Automated Architecture; Speculative Reason in the Age of the Algorithm

Luciana Parisi/Texts/Essays/Automated Architecture; Speculative Reason in the Age of the Algorithm.pdf

P. 1
Automated Architecture: Speculative Reason in the Age ofthe Algorithm Luciana Parisi Early nineties cybernetic culture did not fully anticipate the emergence of the automated form of reason that now subtends the infrastructural architecture of everyday life. After more than twenty years, it has become apparent that what Sanford Kwinter calls the "age of the algorithm"' points to a new form of ratio­ nality not only leading to new methods of calculating economic risk, or performing simulations of social behavior, but also involving the computation of space and time. T his new form of rationality, defined by the convergence of computation and topology, or information theory and nonstandard geometry, has produced a specific aesthetic that can be found in algorithmic architecture. In this field, which includes the pioneering works of digital architect Greg Lynn, 2 among others, computational design has taken inspiration from vector fields so as to model, for example, the speed and direction of a fluid moving through space, or the strength and direction of a magnetic or gravitational force as it changes from point to point. Here the architectural form is the result of the computational processing of bio­ physical variables (the distribution of weight, gravitational pressures, circulation of air, intensity of traffic, frequency of movement). 3 Influenced by second-order cybernetics of evolving feedbacks, algorithmic architecture of the 1990s started to adopt biophysical dynamics as inputs in software programs, and biophysical data became a new parameter for computational design. By closing the gap between mathematical models and biophysical contingencies, computation has entered a field of spatiotemporal connected­ ness-a topology-involving a continual transformation of form without cutting or tearing. Topology, as the geometry of place, here refers to the mathematical 2 3 Sanford Kwinter, Far From Equilibrium: Essays on Technology and Design Culture (New York: Actar, 2008). Greg Lynn, "Architectural Curvilinearity: The Folded, the Pliant and theSupple," in "Folding in Architecture," ed. Greg Lynn, special issue, Architectural Design 63, nos. 3/4 (March/April 1993): 22-29. According to architect Karl Chu, algorithms have been central to the late twentieth-century convergence of computational and biogenetic revolutions leading to the ultimate design of biological and mathematical codes, which promise the embodiment of life, emotion, and intelligence. See Karl Chu, "Metaphysics ofGenetic Architecture and Computation," in "Programming Cultures: Art and Architectures in the Age ofSoftware," ed. MikeSilver, special issue, Architectural Design 76, no. 4 (July/August 2006): 39. This position is not too different from debate about cybernetics in the late nineties that directly engaged the new techno-scientific ontologies of biological bodies. Speculation
P. 2
AUTOMATED ARCHITECTURE understanding of geometrical shapes and spaces, in which connectedness and continuity are preserved under their continuous deformations, such as stretching and bending. lt also refers to the geometrical understanding of form in terms of its evolution in time and thus to the genetic capacities of a form to change through time. From this standpoint, biophysical unpredictability has become superior to mathematical axioms, as the reality of abstraction has slipped behind the concrete­ ness of matter. In other words, the materiality of data and their variations in time have replaced mathematical models in their attempt to represent reality through abstract rules. The digital design of space has superseded the Euclidean matrix of extension by adding temporal evolution to fixed points, developing a topological surface enveloping all points in space and time. The figure of the blob or the fold corresponds to the continuous topological surfaces now defining the dominant aesthetic of digital architecture. In partic­ ular, as Patrik Schumacher has argued, parametricism now represents the new global style for architecture and design. 4 When applied to large-scale urbanism, for instance, parametricism is said to transform the differential distance between points into an integral surface of continual connection. 