Friday, July 17, 2015

Model and Agency

Although we earlier referred to the model's authority, this wrongly suggests that the model is the ultimate arbiter of design decisions. In fact, “a computational engine is not a conclusion but an evolving document which formalizes, refines, and clarifies its authors’ intents.”(1) Moreover, the potential stimuli reactions and invariant relations built into the model are activated to highlight significant concerns and direct the attention of subsequent users toward particular issues, what one might call the 'bias' of the siimulation.(2) When the computational model assembles a heterogeneous array of unit operations that simultaneously embed material, functional, and discursive modes of representation,”(3) it is capable of transitioning from purely quantitative matters of fact into matters of concern(4) that also incorporate associations and intentions.

In contrast to the 'parametricist' approach,(5) the goal is not a correlation of diverse data into a single communicative platform,(6) but to only to frame this information such that it can be positioned(7) within a single conversation. The imperative on the designer is thus to afford multiple types of interaction simultaneously against a changing background. In contrast to a conventional design process, the computational model is not structured as stages with distinct solutions but as an integrated model where different kinds of information engage the designer to address from many angles a single, complex situation.(8)

to flickr
RAK environmental model interface

Finally, we argue that the expanded field of rhetoric provides a more substantial support for design agency, as “a purposeful inclusion of critical practice and the architectural project but also of the more specific use of artificial intelligence techniques in a design setting,”(9) Just as a rhetorician first establishes an argument and then engages in a discussion or debate, so design agency is exercised first as authorial intent asserted through the construction of a configurative model,(10) and then again through an interactive engagement of that system that develops a relational understanding of the situation from a particular point of view.(11) In the chapters immediately following we will examine various way by which a designer might work in this mode. Following that, we will return the question of why this is especially needed in urban design and argue that this thesis is not limited to computational design but can be extended to a general theory of urbanism.


Friday, May 1, 2015

Procedural Rhetoric

Following from this logic, we observe that unit analysis produces a unique mode of engagement and representation. The lack of holistic or consensus meaning in the system suggests a complex interaction for the critic, author, or user, but this is not necessarily new. The combination of such open-ended meaning with the dynamic aspect of computational models, however, is. Procedural media “have to be operated. They are not static objects, but active devices, machines rather than texts.”(1) As such, the understanding of the model must occur through interaction and occur “with an eye toward identifying and interpreting the rules that drive that system”(2) not just an assessment of the end results. This is a particularly salient point in design fields, which are accustomed to separating process from product and to seeing products as singular artifacts rather than series of multiples.(3) Bogost argues that scenario modeling constitutes a “more abstract representations about the way the world does or should function.”(4) than does conventional verbal or visual discourse, because the familiarity of those modes of communication have made their tropes more deeply ingrained. At the same time the active response of a computational model can create a more encompassing engagement that prompts further response and deeper attention than a conventional image.(5)

The goal of procedural representation, then, is to maintain the active dimension of the simulation, directing the user's attention from the product, which is just one contingent state of many, and toward a consideration of the logic behind the scenario being played out, and the simulation that supports it. That is, to encourage computational thinking. “Computation is representation, and procedurality in the computational sense is a means to produce that expression … computer processes are representational, and thus procedurality is fundamental to computational expression.”(6) Computation represents itself best through processes and aims to likewise be interpreted as process rather than images or words.(7) This immediately calls to mind the work of Charles Sanders Peirce, whose semiotic model was not based on a dialectic between the signified and signifier (8) but on a triadic interplay between the representamen, the object, and the interpretant (9) in which the image of the sign brings forth a new image in a process that extends into an infinite series.(10) In this model the interpretation of the sign is not given as meaning but as a continuation of representation.(11) This process, which Peirce called “pure rhetoric,”(12) was not seen as a degradation of nor affront to the real,(13) but as a creative process of reproduction in kind.

