Mario Carpo

Publication History:

“The rise of the non-standard architectural envelope,” in The Faces of Contemporary Cities, edited by Davide Ponzini, 59-71. New York: Rizzoli New York, 2024

The text posted here is a preprint draft and it is significantly different from the published version. Please only cite from copy in print

 At some point in the early 1920s an ambitious young watchmaker, recently arrived in Paris from his native Swiss Jura (and soon to become famous under the nom de plume Le Corbusier) had an illumination.  The first automobiles were expensive, remarked young Corb, because each car was custom-made and made to order by specialized artisans, like a bespoke suit.  Then Henry Ford started to mass-produce identical car in factories, and made cars cheaper; so cheap that his own assembly-line workers could afford to buy the cars they made.  Hence Corb’s idea: why should we not make buildings in the same way as Ford make cars?  Housing is expensive because buildings are hand-made and made to order, mostly on site.  If we could mass-produce standardized buildings in the modern, industrial way, we would make them cheaper—just like cars became cheaper when they started being factory-made. 

One hundred years later we can tell, with a reasonable amount of confidence, that the argument was sound.  However, when Corb started to actually build some buildings, he soon realized that prefabricating houses in a factory was (as it still is) a very tall order.  Hence Corb looked for, and found, a middle way: while the core of the building, its reinforced concrete structure, had to be made on site in the traditional, artisanal way (never mind that reinforced concrete itself was then a very new material), the most visible part of the building, its envelope, could indeed be factory made.  Factories of the time could in theory already mass-produce glass and steel windows, just as they mass-produced automobile windscreens and doors.  As it happens, Corb’s first curtain walls were almost entirely made by hand, being as they were one-off, small-size experiments; but Corb designed them to look as he thought they would look if they had been mass-produced in huge factories churning out millions of identical pieces a day.

It took some time—probably longer than Corb had envisaged—but after the war, particularly in America, the industrial curtain wall did in fact become an ubiquitous feature of modern architecture.  Mies van der Rohe translated its technical logic into a work of art: even if each component of the facades of the Seagram Building (1954-58) was custom-designed, each was replicated identically thousands of times; the Seagram was, at the time, one of the most expensive buildings ever built, but at the same time a monument to the economics of industrial mass-production, and a celebration of the iron law of mechanical modernity: when things are made in a factory, the more identical copies we make, the cheaper each copy will be.

That was the spirit of what today we call high modernism: that brief but pervasive spell of socio-technical optimism, between the end of World War II and, roughly, the assassination of President Kennedy, when a new consumer culture enthusiastically endorsed industrial standards, and standards were universally seen as the image and icon of modernity and progress, and the promise of better times to come.  That optimism, however, didn’t last.  Already in the 1960s some started to find industrial standards a bit boring (Robert Venturi, famously, in 1966); soon thereafter a new quest for visual variations coalesced around the tenets of architectural post-modernism.  Yet industrial curtain walls kept being made pretty much the same, as not much had changed in the technology underpinning them.  PoMo curtain walls were often deeply tinted or mirror-like, and vilipended by all and sundry for a number of different reasons (including, as of the early 1970s, due to new but rising concerns with energy savings and thermal performance).

Then, unexpectedly, as of the early 1990s, a new technology emerged that revolutionized the way architectural envelopes are made, and the way they look.  Computer-based CAD-CAM has already upended the design and fabrication of almost everything—including the design and fabrication of architectural panels, out of which buildings envelopes are made.  Here a brief technical digression is warranted.  While most mechanical making needs casts, dies, stamps, or molds—mechanical matrixes of which the cost must be amortized by repeated use—digital fabrication (whether by milling, 3D printing, or robotic assembly) is for the most part not matrix-based.  In the absence of a reusable matrix, making more identical replications of the same model will not save either time or labor and conversely, variations in a digital design-to-manufacturing workflow do not entail supplemental costs.  This mode of production is called “digital mass-customization.”  Unlike mechanical mass-production, digital mass-customization does not need to scale up to break even: the unit cost of a digitally fabricated panel is the same no matter how many we make; a factory can make one million identical copies of the same panel, or one million different panels (within the same production chain) at the same cost per panel.  This is what economists call a “flat marginal cost” production, where economies of scale do not apply: in theory architectural envelopes could now be made of parts that are all different, or all the same, at the same cost; repeating the same panel everywhere will make many facades look the same, but will not make any of them cheaper.[1]