5 From this standpoint, parametricism implies the inclusion of contingent elements-atmospheric, geological, biological, and physical elements-into the program in the form of variable parameters whose temporal evolution becomes the motor for the trans­ formation of the architectural whole. This means that variables are not only added to the program (as if from the "outside"), but also partake of the software environ­ ment of parametric relations. Parametric programming is not just concerned with the computation of existing elements, but additionally, and significantly, with how feedback relations between finite parameters can engender the infinite variations of an architectural form. Nevertheless, as a style of contemporary digital architecture, parametricism has been particularly discussed as a manifestation of the "cultural logic of late neo­ liberalism," whose topological operations of continual transformation, structural coupling, and mutual correspondence between the inside and the outside define the 4 5 Schumacher recently claimed timt parametricism is the dominant style oftoday's avant-garde, characterizing the power oflarge-scale urban schemes. See Patrik Schumacher, "Parametricism: A New Global Style for Architecture and Urban Design," in Digital Cities AD: Architectural Design, ed. Neil Leach (London: John Wiley and Sons, 2009), 14-24. In computer programming, a parameter is a variable: a symbolic name given to a known or unknown quantity ofinformation, so that the name can be used independently ofthe information it represents and can then be assigned different values in different places. A parameter, therefore, can be used in a new way or in the same way (iteration). Parameters are used in subroutines (a procedure, function, routine, method, and so forth) to refer to one ofthe pieces ofdata provided as inputs to the subroutine. In contrast to standard software packages based on datum geometric objects, within digital architecture parametric software links dimensions and parameters to geometry, thereby allowing for impact ofthe incremental adjustment ofa part on the whole assembly. Speculation
P. 3
LUCIANA PARIS! choreographic arrangement ofdata. 6 Yet this critical view ofparametricism is pred­ icated on the fact that parametricism promises a formally open-ended and flexible space that does not physically match realized architectures, and that it is the direct incarnation of the spirit of the neoliberal market. I propose that parametricism is not abstract enough to meet the possibilities offered by a radical algorithmic for­ malism. If parametricism only relies on the operative function of algorithms and the interactive programming ofparameters without questioning what these material forms of abstraction can do beyond the intentional design of the program, then it remains unable to engage with the unthought-of capacities of these automated systems to determine new spatiotemporal forms. Without denying that parametricism is an instance of the topological aes­ thetic ofgovernance, I want to problematize the rejection tout court ofthe agential character ofcomputer programming and its real algorithmic objects.7 That is, I take algorithmic objects to be necessarily implicated in the sociality that they invisibly structure. The stealthy intrusion ofcomputational programming into everyday cul­ ture requires a close engagement with the nuances ofthe digital apparatus as well as the axiomatic thought-that is, with logic and the formalism that subtends it-that indirectly shapes culture. From this standpoint, the new topological architecture of relations expressed by parametricism is precisely what needs to be challenged in order to reveal the irreversible transformation that algorithmic objects have brought to digital formalism. In other words, I suggest that it is now crucial to readdress emerging types of formalism, parametricism in this case, that characterize the aes­ thetic ofneoliberal capitalism (specifically its having subsumed all culture and life to its own interactive and changing rules). However, instead ofrejecting formalism, I propose that one has to revisit its significance in the light ofradical changes within its computational modus operandi, which may offer aesthetic and cultural possibil­ ities inconsistent with the capitalist axiomatic ofa whole that subtends all its parts. In particular, I argue that these computational systems are populated with what 6 7 Parametricism has been criticized for many reasons, but I will briefly mention only two here. On the one hand, typified by Owen Hatherley's criticism, parametricism is decried as an apolitical self-proclaimed avant-garde. Referring to Schumacher's text "A Glimpse Back into the Future," which accompanied the exhibition "Zaha Hadid and Suprematism" at the Galerie Gmurzynska in Zurich, Hatherley argues tl1at parametricism strips Russian avant-garde architecture ofany social or political dimension denuding it to its bare formal radicality. See Owen Hatherley "Zaha Hadid Architects and the Neoliberal Avant-Garde," Mute, October 26, 2010, www.metamute.org/ editorial/articles/zaha-hadid-architects-and-neoliberal-avant-garde. On the other hand, Schumacher's parametricism has also been accused ofdisengaging from the physical ground of architecture and overlooking the contingencies ofurban planning through an excessive search for formal relations. In particular, it has been argued that the excessive search for the beauty ofform has completely derailed digital design from becoming useful for urban and infrastructural problems. See Ingeborg M. Rocker, "Apropos Parametricism: If, In W hat Style Should We Build?" Log 21 (March 2011). On the problematics of nonphysical design of space associated with a Deleuzian reading of architecture, see Douglas Spencer, "Architectural Deleuzism: Neoliberal Space, Control and the 'Univer-city'," Radica/ Philosophy 168 (July/August 2011). Speculation
