The Entasis of Elon Musk

June 2019

Tamar Shafrir is a design writer and researcher based in Amsterdam.

Read full Bio

For the thickness of the shafts must be enlarged in proportion to the increase of the distance between the columns...the air seems to eat away and diminish the thickness of such shafts...they are sharply outlined by the unobstructed air round them, and seem to the beholder more slender than they are. Hence, we must counteract the ocular deception by an adjustment of proportions...For the eye is always in search of beauty, and if we do not gratify its desire for pleasure by a proportionate enlargement in these measures, and thus make compensation for ocular deception, a clumsy and awkward appearance will be presented to the beholder. With regard to the enlargement made at the middle of columns, which among the Greeks is called ἔντασις [entasis]...an agreeable and appropriate effect may be produced by it.

—Vitruvius, De architectura, 15 B.C., from Book 3, Chapter 3 (Temples and the orders of architecture: The proportions of intercolumniations and of columns)

The idea that an architectural rendering can be ‘real’ or ‘fake’ involves a transference of the logic of one medium—building—to the logic of another—drawing. Architectural rendering has always exploited the potentials of the page or canvas where money, knowledge, taste or gravity proved prohibitive. The most famous historical examples of this exploitation of the image would be Giovanni Battista Piranesi and Étienne-Louis Boullée, predecessors of more modern practitioners such as Le Corbusier, Archigram, Superstudio, Lebbeus Woods, and early Zaha Hadid.

Since the 19th century, when the École des Beaux-Arts formalised the architectural drawing in its modern conception—not only in its aesthetics but also as the conceptual key to the building—one could argue that the architect has become less implicated in building than in drawing as an aesthetic code, an act of creative production, a tool of communication, a set of instructions and a legal requirement. This split was ossified by the increasing technical expertise and clean office environment necessary to work with reproducible drawings like Mylar, cyanotype, pen plotters, and most obviously, computers. Indeed, one could say that architecture is more about communication than building, except for the brief periods of Modernist architecture in the 20th century that coincided with the dire need for new mass housing, namely after WWI and WWII and during Communism.

What is interesting, on the other hand, is that the computer introduced a new dynamic. It is only in the computer that the aforementioned transference between logics actually becomes instrumental, for a few reasons. One is the ability of the computer to aggregate information and automate actions, most recently through BIM [building information modelling] or Revit, the most widely used software. This is a form of digital design based on ‘smart objects’ rather than contextless and unconnected lines drawn on a screen (as on paper). One could draw a freestanding door in any vector program, but in Revit a door object has to be hosted by a wall object. Not only that, but Revit automates the visual language translation from 3D modelling (where the door is a modelled panel in the wall) to 2D modelling (where the door is represented as an abstract door swing and the actual shape of the door is only shown on another drawing called a door schedule, normally in reference to a specific architectural product). In fact, Revit automates as much as possible the process of form-making to architectural drawing production, reducing the inherent redundancy of the architect’s work.

Previously, translating from a 3D model to a 2D drawing was incredibly laborious and tedious, and even more laborious was updating individual drawings—floor plans, reflected ceiling plans, HVAC plans, fire egress drawings, etc.—when something like a partition wall was moved. Earl Mark, my professor at University of Virgina’s architecture school, who worked for Microstation/Bentley Systems in the 1980s, said that the initial philosophy for digital architectural drawing was always to have such an intelligent system, but the computer processing power remained insufficient for decades. Revit only entered mainstream architecture firms in the late 2000s, and until then, the figure of the ‘CAD monkey’ was endemic to architecture—as long as designers could handle the tedious, error-prone, keyboard-oriented practice of AutoCAD, they could theoretically draw anything they wanted.

But when Revit meets computerised manufacture (otherwise known as CNC, or computer numerical control), the issue of the real/fake becomes particularly charged. What one might notice from projects like DUS's 3D-printed canal house (see figure 1) or Gramazio Kohler’s robot drone bricklayers (see figure 2) is that they draw a direct line from digital drawing to mechanical production and assembly, cutting out the role of the human builder entirely. This proposition raises a few problems. One of them is that the act of creating shelter is probably the oldest occupation of mankind, and some conditions of interior space are intimately related to the proportions of the body and its relation to material assembly. Severing that link may have unintended consequences in terms of the inhabitability of our buildings. Second, this tactic reifies the 1960s infatuation with moulded plastic monocoque structures, which granted architects unimaginable formal freedom and made trivial concerns of thermal performance and adaptability—taking for granted that an unlimited supply of energy and material resources would always be available. This tactic inculcates a perception of architecture as space-making through closed surfaces rather than a perception of building as the assembly of materials, utilities, and thermal or moisture barriers.

A third problem is that architecture education today is almost entirely about image-making and very little about actual construction. This means sacrificing a great deal of knowledge about building methods that have worked for millennia, for the sake of taking out more humans from the equation of architecture (and with them their salaries, health insurance, and knowledge). Architects counter this anti-value proposition by racing to make themselves indispensable in the production of more and more dazzling drawings and more and more plywood pavilions. One might notice that the architects who are most interested in actually piling blocks on top of each other in a material- or innovation-oriented practice, such as Peter Zumthor and Frank Gehry, are not noted image-makers. Gehry in particular is known for his cartoonish models of crumpled paper and vague napkin sketches (see figure 3).