In practice, things do not yet really work that way.  For one thing, panels must be assembled after being made, and the assembly of irregular panels is still a labor-intensive and expensive on-site operation, albeit now much facilitated by computational tools.  In the foreseeable future, many of these on-site operations will be carried out by autonomous (AI-driven) robots.  Nobody knows, however, when these technologies will be adopted by the construction industry.[2]

On the other hand, the impulse for the design of more and more convoluted architectural envelopes has emerged over the last thirty years from a range of cultural and formal motivations, which have often driven design intentions above and beyond deliverability.  The rise of digital curviness is an exemplary case in point.  As of the early 1990s a new generation of relatively cheap and user-friendly CAD software allowed the intuitive manipulation of a very special family of complex continuous curves, called “splines.”  Splines are streamlined, smooth curves used in shipbuilding to minimize the drag resulting from the movement of a boat’s hull in water (hence the etymology of “streamlining,” from “the line of the stream”).  Artisan boat makers obtained splines through the mechanical bending and nailing of slats of wood; in the late 1950s and early 60s two French scientists, Pierre Bézier, an engineer by training, and mathematician Paul de Casteljau, working for the Renault and Citroën carmakers, respectively, discovered various methods for notating splines as mathematical functions, thus turning that centuries-old craft into a modern science: Bézier’s maths in particular, published in 1966, was the basis for UNISURF, the spline-modeling CAD software developed by Bézier’s employer, Renault (as of 1968). UNISURF,in turn, was the basis of the CAD-CAM system developed internally by the Dassault aircraft company (then called Avions Marcel Dassault-Breguet Aviation) from 1977 to 1981, when it was renamed CATIA to be marketed by the newly created subsidiary Dassault Systèmes.  At the same time as de Casteljau’s and Bézier’s studies on free-form curves, similar research was carried out at MIT, Boeing, at the British Aircraft Corporation, and particularly by Carl de Boor at General Motors, which developed its own CAD/CAM system in the 1960s; in 1981 Boeing was among the first adopters (or perhaps the inventor) of NURBS, an acronym for Non Uniform, Rational Basis Splines, a new graphic format (and an international standard to this day) which merged Bézier’s and de Casteljau’s maths with earlier studies by Princeton mathematician Isaac Jacob Schoenberg (1946).[3]

Bézier’s, de Casteljeau’s, and Schoenberg’s mathematical notations, when implemented by electronic computers, enormously simplified the design and fabrication of aerodynamic, or streamlined shapes, and it is easy to see why this technology, discreetly but massively adopted by the the aircraft, automobile, and shipbuilding industry as of the early 1980s, revolutionized the production of moving vehicles—cars, boats, and airplanes.  Why would digital streamlining, or spline-modeling, become such a vital concern for architectural designers—as of the early 1990s, and to some extent to this day—is less immediately evident.

One frequently cited reason for this apparently baffling techno-cultural development was the overwhelming influence—particularly among young American designers in the 1990s—of the French philosopher Gilles Deleuze, whose book The Fold: Leibniz, and the Baroque (first published in French in 1988; translated into English in 1992) suggested a relationship between computational design, parametricism, the history of differential calculus, and the curvy shapes (or “folds”) that characterized Baroque painting, sculpture, and architecture.  In 1996 Greg Lynn introduced the term “blob” to define the new wave of digitally modeled “Deleuzian folds,” and computational splines soon came to the seen as the most conspicuous stylistic feature of the new architecture of the digital age.

A less well known, but equally relevant, part of this story had to do with fish.  Frank Gehry’s long-standing interest in fish, carps in particular, is well documented.  Gehry’s first rise to international stardom came in the aftermath of his participation in the seminal Presence of the Past Venice Biennale (1980)—the launching pad for international post-modernism.  Yet, as Jean-Louis Cohen recently pointed out, Gehry—never a fully-fledged PoMo—soon came to resent his accidental association with some of the historicist and revivalist topics then promoted by post-modernists; and, as an expression of his visceral reaction against the anthropomorphic obsession of classical architecture, he started to toy with the image of fish—after all, a much older and ancestral reference than Greek and Roman buildings, and older, in purely zoological terms, than the human body itself.[4]  The big metal fish he built on Barcelona’s beachfront in 1992 was not his first (nor his last) but it was the first where Gehry experimented with Dassault’s noted CAD-CAM software, CATIA. 

