“It may seem anachronistic to bring up the word craft today, especially within the context of contemporary technology and its related modes of production, but there are striking parallels between this medieval concept and digitally driven architectural practice. . . .”
*image: Ramiro DIAZ-GRANADOS / Amorphis, “Go Figure.”
It may seem anachronistic to bring up the word craft today, especially within the context of contemporary technology and its related modes of production, but there are striking parallels between this medieval concept and digitally driven architectural practice. Today the notion of digital craft is in the air, as levels of expertise in the manipulation of computational geometry and matter increase and the gaps between design and fabrication decrease. The word itself—craft—conjures up imagery akin to whittling, where a lone artisan sits with his knife and a raw piece of wood, and carves it into some quaint artifact. However, this conception is a bit superficial. Etymologically associated with the Greek word techne, the word craft historically denotes activities that lie somewhere between art (talent and technique) and science (knowledge). Economically craft is applied to the small-scale production of customized goods; socially it is an autographic endeavor, where designer and maker are one and the same person; and technologically it is associated with hand-related tools and methods that produce variability (albeit, a quality undesired at times). Gottfried Semper defended the craft traditions in his book The Four Elements of Architecture by tying hand craft to “man’s dignity.” But for all of the energy exerted by the defenders of arts and crafts, they could not contend with the sublime scale of production of machines during the industrial revolution and modernity.
This essay drafts three distinct but related dichotomies around the concept of digital craft: autography versus allography, matter versus geometry, and manual operations versus automated simulation. While a medieval notion of craft privileged the former aspects of these dichotomies, I am suggesting that digital craft is a toggling between each, a more topological relationship than a binary one. The aim is to provide a set of working points for the contemporary practitioner. At the very least, what follows will hopefully provide something to think about the next time one sits in front of their screen and delves into the world of their software(s) of choice.
Autographic vs. Allographic: From Brunelleschi to Alberti
In a 2012 lecture at the Southern California Institute of Architecture (SCI-Arc), Peter Eisenman grumped about the lack of authorship being retained in the construction of his Pinerba Condominium project in Milan, Italy. The gist of the story was that the contractor made it clear that he was not going to maintain a high degree of fidelity to Eisenman’s design and construction drawings. One can easily imagine the frustration that would set in. But this anecdote raises the longstanding distinction between the autographic and allographic paradigms, or what Mario Carpo describes as, “The transition from [Filippo] Brunelleschi’s artisanal authorship (‘this building is mine because I made it’) to [Leon Battista] Alberti’s intellectual authorship (‘this building is mine because I designed it’).”1 Clearly Eisenman sides with Alberti and expects a high level of notational identicality to be satisfied. The more drift that occurs in the contractor’s interpretation of the architect’s designs, the less authorship the latter has (or at least feels he has). A more explicit definition is provided by the philosopher Nelson Goodman, who writes, “Let us speak of a work of art as autographic if and only if the distinction between original and forgery of it is significant; or better, if and only if even the most exact duplication of it does not thereby count as genuine.”2 The irony is that architecture and the business of building have, since the Renaissance, primarily occupied the allographic paradigm, yet we continuously struggle to maintain the highest level of authorship of our works. The notion of digital craft seeks to quell this dilemma by circumventing the need for notational representation, thus eliminating the potential drift of translation. A fully crafted digital model can thus be transferred from designer to builder as is, though, of course, this demands that the architect highly resolve all logistical demands and economic constraints (now possible through building information modeling [BIM] software and the algorithm), and that builders acquire the skills to navigate and work with 3-D software.
A short-term (and unsustainable) way around this problem has manifested itself over the last decade in the works of certain young architects and designers, largely through installations and pavilions. This type of architecture has provided a venue for material and tectonic experimentation of digitally produced forms. However, because of the economic constraints and intensive labor demands, installation and pavilion work has resulted in a privileging of the Brunelleschi model (which some do by choice and others out of necessity). Designers, and a small army of unpaid enthusiasts, are thusly forced to build their own designs. Craft, then, becomes a quite literal endeavor, designers often resorting to mixing advanced fabrication with traditional techniques of material assembly in an attempt to compensate for cost limitations. This actually has its benefits, as it introduces a sense of pragmatism to the problem and allows the designer to develop a sensual intelligence that can be folded back into the design process in the future. It also conjures up the specter of William Morris, the 19th-century English textile designer who, after years of supporting the Ruskinian philosophy of wholesale rejection of industrial manufacturing and a return to handcraftsmanship, found a way to tactically combine the two.
