue of Flat Out Peter Eisenman wrote a critique of my recent book, The Second Digital Turn. In the form of a letter addressed to me, Peter’s review tackles some key aspects of contemporary digital design that are now best illustrated by your work: your respective theories of “mereology” and “discretism” extend, reinterpret, and continue, in a sense, my plea for a “discrete” turn in computational design, and bring some new arguments to bear, both theoretical and design-based. This is as it should be; when I first conceived that book, during my last years at Yale, I knew Peter well, but I had never heard of either of you—not surprisingly, as your computational work did not yet exist. So let me hand down the discussion of this matter another generation, so to speak, and relay some of Peter’s queries to you directly, with some of my own added.

Peter’s letter raises some general points about the meaning of forms, and the political meaning of architectural forms in particular; and from those premises Peter proceeds to denounce digitally discrete architecture as meaningless at best, in the literal sense of the term—if not politically unsavory. In short, Peter’s argument reads as follows: smooth, curvy, and continuous forms are inherently compliant, whereas rough, angular, and discrete forms are inherently oppositional; hence smooth forms are a way of saying “yes”—and in architecture, in particular, of saying “yes” to power. Likewise, rough forms are a way for designers of saying “no,” and of saying “no” to power—of being critical of power, and challenging it. Peter seems to imply that this may be due to some universal, timeless meaning of forms, but that is unproven: philosophers and psychologists have been arguing on that matter for quite some time, with no clear-cut conclusion to date. But Peter adds another, strictly technical and more straightforward argument: architectural forms also have meanings that depend on how easily, or not, we can design and construct them. For example, complexity (or disjointedness, fragmentariness, etc.) went against technology (was “oppositional”) in the 1970s and 1980s, when complex buildings were relatively difficult to realize using traditional handmade drawings and the mechanical construction tools of the time; but complexity is way more technology-friendly (hence “compliant”) today, as digital tools now make all kinds of intricate forms easy to notate, calculate, and fabricate. Indeed, using digital tools today we can build identical shoeboxes, pliant blobs, free-form potatoes, or the apparently random, messy aggregation of billions of different voxels, all at the same cost (at least in theory)—so none of these forms today is more “oppositional,” or “critical,” than any other, and none, when individually taken as a supposedly meaningful form, can express any degree of “resistance to power” any longer (Peter’s words).

Peter’s argument is sound, and can be generalized. Think of Mendelsohn’s Einstein Tower, or Kiesler’s Endless House: their organic forms were meant to be almost impossible, or at least very difficult to build, when their designers first conceived them. Back then, that was  a way for Mendelsohn and Kiesler to spell out their rejection of the dominant technologies of their time—that was their way of saying “no” to the technical logic of mechanical mass production. But when similar organic shapes were adopted by Greg Lynn, for example, or other blob makers of the 1990s, that was a way for them to display their enthusiastic endorsement of the digital technologies of their time—that was their way of saying “yes” to the technical logic of digital mass customization. The same organic forms would hence have been “oppositional” in the early twentieth century, and “compliant” in the late twentieth century: same forms, but with different meanings at different times. On the other hand, it is equally evident that this interpretive model only works in broad strokes, or when seen from afar: when one gets into the nuts and bolts of design and fabrication there is always a solution that is technically easier than another; each tool, software, and technology favors some forms and make others more laborious to obtain, and within each technical system, mischievous designers will always find ways to act mischievously (or oppositionally, or critically, or dialectically).

Assuming, that is, that architects should have an irresistible inclination toward being permanently at odds with the rest of the world. Daily experience suggests that this is indeed often the case; but this was never a legal mandate, nor does it appear to be a law of nature. Most human beings can tell the difference between stuff they like, and stuff they don’t like—and they can be alternatively compliant or oppositional, as need be. Why shouldn’t architects do the same? High modernist structural linguists, pursuant to the Saussurian dialectics of langue and parole (code and creativity), would have had an explanation for that—but let’s leave that discussion for another day. And at any rate, it is a fact that back in the 1990s the smooth and continuous forms of early digital design, on top of making a very alert use of new spline-modeling technologies for the design and fabrication of aerodynamic surfaces, were also often meant to symbolize the very “compliant” political stance of their designers. We know that because some protagonists of the early digital scene clearly said so back then: the late 1990s may have been the only period in the history of the world when an architectural avant-garde emphatically endorsed the political status quo. As a result the digital style of the late 1990s, now often called “parametricism,” but which should more pertinently be called the style of digital streamlining, is to this day often associated with neoliberal, neoconservative, or libertarian ideologies.