The connection between CATIA and fish may appear on first impression incongruous, but it follows from a perfectly rational technical argument.  Fish bodies are naturally streamlined so that fish may more easily move in water.  The streamlining of fish happened over time by dint, we are told, of evolutionary adaptation.  But, having to build a big (56 x 35 meters) metal fish hovering high over Barcelona’s marina, Gehry observed that fish move in water in the same ways as the hull of a boat does, and concluded that the software now used to design the spliny hulls of boats should be able to emulate the natural splinliness of fish.  In the end, Gehry did not find the software he needed from shipbuilders but from an aircraft maker, and the rest is history: CATIA-based spliny curves became the hallmark of Gehry’s style, throughout his multifarious architectural production, and almost to this day; after the resounding success of his Guggenheim Bilbao (designed 1991-94; inaugurated 1997) the technological expertise acquired by Gehry’s architectural office was spun off to an independent consulting company, Gehry Technologies (created in 2002, sold in 2014), which released in 2004 a version of CATIA specifically tailored for architectural design, and became a BIM service provider to architectural firms around the world.  If the story of Gehry’s relation with Dassault is known, at least anecdotally, and it has been told many times, nothing is known of the story of Gehry’s relationship with the Italian company Permasteelisa, specializing in the project delivery of special architectural envelopes.  Permasteelisa started collaborating with Gehry for the Barcelona Fish and has built, to date, 10 of Gehry’s best known masterpieces (as well as hundreds of other very complex envelopes, curtain walls and facades, including many designed by well-known stararchitects around the world).

The invention of the non-standard, post-industrial architectural envelope is then due to the unlikely encounter of a Canadian born, Los Angeles educated designer who managed to be at the same time anti-modern and anti-classical, with a special interest in fish; of a Paris-based aircraft maker leveraging the great mathematical tradition of French polytechnical engineering; and of an Italian fabricator that could adapt its delivery to the intention of the former, and to the technology of the latter.  Gehry’s panels are famously not only all curved, but also all different from one another; the “folding” draperies in the envelope of Gehry’s Eight Spruce Street residential high-rise in New York City (2003-2011), for example, are made of around 11,000 metal panels, but each panel has been repeated, on average, no more than 5 times,[5] and Permasteelisa’s recent catalogues and technical literature pertinently highlight the firm’s expertise in the delivery of envelopes where most components are singular one-offs.

This is indeed what the digital pioneers of the early 1990s had envisaged and predicted: the non-standard envelope of today’s computer-driven global architecture is the symbol of the post-industrial logic of digital mass-customization—just like, two or three generations ago, the standardized curtain wall of high modernism symbolized the technical logic of mechanical mass-production.  And just like industrial mass-production always was inherently unsustainable, digital mass-customization is, at least in theory, inherently sustainable: digital fabrication does not need scaling up to centralize manufacturing in remote, low-cost production hubs; on the contrary, the potential of digital manufacturing can be better exploited by distributed, neo-artisanal micro-factories, having access to locally sourced materials, energy, and labor.  Stay tuned: the best days of the post-industrial, non-standard architectural envelope may still be ahead of us.        

[1] Carpo, Beyond Digital (Cambridge, MA: The MIT Press, 2023), 6-16.

[2] Alejandro Zaera-Polo and Jeffery S. Anderson, The Ecologies of the Building Envelope (Barcelona and New York: Actar, 2020), 262, and passim.

[3] Carpo, The Second Digital Turn (Cambridge, MA: The MIT Press, 2017), 58-64 .

[4] Jean-Louis Cohen, Frank Gehry. The Masterpieces (Paris: Cahiers d’Art-Flammation, 2021), 133-34; Alejandro Zaera Polo, “Conversations with Frank O. Gehry,” in Fernando Marquez Cecilia, Richard C. Levene, eds., Frank O. Gehry 1987-2003 (Madrid: El Croquis, 2006), 39.

[5] Permasteelisa Group, Architectural Envelopes 2015, 314.  The numbers cited by the company likely exclude the smaller, standard panels used for the flat cladding extending all over the south-west side of the building.