Suggestively, Ian Bogost, has proposed the name “procedural rhetoric” to cover techniques “for making arguments with computational systems and for unpacking computational arguments others have created.”(14) Following the category of “expression that represents processes or systems with processes or system,”(15) (and Peirce's thesis that the interpretant further produces new representations (16)), procedural rhetoric covers both how “simulation authors … think about their objects as systems and consider which are the laws that rule their behaviors,” and the ways in which “people who interpret simulations create a mental model of it by inferring the rules that govern it,”(17) neatly bridging the problematic split between the designer and user discussed above. Ultimately, the confrontation between the authorship of the creator against the application by the user is flattened out in the jump to infinite series enabled by the authority of the simulation itself.(18) Bogost frames procedural rhetoric primarily as a persuasive tool, with a primary interest in how videogames can be employed as a medium of critique or political statement.(19) From this perspective, he writes, “Persuasion is related to the player’s ability to see and understand the simulation author’s implicit or explicit claims about the logic of the situation represented.”(20) Interacting with a simulation, requires one to make attempts to understand the logic (21) and “to analyze, contest and revise the model’s rules according to his personal ideas and beliefs.”(22) Though many procedural media restrict the possibility of rewriting the logical rules of the simulation itself, such calibration of procedures, responses, and degrees of freedom is often possible within—even typical of—a computational design process. “The iterative, reductive blending of the model’s system of transition functions over the course of the design process produces an explicitly structured and strategically searchable solution space” (23) This 'intention space' formalizes the model's ideological position, and reframes the persuasive role of procedural rhetoric “from the simple achievement of desired ends to the effective arrangement of a work so as to create a desirable possibility space for interpretation.”(24) These interpretations (or interpretamen) reflect the Peircian rhetorical mode, and can themselves related to one another diagrammatically as a non-Euclidean spatial figure,(25) which can illustrate trends or groupings of potential parameter states. A better understanding of the parameter space can be used in a more traditional argumentative role to construct a more explicitly comprehensible parameter space (26) and “to address the logic of a situation in general, and the point at which it breaks down and gives way to a new situation in particular.”(27)


Sunday, March 22, 2015

Unit Operations

By considering computational models as complex manifold of actions and agencies, we position the model closer to the conceptual idea of the city as an endlessly reconfigurable assemblage (of assemblages) not structured by an overarching law. This does, however, create some difficulty in assessing the success or weaknesses of an urban design. Ian Bogost introduces the concept of 'unit operations' as an interpretive tool that engages with procedural media on their own terms.(1) As is the assemblage in assemblage theory, the unit in unit operations is very loosely defined: “In essence, a unit is a material element, a thing. It can be constitutive or contingent, like a building block that makes up a system, or it can be autonomous, like a system itself. Often systems become units in other systems.”(2) More significant is the way that multiple units relate to one another within a work. In contrast to hierarchical systems that “regulate meaning for their constituents,”(3) these units aggregate into “a configurative system, an arrangement of discrete, interlocking units of expressive meaning.”(4) The important detail here is that the level of operational control remains in the individual unit and though these units may be said to form systems, the systems themselves do not become structuring forces but are “the spontaneous and complex result of multitudes.”(5)

to flickr

Thus “rather than attempting to construct or affirm a universalizing principle, unit operations move according to a broad range of diverse logics, from maximizing profit to creating new functional capacity.”(6) The diversity of logics in play requires close attention to the ways that individual units are positioned within the system as well as how their influence spreads through the network. In the first case, rather than trying to interpret the overall meaning of a network, one might analyze how a particular point of view manifests itself within the context of the network. In the second case, a study of the indirect effects that arise from an isolated action may be called for, or tracking of how the coherence of specific information changes over time. In either case, the emphasis shifts to an exploratory or interpretive response of the situation and away from the “attitudes or values that inform the approaches that created the systems in the first place.”(7)


Sunday, March 15, 2015


One of the defining features of computational design is that it is inherently marked by a divided or displaced authorship. “to write procedurally, one authors code that enforces rules to generate some kind of representation, rather than authoring the representation itself.”(1) The rules then generate a specific instance through the “intervention of some additional agency that may be other than, and even unrelated to, the … designer.”(2)
Mario Carpo defines this relationship as the 'Split Agency' of computational design and, though he typifies the product as an “an open-ended algorithm, or a generative, incomplete notation,”(3) he confers a hierarchical blessing on this objectile(4) naming the programmer as the “primary author”(5) or the “real digital author.”(6) Meanwhile the user or operator of the system is relegated to becoming an 'interactor'(7) who “exerts only a limited and ancillary form of agency”(8) in adjusting the scenario to a specific case or to personal taste.

