chunks of data
chunks of data
C: One view of Parametricism is that you design a system that you feed inputs into to produce outputs. You can vary the parameters to change those outputs but it is essentially a top-down process. From our brief understanding of object oriented programming, the way it speaks about objects is more of a bottom up system. I was hoping that the two of you could expand on this difference, or if it’s not a difference?
ML: Although you mentioned Patrick Schumacher earlier, I understand that you are not referring to Parametricism as a style but to parametric design as a design methodology. Parametric design, or in another word, designing with a set of parameters and their relationships, is nothing new to the architectural design process. Implicitly it is neither bottom up nor top down. Depending on how these relationships and interdependencies are defined, the design approach may be either bottom up or top down or both. Therefore, Object Oriented Design (OOD), which is understood as using Object Oriented Programming (OOP) to design, is also classified as parametric design. OOP associates objects as data fields and behaviors, defining a system of interacting objects. It’s more of a “collective” data assemblage than an I/O system, therefore OOD may potentially address a bottom up design approach.
However, here again whether the design process is bottom up or top down depends on how these objects and their relationships have been encoded in relation to the design problem, therefore OOD could lead to a design approach that is either bottom up, top down or both.
BB: In the past decade there has been a tendency among architects, which perhaps goes back to Stan Allen’s “From Object to Field” article, to assign a negative connotation to the top-down process and a high regard to the bottom-up. But it’s usually not that black and white. When writing the most flexible algorithm in OOP certain variables are declared and conditional statements set. Therefore, when testing design objectives there has to be a relationship between the top-down and bottom-up systems.
C: Perhaps instead of top-down and bottom-up, it would be better to use Graham Harman’s “overmining” and “undermining” analogy in which a top-down system would be considered overmining, in that you’re focused on the system rather than its parts, whereas in Object Oriented Programming the system is created through the parts and specifically their interactions with one another. Could you explain Object Oriented Programming as it is different than other languages or design principles?1
ML: We could argue that as opposed to some more traditional architectural and computational approaches, OOP offers an alternative to symbolic operative modes by focusing on material organizations and agencies.
Through Object Oriented Programming we can produce complex and consistent organizations using simple rules of interacting objects that communicate, self-organize and develop ad-hoc communities. The distinctive feature of Object-Oriented Programs is that they are “flat” networks of actors and objects gathered up into assemblies. They act through simple, local rules, processing sensorial and physical data, figuring heterogeneous yet consistent wholes. But since you quote Graham Harman, there is also the question of what this part or “object” is in a design space. For instance, are non-physical or non-geometric instances potentially also design objects? And if so, can OOP help us explore complex design problems through the assemblage of non-relational data in the same design space?
BB: Graham Harman’s theory of objects directly builds upon the Heideggerian question of “what is a thing?” Heidegger has been used in the past as a theoretical pillar to describe the relationship of architecture and technology. For architects this return to the theory of objects is due to the introduction of Object Oriented Programming platforms to the design process, which was initiated about a decade ago in architectural schools like Columbia. Programming platforms such as Processing allow writing and compiling code in an efficient manner that allows the generation and simulation of larger populations. This has made the OOP platforms favorable in experimental architecture.
Architects have always operated in multitude of domains at once, and recently they have been drawn to Object Oriented theorists, such as Graham Harman, and Ian Bogost and the influence of the movement in art and literature. This is due to the fact that currently architects are trying to place the computational work generated using OOP in a theoretic context and developing this notion of Object Oriented Design. The Object Oriented Ontology (OOO) camp has also been interested in drawing correlations between ontology and the design aspect, since ultimately the interest is to create discrete behaviors and entities that thrive in relationship to one another.
ML: Yes, it is interesting to look at the parallel discourse around OOO, OOP and OOD. Bruno Latour claims to go back to things (objects) by proposing gatherings of hybrid ecologies. He rejects institutional politics and claims for “Object-Oriented Politics” as a much more effective way to represent the contemporary pixelisation of politics. Within similar lines Peter Sloterdijk proposes an understanding of contemporary politics by absorbing the multiplicity of positions through a material mediation with no need to regularize or homogenize them. In this approach, politics is understood as a mass of hybrid forums that proliferate. On the other hand, in architecture we have mainly focused on the search for new design sensibilities that emerge out of the new digital design paradigm, although we left aside the style question at the beginning of the conversation.
Since you started with the word Parametricism, it might be worth also reflecting on how successful or relevant the exploration on the aesthetics of OOD has been so far, or perhaps it’s all about its aesthetic inconsistency, or the questioning of whether it is ultimately not about style any more but about how the design problem is stated. For instance, Mario Carpo recently has described the digital design aesthetics derived from big data. He refers to a discreet voxel-like architecture as a possible embodiment of an approach that deals with the management of large chunks of data—“objects” that result in a cellular and coarser design aesthetic.