But that was then. Times have changed, and so has the technology. Discretism is now the best way to marshal the unprecedented power of artificial intelligence, robotics, and machine learning, and the discrete forms that often result from the pertinent exploitation of those technologies simply show that we are trying to make the best use of the best technologies now available to us (and again, a linguist would say that such meanings are conveyed by direct indexical denotations). When I say “best use” I mean producing better things for more people at a lesser economic cost, and at a lesser cost for the environment. Assuredly that would appear as a fairly consensual project—one that no rational human being would oppose.

Yet many today do, because we do not live in rational times. Fascism is not a rational ideology; it wasn’t in the 1920s and 1930s and it isn’t now. Nor are racism and nativism—they never were. But these are the times we live in, so our challenge today is to say “yes” to the best use of technologies that can serve us well, and “no” to the fascist ideologies that are, once again, bent on destroying our civilization. We should be compliant with technology, and oppositional to politics.  

One could think of different ways to do so. On the one hand, one could argue that, by simply using today’s technologies at their best, we would be already pursuing a project of rationality that goes counter to the irrational politics of today’s fascist ideologies. Therefore, by simply saying “yes” to today’s technology (as above defined: in a comprehensive, holistic balance of all societal and environmental factors and of their costs) we would already be saying a strong “no” to today’s fascism (and indeed, as a proof a contrario, today’s fascism is viscerally averse to any such use of computational tools for automated manufacturing). But others would argue that architecture must be more than that—as it always was. That architects must add symbols, figures, and signs, to communicate, corroborate, and amplify the statements that architecture makes by dint of the purposes it serves (or sometimes, perversely, to voice the designers’ disagreement with regard to the assignment they received, accepted, and carried out). Many professionals today agree that there is no way to avoid the use of computational tools for design and construction, because these are today the most effective tools of our trade, and if we renounce them, we effectively renounce our trade—or delegate it to others. But how can we make an intelligent use of these tools and still express our “resistance”, or “challenge” (Peter’s word), to today’s political autocracies, which we oppose and resent? 

The unprecedented power of today’s artificial intelligence is already out of sync with the limited number-crunching performance of human intelligence. Many traditional mereological tools, like proportions and ratios, were for centuries used as shortcuts to deal with numbers too big for human handling—to convert big numbers into smaller numbers, and put some human order to nature’s apparent chaos. But those mereological devices were made-to-measure for the limited processing power of our mind; and those limitations no longer apply to today’s artificial intelligence. Computers today can deal with any amount of data without any need for sorting, hierarchical organization, formalization, generalization, or structure. What will be the role of mereology in a post-human computational logic that no longer needs proportions, nor ratios, of any kind? 

On the other hand, artificial intelligence applied to robotic machinery already allows us to envisage artifacts composed of an unimaginable (literally) number of parts, to be assembled by an equally unimaginable number of robotic gestural movements. Parts readied for robotic assembly need not be identical, nor identically reproduced; likewise, automated, robotic gestures need not be identically repeatable: in both cases, endless variations may come, theoretically, at no extra cost.  Why is it then that brutalist chunkiness is now often seen as the marquee visual identity of experimental robotic? Why are so many talented computational designers today tinkering with modularity and prefabrication? In pure technical terms, computational automation today needs none of that: the marginal costs of computational production are (in theory) flat; using computational tools, making more of the same will not make anything cheaper. And if the reason for today’s computational brutalism is eminently non-technical, what is it then?

You are the designers, so these questions are for you. I think I have some inkling of what you are up to; but I also feel that you can better explain that in your own words. How can we be computational AND critical today? Answer this, if you can, and you will have answered Peter’s query (and mine). 




Gilles Retsin Architecture, Real Virtuality, Royal Academy, London, 2019.

View of installation made from CNC-milled plywood kit-of-parts held in place by lateral steel rods.

Photo by Studio NAARO


Daniel Koehler (with Genmao Li, Chen Chen, and Zixuan Wang), Project WanderYards, Coder le monde (Coding the World), Centre Pompidou, Paris, 2018.

View of model showing a building conceived as a city made from architectural parts (i.e., mereologies), 2017. 3D-printed polylactic acid, 1:250, 78 x 88 x 107 cm.

Produced in Large City Architecture Research, The Bartlett School of Architecture, University College London. Photo by Daniel Koehler.





Flat Out 4


“The Challenger. Dear Daniel and Gilles… ” Flat Out 4 (2020): 28-30