Immediately, however, we recognize that this simple division does not hold in any practical sense. Carpo admits that “in practice, these two stations of agency are often merged into one, as a single agent often does both jobs – first designing the general program, then finalizing one or more specific objects designed and made with it. This is normal and to some extent inevitable.”(9) Yet, this still ignores the fact that the programs that he portrays as determining is itself always designed within the constraints of other software or coding languages and is often extended into unintended functionality through independently developed plug-ins or libraries,(10) or more prosaically through user scripts.(11) Moreover, there is no room in the two-level split agency for cases where the computer itself plays out its own scenarios through programmed automation or generative solver.(12) These cases highlight the fact that in all events, the model itself exercises an agency that does not properly belong to either of Carpos two agents. Neither is the programming nor the application able to detach from the other as an entirely inclusive, self-controlled activity, but both are conducted as inter-active acts.(13)


Monday, March 2, 2015

Unique Traits of Computation

While certainly not the only solution to these criteria, computational modeling is uniquely suited to provide insight to the task of a complex, autonomous, urban design. As an inherently,(1) even uniquely,(2) procedural medium, it is well-suited to simulating behaviors, and in the automation and wide variety of possible responses, computation is more capable of producing formal, yet complex, systems than other methods, for example participatory or game-based planning.(3) Lorenzo-Eiroa argues that computation has “shifted the mapping of extrinsic content to the coding of emergent content or agency.”(4) That is, in place of translating external forces from the environmental, technological, or political realm into notational representations that inform design,(5) these forces are being explicitly formalized as inputs that are increasingly absent from the representational aspect of computational modeling, which has shifted its attention to the result of playing out these forces (and others).(6)

Auxin Flux Canalisation Process on Vimeo

Procedural representation, as a medium, is not quite that simple, however. ”Procedural systems like computer software actually represent process with process. This is where the particular power of procedural authorship lies, in its native ability to depict processes.”(7) In this sense, the representational mode of computation is not constrained to a simulation but is at the same time, an enaction.(8) This allows representations to more closely parallel their objects and produces a more tangibly comprehensible image of the logics depicted,(9) but also stands on its own as an active environment(10) which can be engaged with independently of its role as a representamen.(11) While representation is always a creative process, putting forth an “independent object” that “defines an agency,”(12) computation hones this point by establishing a self-reflexive representation, aspects of the program are represented within the code to other elements or objects within the program. In the same way, the computational model “represents a formal logic which governs the formation of a category or type,” and at the same time “a unique solution.”(13) In contrast to instances that become “problematic when there is no agency at a representational level, such as when the content represented is extrinsic to the performance of its medium,”(14) computational agency is located precisely in the representational register. This fact motivates the close focus “on the emergent quality of code”(15) and the particular attributes and constraints it engenders as a form and an “autonomous logical system.”(16)


Thursday, January 29, 2015


The urban model that we put forward, then, must proceed from assemblages that satisfy the parameters set out in the previous chapters: sensitive to various contingencies and able to develop alongside the surrounding environment through a mutual production of the city(1) but yet not simply relational entities defined entirely from without. Rather they must posses an internal consistency through which external pressures are absorbed, translated, and made sense of. Unlike some recent applications of object-oriented philosophy within urban design, we are not especially concerned with classifying the type of object that best represents the city,(2) nor do we advance a naïve formal figure of objects based on a kind of metaphorical materiality.(3) Instead the imperative is to develop an urban model whose components possess degrees of individual autonomy from the urban plan as a whole.(4)

to flickr
Process screenshot in GenerativeComponents from my Master's thesis

In particular, it is necessary that the constituent urban assemblages are able to react freely to situational conditions so as to enable the kind of spontaneous formation that is the goal of Assemblage Urbanism. “The indeterminacies of the formative processes of urbanism require methods which specify its propositions provisionally.”(5) For this reason, while the city or masterplan as a whole does constitute an assemblage of its own, such a frame is less appealing to this thesis, as it renders all of its responses subject to a single endo-consistency. The aim, then, is not only that the assemblages afford some possibilities,(6) but that the internalization of behaviors and affects ground an object agency that registers its own apperception and exercises a decision-making capacity with significant impact on its own development and as an action or force “that leads to one particular enactment of the city”(7) Additionally, the elements of the model must be discrete individuals independent from one another, in order to establish their identities through the selective inclusion or cultivation of their environment, their parts, their relations, and their characteristics.(8)