J: Important to OOP is the idea of discrete objects. As Mario Carpo said in “Breaking the Curve,” “the inherent discreteness of nature (which, after all, is not made of dimensionless Euclidean points nor of continuous mathematical lines but of distinct chunks of matter, all the way down to molecules, atoms, electrons, etc.) is then engaged as such, ideally, or in practice as close to its material structure as needed.”2 However, theoretical physics has continued to try to further decompose and reduce our universe to strings and quarks. I think code could be accused of the same thing—reducing agents to specific behaviors or each agent has a desire to do x number of things—lines, functions and parameters. So, at what point is it an object? If it can be reduced to specific intents, what makes it an irreducible thing from the ontological point of view?
BB: An irreducible characteristic of an object is in its definition a cluster of variables and its equal value to other objects. These objects with missions are often referred to as “agents.” As a designer when writing an OOP instance, you break the interactions and conflicts of your system into different objects. You think about how the variables of each object will interact with one another. But, when you assign too many rules to an object, the possibility of what it can become is limited. Complexity in this case is not defined by the number of rules each object inherits but could be defined as how the overall system performs as a whole. When writing a multi-agent system as architects, it is important to test the interactions by visualizing the relationships and often we assign geometric primitives to various objects. Problem arises when we short-sightedly refer to the visualized objects as building or ‘Architecture’. In Object Oriented Design not all the exchanges and layers necessarily have to be visualized. For instance, in video game design when referring to Object Oriented Design, we are referring to more than just the aesthetics of the game but more precisely referring to the chunks of data that are being exchanged. The popular videogame Minecraft was recently rewritten as an object based interaction. Minecraft world, with its sophisticated voxel engine, allows constant exchange between the parts, but as a player you don’t have to know or see all the micro relations. These exchanges become visible when certain conditions occur and new elements emerge.
This notion that not everything has to be visualized is a very difficult concept to grasp for us designers. Often in architecture schools courses that teach OOP require the first time programmers to model an existing multi-agent system—slime mold, neurons, flock of birds—and the first reaction is to visualize every single instance of the objects. Beyond the aesthetics question, it is essential to understand how many types of objects are necessary for an OOP simulation. Going back to our top-down and bottom-up discussion in OOD, if you have too little or too many object interactions in a system, then after a while the relationships become predictable or the system goes nowhere. So, as mentioned before, it’s always about this top-down and bottom-up balance, or learning overtime how much is enough, or what really kills a system.
J: Yeah, I mean it would be interesting to know what tips the scale—when it stops being object oriented and when it is too top-down. One of the main takeaways from the lectures by Gilles Restin and Graham Harman was this term “emergence,” and you used that term when referring to Minecraft, that it was being rewritten to be object oriented and so these things emerge.3 But that’s only possible if it’s not being presupposed or predetermined, so it cannot be too top-down. Is it just a matter of feeling that balance out?
BB: Exactly. I think that’s where the design problem becomes interesting. On the other hand, we are always faced with the question of the subjectivity of the designer.
We can talk about OOP from an engineer’s standpoint in order to solve a problem, or from a scientist’s perspective in order to discover new theories.
C: This idea of subjectivity seems to always circulate around the idea of computational design, and it seems as if you’re saying that you don’t see this way of working as any sort of removal of subjectivity. But the term “emergence” implies an objectivity to the design.
J: We talked about this idea earlier, that there is this ongoing process that keeps running. This is especially true in Processing—the simulation keeps running...
ML: But you stop it...
BB: … and this is very subjective. Often what you want to do with a simulation before you hit play is to make predictions based on your initial objectives. If your final result is similar to your initial predictions, are you really making good use of the computational power? As designers sometimes we are post-rationalizing these processes in a pseudo-scientific way, limiting the possibilities of a design space. In any OOD you do have certain criteria that could very well be quantities or qualities. But how do you evaluate the outputs if it’s not subjective?
If you look at the outputs as your matrix of possibilities, then from this spectrum you select based on the initial objectives which could be budget, efficiencies, etc. But, as mentioned before, it is still very difficult to make this selection as a designer.
ML: What we often face in architectural projects is that unless the computational process is described through what we could describe as an “engineering” approach, say, defining a performance oriented goal and a specific judgment criteria, it often becomes very complicated to argue for it in front of an audience. However, especially at the early stages of the design process when we deal with an open design space, the available data is often diverse. Therefore, computational explorations are performed aiming at discovering new relationships and consistencies.
BB: Sure, ultimately it is about the convergence of these contradictory and yet complimentary relationships.
1 Harman, Graham. “Strange Objects Contra Parametricism.” SCI-Arc. Los Angeles, September 18th, 2013.
2 Carpo, Mario. “Breaking the Curve.” Artforum. February 2014. pg. 172
3 Retsin, Gilles. “Object Oriented Design.” CAAD Lecture, ETH. Zurich, November 9th, 2012