On the other hand, architects like BIG, headed by Bjarke Ingels, make their claim almost entirely on the power of the drawing, whether a sleek rendering or a clever diagram. But what happens when that drawing becomes real? Compare, for example, the rendering of his installation for Burning Man against the reality (check the comments on this article). Should this gap between the promise and the delivery in a festival structure give us qualms about his plans for Mars settlements? Such design for outer space is a good case study, because it is the place where human builders are least likely to have a hand in materialising the visions given to us by these architects. Therefore, it’s also the place where drawing will most literally transfer into building in the future—in other words, where the technical ability to command what happens in the computer (from modelling to rendering to CNC manufacture and assembly) translates into the ability to shape the built environment.

A more worrying example is the design for a Mars colony (see figure 4) by Tesla and SpaceX founder Elon Musk. Consider how little has been learned from city planning in more than 50 years: the rendering is reminiscent of Le Corbusier’s plan for Chandigarh, based on a human body plan and critiqued in Christopher Alexander’s 1965 essay “A City is Not a Tree”.

One of the unexpected benefits of the inefficiency of the AutoCAD process is that the drawing was just a set of instructions, and a great deal of ad hoc problem-solving could take place on site. People with experience in building could look at a design that appeared viable on paper or screen and know that it might be completely impossible to execute in reality, even if it was something as banal as not being able to fit a screwdriver into a gap in order to attach a panel to a base structure. Do we really want to entrust even more of our built environment to people who are trained as visual communicators? Even an architectural master like Alvar Aalto was guilty of designing a horizontal chimney in his Paimio Sanatorium, conveniently forgetting that hot air travels up, not sideways. And in the case of Elon Musk, do we want to collapse the distance between those who can make intelligent machines and those who decide what they are used for?

Of course this argument relies on the fantasy of machines running flawlessly, which is exactly that—a fantasy. However, the fake/real rendering debate does not address the positive aspects of their falseness. In some cases, what we are implicitly criticising in rendered fantasy images of exotic architectures is the inability of architects to reproduce them perfectly in reality. Of course, that gap will get smaller and smaller as the computer becomes the driving force throughout more of the process of architecture and construction—but the narrowing of that gap is a threat to the quality of our built spaces. Two recent examples spring to mind, one very small, the other very large.

Firstly, at the very small scale: if construction processes become increasingly reliant on CNC manufacturing then we will be tied to what we can draw on the computer. Imagine drawing a curve by hand versus one on the computer versus crafting a curved component. Drawing a curve by hand is constrained by the body, its proportions and motor skills, and the qualities of the pen or pencil and the medium, whether paper or glass or sand. In order to draw a digital curve, meanwhile, the computer has to be able to calculate mathematically the coordinates of every point on that curve. Of course, there are different algorithms and methods to draw and calculate the curvature such as parabolic, circular, spline, Bézier curve and so on, but ultimately the mathematical precision of the computer is as much a limitation as it is an asset to creativity in complexity. The surface, topological and formal vocabulary of the architect is given artificial limitations by the software and available manufacturing processes. Add in the further computation of CNC cutting paths and the necessity to use only materials that can approximate the grainless uniformity of digital ‘matter’, and we can start to contemplate how an entirely computer-driven process negates more possibilities than it creates.

At the very large scale, consider how Rem Koolhaas talks about the OMA skyscraper De Rotterdam in this Guardian article:

"The most important thing about this project is your perception of its size and mass as you drive over the bridge" says Koolhaas, hunched over the wheel...as we begin our ascent of the boomerang-shaped road of the Erasmus Bridge. "We calculated how the view would change as you approach along this curving path" […] "That's all you need to see. The rest is just a cheap office building", he says, before leaving me to explore the interior for myself.

Funnily enough, there is another building that was also designed largely around how it would be perceived by its audience—the Parthenon in Athens, probably the most enduring image in the Western architectural discipline. However, what the Parthenon builders knew is that our perception is not ‘reality’. In order to make us perceive the building as 'perfect' and 'symmetrical', they had to employ all sorts of geometric acrobatics, such as the entasis (swelling), angling, and altered spacing of the columns, as in the following diagram (see figure 5).

Of course, the Parthenon is valued because it is a site of multiple religious significances, impressive sculptures, embedded social rituals involving the community, worthy and long-lasting materials, and cultural heritage protection. We do not value the Parthenon simply because of how we perceive it from a distance, so why should we take Koolhaas’s argument at face value? De Rotterdam hints at a future of luxury flats (in price but not in any other meaning of the word), designed to be seen from across a bridge, and Elon Musk’s Mars colony goes a step further: it is designed to be seen from a satellite image downloaded to a computer on another planet. Perhaps Archigram were right to focus almost exclusively on architectural renderings: their 1967 Cushicle (see figure 6) treated the interior surface of a small immersive media pod, built around a recliner, as a screen for projected images and a chamber for piped-in music.

That 'rendered' reality might be the best that architecture can offer the 99.9999% in the future. And at least, in Archigram member Michael Webb’s rendering (see figure 7), we’ll have blankets to keep us warm, covered with brand logos to give us something to look at when the power supply shuts down.

(The commission of this essay was generously supported by Molior London)