P. 4
AUTOMATED ARCHITECTURE Alfred North Whitehead calls "eternal objects," which I conceive as incomputable entities that possess an infinite capacity for the eventuation of change and for rad­ ical contingency. 8 My aim here is to show, via Whitehead, that parametricism, as a particular type of formalism, may offer, despite its ostensible homologies with neoliberal capitalism, a non-intentional degree of "interference" that breaks open from within the axiomatic wholeness of neoliberal capital. A close analysis of this transformation can help explain how structural changes in programming are not only not negligible but are in fact ontological expressions of computational culture and power. This analysis highlights the incongruence-the asymmetry and non­ equivalence-between software as a totalizing system of governance and algorithms as a series of fractal or inconsistent events. One could argue that parametricism immediately transforms formalism because it incorporates contingencies into its model by using real time data. Yet introducing temporality into programming does not fully challenge the formal nature of the enterprise; it only affords formalism the pretension of describing how mathematics can incorporate physics by creating a totalitarian system of relations, according to which a few mathematical mies are able to lead to the evolution of com­ plex structures by establishing continual feedback with the environment. This means that parametricism's topological qualities do not in fact challenge formalism: instead, insofar as the computation of biophysical contingencies only serves to construct the program as being open to change, they are but the reification of its formal logic. The introduction of qualitative and real-time variations in algorithmic pro­ gramming aims at anticipating spatiotemporal changes, which means designing the potential transformation of scenarios. In other words, software interactions with the real data of the environment have become constitutive of the aesthetic of topological governance based on the effects that potential data can have in its internal architecture of relations. Instead of simply reducing biophysical variables and contingencies to sets of binary codes, which are unable to process the gray areas between states, digital design now uses the topological integration of differential relations, or intensive data, to build spatiotemporal connections and anticipate incipient changes. The use of evolving algorithms or open-ended instructions that respond and adapt to the external environment also reveals that governance now relies on the calculation of differentials, or the variable qualities or infinitesimals found in the relation between states. Speculative Computing lt is important to specify that interactive algorithms and responsive computation occlude, rather than reveal, the what and the how of algorithmic objects-objects deemed to remain passive in the face of an ever changing governance of continual 8 See Alfred North Whitehead, Process and Reality (New York: Free Press, 1978), 22ff. 366 Speculation
P. 5
LUCIANA PARIS! connectedness. We turn therefore to information theory to address algorithms as sequential spatiotemporal data structures conditioned by infinite amounts of infor­ mation. I believe that these data structures are actual spatiotemporalities and are precisely the objects of algorithmic architecture. This means that an algorithmic object is more than a temporal form or the result of interactive inputs, that it defines spatiotemporal structures as stemming from the increasing amount of automated data exploding in our computational culture. 9 In order to understand the status of algorithms in digital architecture, I suggest that algorithms need to be considered in Whitehead's terms as actualities: spatiotemporal data objects.' 