The effect of inconsistent assemblages engaged with one another is an ecology of nonlinear interaction needed to provide our urban model an analogue of the dynamic behaviors of actual, lived urbanism. The key is in enacting meaningful responsive(9) interaction, that is, responses that “make a difference”(10) in their execution in contrast to the perfectly coordinated responses of linear effects that do not sufficient distinguish between individuals but are more typical of the internal actions of a highly regulated assemblage. As we said earlier,(11) urbanism operates also through the openings created by the heterogeneous differences embodied within an assemblage and the potential for transformation present when assemblages encounter unexpected or conflictual reactions. Apart from conflicts and effects of translation, the interaction between model elements become nonlinear when they do not follow a single generative timeline or scalar progression, but operate at multiple levels simultaneously or feed impacts back onto the process inputs requiring a readjustment of the current state.


Sunday, November 23, 2014


Inclusion, according to Deleuze, is formed by the monad's apperception. Inclusion carries events into the monad,(1) enabling exo-relations and promtping individual manifestations. Because identity for the objectile is not a recipricol definition but always a vector,(2) perceptions advance differentially: infinitesimal variations of perceptions that develop inclusion.(3) Of the different types of inclusion(4) we are interested here in how the monad includes the world within itself. Consistent with the object-oriented position, there is no object which functions as a universal world, one that can contain all others. Despite the fact that, for Leibniz, every monad includes the whole world, the “the reason of the series... is not. The limit remains extrinsic and appears only in a harmony preestablished among the monads.”(5) In this formulation, every monad is a singular subject; however, these subjects are themselves without objects, “these are minute perceptions lacking an object, that is, hallucinatory microperceptions.”(6) Despite the inversion of intent, the result retains a correspondence with the object-oriented goal of “subjectless objects” because no monad is thrown under another subject as its correlate, but each exists only for itself.(7) What is changed is that there is no longer a guarantee nor a necessity of overall harmonious convergence across the totality of monads.

For Leibniz, the convergent harmonization of these series was required by the imperative of a single compossible world that is sharply delineated from all others. Deleuze, drawing from Riemannian manifolds, introduces “a fibered conception according to which 'monads' test the paths in the universe and enter in syntheses associated with each path... a world of captures instead of closures”(8) In this model “bifurcations, divergences, incompossibilities, and discord belong to the same motley world,”(9) or rather, a plurality of non-exclusive worlds.

to flickr
Fruit cultivation Wanmu Orchard, Guangzhou

The relation of object to world is a complex one; “there is always a double antecedence: the world is virtually first, but the monad is actually first.”(10) To clarify, we would say that the point of view of the monad precedes the individual object(11) as a potential series of interaction between the monad and surrounding objects, but that the world, or environment(12), that it occupies does not pre-exist as such. Part of the generative ability of objects includes their “active role in constructing their environment, both through determining relevancies in the environment and through actively changing their environment.”(13) For each and every object, therefore, there is a unique environment, which it includes.(14) However, in the same way that objects that become components of an assemblage do not give up their distinct identity or agency to become docile parts only, so do the environmental conditions “exceed the object, they are equally the conditions involved in other existing objects, and that cannot therefore be specified as belonging to that object alone, nor as terminating in it.”(15) In particular one can say that “while objects construct their openness to their environment, they do not construct the events that take place in their environment”(16) and relate to it in a feedback cycle of construction and constraint.(17) The need for objects to form “contingent strategies for contending with the environment”(18) constitutes the ground of exo-relations.

Though Deleuze prefers the metaphor of the fold to convey the complexly implicated interior of the monad,(19) the virtual dimension from which objects are unfolded is not a pre-individual stratum that is continuous like a sheet of fabric. Instead it is like an entangled knot or rhizome: not everywhere continuous but, through a complex selection, continuously interconnecting. “This genesis is a genesis from other objects or discrete individuals, and in many instances is productive of new individual entities.”(20) It is perhaps better to use Leibniz's own images of every portion of matter as teeming with individuals “like a garden full of plants and like a pond full of fishes”(21) in order to remind ourselves of the complex plenitude of components at every scale. As an alternative to assembling or constructing, then, we might speak of objects as 'cultivating' their environments.