(re)consideration
I am all too aware that there has been little activity on here in the past month. This is not an indicator of lack of work, it probably should have been raised a flag for a lack of clarity on my part for the procession of this project. The following will try to explain my understanding of how my thoughts/interests have progressed, and how I intend to bring my somewhat disparate explorations together for my final project.
This project proposed is an architectural installation in the School of Architecture using the methods I learned throughout the semester. More information on the site will be posted later.
In the last meeting I had with my project advisor Professor Ambrose we had a heated discussion about how/what I was going to build for the final project. What I proposed at this meeting was a “growth” system based on an algorithm that I had been working on which would take an object, allow the user to identify connection points/surfaces on that object, and then the object would “self-replicate,” attaching to itself to these connection points on its copies. The growth would be controlled/directed by a bounding box dictated by the site that the object could replicate within.
This approach seemed fruitful at first, but after some reflection I realized it was quite flawed. I now I realize that had I attempted to proceed with this route, the result (or lack thereof) would have likely been disastrous.
The first fatal flaw was that there is no inherent organizational structure to the system I was building - fundamental for any complex system (especially if I expected to build it). As a material I had proposed to use steel “L” and “T” brackets from Home-Depot in various sizes and connect these brackets to each other (not to wood) using the algorithm. Without any sort of code to detect the fitness of a given connection or to regulate the angle that the pin connection would take, the generation would have been blind (heavily reliant on randomization), sloppy (difficult to guarantee logical connections), and very probably unstable. The structural stability was a worry of mine from the beginning, and knowing the difficulty of making it stand, and uncertainty of writing viable code that would be able to manage such complexity from the top down (like a giant brain) including the connections for several hundred pieces of different types and the subsequent recursive reorganization of the system based on fitness, I chose a material (steel) which I thought might be able to stand simply because of its material nature.
This forced reliance on such ambiguous tectonic considerations was also not a clean one and points to the second problem…I would need to choose a material for a system whose structure/geometry I could not know ahead of time, but could not test without the randomly choosing a starting geometry. If the connections were to remain simple, which they would have to for coding sake and the scope of this project, I needed a simple 3 dimensional shape that was structurally rigid, strong, light and standardized. This, however, wasn’t the real problem and even if I had discovered a viable object the form generation would still have to rely heavily on a randomization, at least for the initial state.
The real issue I now believe, was a lack of clarity in what I had been studying and how these topics might lead to a cohesive final project. Much earlier I had realized that I had not seen anyone attempting what I had proposed, likely because it was not really indicative of anything except an algorithm and a random seed. This lack of clarity prevented a true juxtaposition of my intentions to those of the precedents Michael had suggested, namely Nader Tehrani’s Immaterial installation at Harvard, and Shop Architects’ prefabricated wall cladding for an installation at MoMa (?). My understanding of these two projects are as follows:
- Tehrani - A materials study, yielding a part that was used to create the whole, probably not deterministically, that is, the final shape was probably not modeled precisely via an algorithm prior to the installation.
- Shop - apply a decompositional algorithm to a predetermined form (wall) to yield a standardized kit of parts which provide a nonuniform texture for the object while remaining time/cost effective.
My proposal was unlike either of these. I had neither focused on material, nor decompositional algorithms. This forced me to revisit my intentions/interests for the project.
From my notes following the discussion with Michael:
- Intensions:
- Demonstrate possibility for nondeterministic computational design
- Part to whole relationship/ self similarity
- Emergent Complexity from simple rulesets
- Issues:
- Simple construction through digital fabrication
- need a simple system
- grid
The majority of the studies I have done this semester have been fundamentally about a surface condition, whether it was the description of a surface condition through the repeated application of an object to that surface (Sept-Oct postings), or a the creation of a surface condition via the arrangement of objects between vectors, and the effects on this condition following the subsequent manipulation of those vectors (Nov. postings). I also realized that there had been some overlap between these two types of studies that I had not actually “published.” Beginning with the surface normalization script, I had collapsed the the arrayed objects to lie on a flat surface to see how the collapsed surface still exhibited the spatial condition of the curved surface.
Simply stated, I am interested in letting the part determine the spatial condition of the whole through an implied surface condition. Earlier research explored the implication of a spatial condition through part to whole orientation, and I would like to stay as true to that as possible in this next phase.
Perhaps the most important principle that I should try to adhere to when working with these complex systems is simplicity. This is not an oxymoron. The power of computational design lies in the ability to leverage computation to yield incredibly rich and sophisticated systems from simple rule sets.
To this end, the first phase of the reconsidered final project will explore:
- A system with a grid, or other simple organizational system.
- An autonoma system, perhaps similar to the flash example posted in October, where the surface orientation of the part reveals the greater spacial condition through a localized rule set that governs each part based on that parts immediate context.