Those who carve a niche in this genre don’t have to worry about scaling up and can freely evolve their craft techniques, but are ultimately faced with the pressure of the new. As installation and pavilion work is temporary by nature and so ubiquitous, its aesthetic and cultural effects wane quickly. It is a cultural practice that demands the rate of change of fashion and music, and yet is still tied to that of architecture, which is closer to the pace of geology. Those who use it as a stepping-stone toward larger (and perhaps more legitimate) projects face the problem of scale. With an increase in scale comes a whole host of problems related to building conventions, budget, and legal restrictions (codes). And since none of these are getting any simpler, the gap continues to widen.
In a prophetic paragraph in his 1989 essay “The Death of Drawing,” architect William J. Mitchell posits:
We have not quite reached the end of this story. Usually, in practice today, the end product of a computer-aided design process is a set of drawings plotted from a database. Indeed many architects shortsightedly think of computer-aided design as essentially a technique for fast drawing production. But there is little reason to doubt that architects will soon adopt the practice that is now commonplace in manufacturing industry and deliver, instead, databases in machine-readable format. The contractor can then process this to produce plots and images as required, use it for input to cost estimation and construction planning and management software, and even use it directly to program CAD/CAM systems and construction robots. When this eventuates the eclipse of the drawing will be complete.3
Matter vs. Geometry: From Intensive to Extensive
Since the Renaissance, the relationship between matter and geometry in architecture has been far from contingent. Until recently, geometry has had the upper hand, disciplining matter in its own image. With the allographic paradigm mentioned above, geometry and its related notations necessarily become the primary means to deliver the principles of identicality. Furthermore, geometry, at least in its Euclidean and pretopological incarnations, abstracts the material world in order to gain control over it and operates within a conceptual framework. For example, the nine-square grid problem initiated by John Hedjuk and the Texas Rangers is a classic exercise that privileges Euclidean geometry in a Cartesian coordinate system. As points become columns and lines become walls, it is still geometry (and proportion) that determines thickness and overall organization. Matter waits for geometry’s instructions. Points, lines, planes, and volumes occupy static and extensive positions in space (as void). Manuel DeLanda, an astute materialist, convincingly makes the case for the active role of matter in the production of architectural form and for the value of the craftsman for contemporary discourse.4 This inversion is not merely a reaction to the previous paradigm, but one that aligns itself with the advent of computation and developments in allied fields such as science, technology, philosophy, mathematics, and art.
The introduction of topology, calculus, and simulated physics to architectural form has completely unnerved the discipline’s sense of control over geometric organizations. In this model points, lines, planes, and volumes are exchanged for vectors, curves, surfaces, and mass. The extensive differences of the former are confronted by the intensive ones of the latter.5 Rather than the disciplining of matter through geometry in a top-down hierarchy, intensive differences are managed and form is coaxed into resolute expressions. An analogy might be useful here. Leonardo Da Vinci’s famous Vitruvian Man drawing, for example, is a clear depiction of extensive control over a body. Here, a geometric system is projected onto the body, producing an essential figure. Geometry reorganizes the body into an ideal image of beauty, proportion, and harmony. Butchery diagrams, on the other hand, coax geometry from matter, in this case flesh. In locating and drawing the prime cuts of an animal, the butcher identifies the general boundary of meat according to intensive (muscle) properties such as fat content (leanness), flavor profile, density (of muscle), and so on. The resultant diagram is different for each species and serves as a loose cartoon for the more visceral complexities of matter. This is only possible with, and directly tied to, the level of craftsmanship of the butcher. The geometric complexity of the diagram is dependent on the level of the butcher’s knowledge of his animal (material) and on his skill with knives (tools).
For more than 5,000 years man has been engaged in basic crafts (agriculture, medicine, metalworking, weaving, dyeing, perfumery, and glassmaking being among the oldest). Historically, the practical craftsman has occupied a natural opposition to his scientific counterpart. If we understand geometry as a scientific and theoretical endeavor, then philosopher Stephen Toulmin and historian June Goodfield are accurate in their assertion that,
Science has again and again been in the position of debtor, drawing on the craft tradition and profiting from its experience rather than teaching craftsmen anything new. It has been said that “science owes more to the steam-engine than the steam-engine owes to science”, and the same thing is true more generally. In its early stages, especially, the craft tradition was—so far as we can tell—devoid of anything which we would recognize as scientific speculation.6
A notion of digital craft posits that matter obtains at least equal status to geometry and that computational geometry is “live”—that is, laden with heterogeneous material behaviors that require the coaxing of form rather than its imposition. Working exclusively with default geometries in digital space (like the Voronoi diagram) has resulted in homogeneity across the discipline and produced clichés quickly. A return to matter through digital production is really about putting sensual (manual) intelligence on the same plateau as rational (cerebral-cognitive) intelligence. It cannot simply be an inversion of the dominant paradigm or a wholesale return to medieval practices. The potential feedback loops within a matter-geometry complex require craftsmanlike dexterity with scientific precision and projection (of geometry, force, and thought).
