Archive for 11/11/07 - 11/18/07

Algorithmic body... reactions.

Monday, November 12, 2007 § 1

It is amazing how fast information can travel today, and it is even more important that this can promote dialogue and exchange of opinions, in a way that would have been impossible using just the traditional “printed” means. This was the first thought that occurred to me when I was browsing archiblog today and I came across the article entitled “Of Buildings, Computers and Telescopes” published originally on the HAD blog, by Rafael, a computer science student in the University of Sussex (unfortunately no further information is available). Subject of this article is to justify how one of my projects, Algorithmic Body, is using “invalid”, “grossly-erroneous reasoning methods” and totally fails to meet its aims by “contradicting its architectural intentions”. The above comments of the author are used in order to justify the thesis of the article which more or less is that:

it is possible for a computer scientist to acquire the knowledge necessary to produce software that automatically designs buildings” while

it is not possible for an architect to acquire such knowledge without formally studying computer science, or becoming a computer scientist”.

I will not stay to the fact that academic dignity requires at least a full identity when attacking somebody else’s academic work, and especially when using terms as the ones I just mentioned. That is not really the point, and as I already mentioned, the nature of the internet is totally different from the nature of printed media. But I think that it does make sense to try to answer to the claims of the article, since that would bring forward some important issues that architecture is facing today. And I will start from the first part, my “grossly-erroneous” project:

The main argument that justifies the grossness, according to the HADblog posting, is that I state in the description of my project that “one of the most interesting characteristics of algorithms is that they can be totally self-referential. An internal quality that can produce a process independent and autonomous, that relies only on its own rules.” According to the HAD posting by that I:

ascribe to algorithms a property that is fundamentally incompatible with the concept of "algorithm" (i.e. "total self-reference").” And I

Judge this non-existing property in algorithms ("total self-reference") as "an interesting feature of algorithms".”.

Quite interesting I must admit…Now let me elaborate the way I see it: the algorithm that I was referring to, and which I used in my project, is a cellular automaton algorithm. Now, it might be because I am Greek, but to me it seems rather self-evident (which in Greek would be auto – noito) that the word “automaton” (which in English would be self-powered) is inside the expression “cellular automaton” for a reason. But since not everybody is able to understand the reason behind the use of words I will try to explain: The word automaton is made out of the word ‘auto-’ which means self and the word ‘-maton’ which means power. Therefore: self-powered. To clarify things here is what has to say:


1. a mechanical figure or contrivance constructed to act as if by its own motive power; robot.

2. a person or animal that acts in a monotonous, routine manner, without active intelligence.

3. something capable of acting automatically or without an external motive force.

[Origin: 1605–15; L: automatic device Gk, n. use of neut. of autómatos spontaneous, acting without human agency, equiv. to auto- auto-1 + -matos, adj. deriv. from base of memonénai to intend, ménos might, force]

Therefore a cellular automaton is an a priory self- powered system. While a cellular automaton algorithm is running, it is using only the rules found inside the algorithm itself in order to define its next generation. Therefore it is always and totally self-referential. Of course it needs an initial state (input) and it provides a final result (output), but obviously enough, the first comes before the algorithm (and can be defined in many ways, algorithmically or not) and the second comes after the algorithm (and of course can be manipulated in any way in order to be transformed into a building or anything else). The algorithm itself remains an automaton. In other words, a cellular automaton algorithm is by nature self-referential; other algorithms may not be. However, all recursive algorithms tend to be self referential. (1)

In any case, what is important in an architectural context is not the question whether the algorithm is self-referential or not, but the fact that the architectural process is self-referential. Therefore the process in my project is a process using rules that are internal to it and are not referring to any of the traditional external architectural factors, like program requirements or site conditions.

Nevertheless, there is a even more important problem in the thinking behind the discussed article, that doesn’t have to do with the correct or erroneous use of words: The problem is that the concept of algorithm as employed in architecture is evaluated in relation to the concept of algorithm as used in computer science. Concepts and ideas transferred from one field to another, not only can be altered and redefined, but they have to undergo such a process, if we want them to produce new ways of thinking. Of course that is a position, that underlies my work, and is largely based on the writings of Gilles Deleuze and Felix Guattari (look for example the way that they use concepts like rhizome, diagram, machine etc). Supporters of ‘universal truths’ and “pure reason” will of course disagree. And in that context it makes sense that the posting refers to Alan Sokal, author of the notorious book Fashionable Nonsense, where he tries to render the work of post-modern thinkers like Deleuze, Lacan, Kristeva and Baudrillard as nonsense, based on the idea of a ‘misuse’ of scientific terms and concepts. What does not make any sense at all is that on a previous article on the same blog, Manual DeLanda’s article “Gilles Deuleuze and the use of genetic algorithms in architecture” is also used in a similar fashion. But that is a whole different discussion.

The above observation is actually answering why computer scientists, unfortunately, will never be able to produce architecture. Because science, as Deleuze and Guattari point out in their last book “What is Philosophy?”, must always isolate a small fraction of reality in order to operate. On the contrary, philosophy, and I will argue, architecture, have to deal with an extremely broader field, which tries to include almost everything. If one succeeds to understand that difference (obviously Sokal and the author of the article is far from success in that aspect) one may very well become an architect without a formal education (in a similar way that Le Corbusier or Tadao Ando did). Similarly an architect can use computer science in architecture, as long as he understands the qualitative and quantitative differences between the two fields.

I will use a quote from an interview of Marcos Novak to explain myself better: “To be fair, though, there are at least two architectures, the architecture of accommodation, and the architecture of excess. Accommodation produces buildings, excess produces 'Architecture.'”(1) So, it might be possible for someone to write a software that will produce buildings; it is a whole other story to produce Architecture. Architecture goes far beyond problem solving, to suggest new ways of thinking and inhabiting space.

I will conclude with a phrase from the beginning of the HAD posting that triggered this text. And let me replace the words computer science with the word architecture:

“It is understandable that people who are not exposed to architecture miss the breadth of the discipline and relate it to something more familiar.”


(1) “A recursive definition or a recursive algorithm is characterized by self reference. Typically with recursion, a function is defined in terms of an earlier version of itself” Cragal,J.M. Descrete Mathematics for Neophytes: Number Theory, Probability, Algorithms and Other Stuff.

(2) Interview with Marcos Novak by Knut Mork, (c) 1995 on :