0 Algorithms are not simply instructions that have to be executed by a program, but are actualities spatiotemporally constrained by the data of which they are composed. This spatio­ temporal conception of algorithms helps us to rethink algorithmic architecture as the performative vehicle of the spatiotemporalities of data structures constituting algorithms. In other words, in order to define the aesthetics of algorithmic archi­ tecture and the formation of new spatiotemporalities, one must first confront the spatiotemporal matrix of algorithms. Contrary to the view that computation is a reductive form of rationalism, I want to rethink computation in terms of speculative reason, to borrow another term from Whitehead." Computation as an instance of speculative reason does not correspond to the topological order of potential connectedness between points. On the contrary, Whitehead's understanding of speculative reason explains that the function of reason is to add new data to the continual chain of cause and effect-here, the continual processing of data. Similarly, a speculative view of computation implies that such pro­ cessing is not a mere compression of data or a structure of relation defined by sets and subsets. Instead, a speculative understanding of computation implies that each set and subset of instruction is conditioned by what cannot be calculated, the incomputable 9 This is an important argument against the idea that algorithms are merely temporal forms destined to disappear in the background ofubiquitous computation. Computational design problematically embraces the logic ofprediction and the calculation ofprobabilities, which are already set data and are unable to explain novelty in spatiotemporal experience. 1 argue against this form ofmeta-computation and against rational logic that is based on few unchangeable rules, the combination ofwhich produces all forms of complexity. 10 For Whitehead, the actual world is composed ofactual occasions, or actualities. These are grouped in events and become the nexus ofactual entities that are "inter-related in some determined fashion in one extensive quantum." Events therefore explain the togetherness of actualities, which Whitehead calls the "nexus." But every nexus is a component part ofanother nexus that emerges as an unalterable entity from the concrescence ofits component elements, and stands as a fact, possessed ofa date and a location. See Whitehead, Process and Reality, 109, 230. 11 According to Whitehead, "the history ofmodern civilization shows that such schemes fulfil the promise ofthe dream ofSolomon. They first amplify life by satisfying the peculiar claim ofthe speculative Reason, which is understanding for its own sake. Secondly, they represent the capital ofideas which each age holds in trust for its successors. The ultimate moral claim that civilization lays upon its possessors is that they transmit, and add to, this reserve of potential development by which it has profited." Alfred North Whitehead, The Function of Reason (Boston: Beacon Press, 1929), 72. Speculation
P. 6
AUTOMATED ARCHITECTURE data that disclose the holes and gaps within (and not outside) the formal order of sequences. Algorithms are conditioned by incomputable data because that which cannot be calculated in advance is the internal limit of any computational process, manifested by the incapacity ofa program to include all possible answers in its premises or by the failure offormula to determine all ofits expressions. This means that a notion ofspeculative computing is not concerned with quantifying probabilities to predict the future, but rather with a concrete system ofalgorithmic objects defined by randomness or incompressible quantities ofdata. Thus, a notion ofspeculative computing is not to be confused with the capacity ofalgorithmic architecture to create temporary forms that simulate what spatiotemporal structures and infrastructures could become. On the contrary, I advance the notion of speculative computing to suggest that random data-which in information theory means non-compressible data-are the contagious architectures ofthe present. These data architectures, far from withdrawing from the present as temporal forms that appear and disappear, are instead actual objects that, even when they cease to be present or function, nonetheless remain objective data to be inherited, evaluated, and selected by subsequent algorithms. I have so far argued that the scope of algorithmic architecture cannot be confined to the specific forms ofbuilt or designed environments. Algorithmic architecture inevitably unleashes the question of what and how algorithms are, thus exposing the invisible dynamics of spatiotemporal data structures that subtend and drive the increasing use ofalgorithms to design environments. lt is therefore possible to suggest that algorithmic architecture reveals the imma­ nent reality ofpatternless data, exposing the inconsistent unity of algorithmic objects and thus determining a fractal and not topological arrangement ofspatiotemporalities in our computational culture. What is most important here is that this reality cannot be encompassed by the totalizing and invariant function of a topological model of continuous surfaces. From this standpoint, algorithmic architecture does not just reveal the neoliberal aesthetics of topological relationality; rather, this newly auto­ mated aesthetics consists of the abstract reality ofalgorithmic events emerging from the automated selection and production ofincomputable data: random information. Like Whitehead, I contend that the aim ofspeculative reason is the production of an abstract scheme, the concrete arrangement of relations. 