Manual Operations vs. Automated Simulations: From Direct to Mediated Control
The rise of scripting and parametric software in architectural design and production has certainly provided unprecedented power over the control of vast amounts of geometric and tectonic information. To be clear, architecture has always been parametric (if understood as the establishing of limits or boundaries in relation to one another), but the seductive power of its computational form allows local variations to be tracked and managed globally in an automated instant. This is where a set of rules is respected. Parametric modeling clearly has its benefits in the optimization sectors, where all of a project’s geometric, tectonic, and performative systems are linked to each other, as well as to cost-control spreadsheets through BIM software packages. But this is perhaps more useful for the back end of production. For the front end, during the earlier stages of the design process, scripting is perhaps more generative. This is where a set of rules is written. These models provide for mediated control through simulation. Parametric modeling simulates the physical behaviors of change while scripting simulates the physical behaviors of growth. Both are useful and productive, but have their shortcomings.
One of the draws of scripting is that it is, “An efficient way to produce differentiated repetition in digital modeling that would otherwise require a great deal of time and effort. At its essence it is a method for reducing the number of keystrokes required to model, alter, and then repeat a particular form.”7 This is also its danger. In being able to generate large amounts of geometry quite easily, scripted organizations tend to quickly dominate a scheme and become legible as such. Scripting produces a totalizing and hegemonic effect where the overall imagery appears complex, but local conditions are often simplistic at best. While it is possible to write multiple scripts for all scales and systems of a project, it would require an extraordinary amount of time and likely impede on the nonlinear flow of the design process. Scripting seems most successful when applied selectively in a hierarchical design chain.
Manual operations, on the other hand, define a more sculptural approach, where direct transformations on control geometries take place. A simple example is in connecting two different geometric systems by moving, one at a time, the control vertices (CVs) of one surface over to another. Working this way requires the initial definition of parameters or constraints as well as a loosely defined target that is being moved toward. While this method provides for local control, things can get quite messy without a good (mouse) hand. To work primarily in this way one needs to be a virtuoso. The problem is that the interfaces of 3-D modeling software are becoming smoother, requiring less discipline and rigor in generating superficially complex form. With a few tricks, the software can make anyone feel like a virtuoso, when in fact they are quite rare.
Digital craft requires the simultaneous distribution of manual operations and automated simulations; which one to start with does not really matter. At some point the designer must get dirty with scripted geometry by directly and manually taking control, or, conversely, must discipline their sculpted geometry by mediating it through automation.
Whether or not we are in the midst of a medieval comeback, it seems that a productive tension exists between the worlds of the mouse and the hammer. Just as we are coming to terms with an emerging model of nature as computational, we must also come to terms with computation as material. Far from being a nostalgic call for the golden age of medieval craft practices, this is one of merging the sensual intelligence of direct making with the mathematical intelligence of mediated simulation. As contemporary building problems related to performance, economy, assembly, regulations, and so on become increasingly complicated, a digital craft design ethos promises to maintain the vitality of speculative discourse while the building industry at large, including the manufacturing of materials, catches up to this model of architectural production.
1. Mario Carpo, The Alphabet and the Algorithm (Cambridge, MA: MIT Press, 2011). ↩
2. Nelson Goodman, Languages of Art, An Approach to a Theory of Symbols, 2nd ed. (Indianapolis: Hackett, 1976). ↩
3. William J. Mitchell, “The Death of Drawing,” UCLA Architecture Journal 2 (1989): 64–69. ↩
4. Manuel De Landa, “Philosophies of Design: The Case of Modeling Software” in Jaime Salazar et al., eds., Verb Processing: Architecture Boogazine (Barcelona: Actar, 2001), 130–43. ↩
5. For a clear explanation of “extensive” and “intensive” differences see For a clear explanation of “extensive” and “intensive” differences see: Jesse Reiser and Nanako Umemoto, Atlas of Novel Tectonics (New York, NY: Princeton Architectural Press, 2006), 104. ↩
6. Stephen Toulmin and June Goodfield, The Architecture of Matter (Chicago and London: The University of Chicago Press, 1962). ↩
7. Jason Payne, “A conversation between Sanford Kwinter and Jason Payne,” in Tomoko Sakamoto and Albert Ferré, eds., From Control To Design: Parametric/Algorithmic Architecture (Barcelona: Actar, 2008), 219–39. ↩