12 In other words, for reason to be truly speculative, the schemes that are produced and realized must be able to encounter their finitude and limits: to account for the incomputable, the dis­ crete parts that cannot be incorporated into the continual evolution ofthe whole.'3 12 13 "The true activity ofunderstanding consists in a voyage to abstraction which is in fact a voyage to the more fully concrete: to the system in which the fact is enmeshed. The system as conceptualized may be more abstract than the fact itselfin that it is more general, but the real systematic context is more concrete, and its elaboration yields more about the existential relations ofthe fact." lbid., 76. "Abstract speculation has been the salvation ofthe world-speculations which made systems and then transcended them, speculations which ventured to the furthest limits ofabstraction." lbid. 368 Speculation
P. 7
LUCIANA PARIS! What can be learned from understanding computation in terms of speculative reason is that the contemporary form of governance ventures toward the actuality of incomputable probabilities (infinite discrete unities that are bigger than the totality of the whole sequence of algorithmic instructions) that lie behind the digital ground of universal computation and the continual surface of topological connectedness. The age of the algorithmic governance precisely marks the moment at which the unintended consequences of ubiquitous computing unleash the limit of data pro­ gramming (the limit of computation) into the everyday, whereby new spatiotem­ poral structures that cannot be assimilated into what we know or perceive take over. lt would be wrong, however, to view these incomputable structures of data with naive enthusiasm and thus rest in them our hope of breaking open the topo­ logical continuity of governance. Instead, it is important to address the reality of algorithmic objects without overlooking the fact that the computation of infinity is at the core of a new logic, rationality, and ultimately governance. This concern with incomputable probabilities is therefore ultimately a concern with the transfor­ mation of the formal logic of computation. This is not to be confused with a call for an underpinning mathematical or digital ontology or politics. Instead, I prefer to follow Gregory Chaitin's discovery of Omega, an incomputable probability that shakes the mathematical or digital ground of truth by revealing that the probability for infinity is an algorithmic affair, exposing the limit of reason and perhaps the irreducible presence of a non-human thought.' 4 This also means that what subtends the algorithmic architecture of contemporary culture is the computational process of calculating that which cannot be predicted in advance, and that the computa­ tional idealism of efficiency is in truth internally challenged by its limit. However, this is not simply to say that the incomputable always already escapes computational formalism because it remains ungraspable and fully dependent on the physicality of experience. Instead, I suggest that the incomputable is nonetheless apprehended in the last instance by a computational process whose internal limit has become the condition for any form of calculation. What enters the cultural field therefore is not the failure of computation to incorporate the incomputable, but the capacity of the incomputable-the uncompressible and entropic expansion of data-to drive what can be computed. To conclude, I want to emphasize the argument that algorithms expose pre­ cisely the inevitable reality of automated modes of thought through automated spatiotemporal data structures that cannot be subsumed under a totalizing prin­ ciple of computation. From this standpoint, if we want to understand algorithmic 14 T he notion ofthe incomputable is a development ofthe halting problem defined by Alan Turing in the 1930s. Chaitin explains that the definition ofthe halting probability is based on the existence ofprefix-free universal computable functions, defining a programming language with the property that no valid program can be obtained as a proper extension of another valid program. In other words, prefix-free codes are defined as random or uncompressible information. See Gregory Chaitin, Meta Math! The Questjor Omega (London: Atlantic Books, 2006), 130-31. Speculation
P. 8
AUTOMATED ARCHITECTURE architecture as an extended practice of governance, it is important to challenge the assumption that governance, in its conjunction with computation, corresponds to a topological form (the continual variation of a whole connecting every point). Instead, a closer look at automated architecture suggests that algorithms, or the stuff of computing, are parts that can be bigger than wholes, as much as wholes can be smaller than parts. Hence, automated algorithms have the opportunity to turn the topological aesthetics of governance into a fractal arrangement of incom­ putable infinities. 37° Speculation