Architecture machinae

Essay on the ‚object‘ of architectural theory

The notion of ‘architecture machines’ has been around at least since the formation of the Architecture Machine Group [AMG] at the MIT. In the early days of digital design and fabrication, this group led by Nicholas Negroponte experimented with the construction of architecture machines as “all-purpose cybernetic design assistant[s] ”[1. Mario Carpo, The Alphabet and the Algorithm, 1. Aufl., Writing Architecture (Massachusetts: The MIT Press, 2011), 35.]. The sort of machines AMG had in mind were instruments to facilitate the design of architecture; in the course of development these machines were integrated also with the production of architecture by way of a digital chain. These design-production instruments are not what this essay sets out to address. It is not about machines used to produce architecture, but about the machines architecture itself consists of; it is not about technology but about society which relies on the cultivation of highly diverging machines.

A machine in the popular sense of the word is a technical object that can be fully explicated by its material determination and its orientation towards a specific purpose, its telos. This image of the machine as essentially mechanistic started to get predominant in the 13. century and gradually deployed within the modern age and the subsequent mechanization, industrialization, and digitalization. In an ideal image, a mechanic is characterized by its full determination; it may be plenty of work to figure out in which state a complex mechanic is and to deduce its preceding or subsequent states, but it can be done in principal, at least by God or Laplace’s demon.

The paradigmatic analytical model of such a mechanistic machine is the Turing-machine: an idealized, symbolic machine, which produces defined symbolic outputs on the base of defined symbolic inputs by way of an algorithmic procedure. It is a closed system which only operates on a defined, finite set of possible inputs, and accordingly it is only capable of producing a predefined set of possible outputs. If something can be processed by this machine depends on whether it can be formulated in the input language. The general strategy of Turing mechanics is to break down complex problems into atomic decisions which can be solved unambiguously―yes or no, true or false―, and to assemble these solutions to answer the initial complex problem. The gordic knot of Turing-machines is the Entscheidungsproblem: there is no general way to predetermine whether a Turing-machine will terminate after a particular input, as recursive patterns may force it to run infinitely. And the crux is that the Entscheidungsproblem is not restricted to complex esoteric problems, but already is inherent in the formalization of basic arithmetic, as Gödel has proved.

But the reading of machines as mechanistic machines is a shortening of the historical meaning of the term. The Greek concept μηχανήmechanē―and its Latin successor machina denoted a mean, an artifice, or a device in both a material and immaterial sense. A machina can be anything which bears the potential to alter a situation, be it in politics, in society, in the sciences, in the arts, in urbanism or in any other realm. Raunig calls these machines that are essentially undetermined and know neither defined inputs nor outputs “abstract machines”―the rather anarchistic relatives of mechanic machines of whom it is not clear whether they even share the same bloodline.

A paradigmatic model of an abstract machine can be seen in the prop in theatre. A prop is everything that furthers the plot: it can be a material object like a fake gun, an immaterial motif such as jealousy, a fantastic devise such as seven league boots, or a spatial setting that includes the audience in the plot; all these props can be decisive for the deployment of a play. The prop is a machine which catalyses the play by concatening persons and deeds, theatrical fiction and everyday life experience, function and appearance [Erscheinung] in an open, playful dynamic.

Mechanistic machines and abstract machines can be differentiated along the difference between instruments and media: a mechanistic machine can be conceived of as an instrument which is supposed to overcome a deficiency; as a prosthesis for the desires and needs of deficient creatures. This way Arnold Gehlen conceives of technology in his anthropology and Marshall McLuhan conceptualizes media as extensions of human communication. On the contrary, an abstract machine is similar to a medium in the sense of Niklas Luhmann: it is not a mean for a particular end, but the precondition for every end in general, as e.g. the binary opposition between truth hood and falsehood is the precondition for each particular scientific hypothesis. Abstract machines are relatives of Kant’s notion of the transcendental schemata as they are the “third thing” (Kant) in which conceptual and empirical knowledge intersect, but with the difference that abstract machines do not reside in the transcendental realm of the a prioric faculties of subjective knowledge [Erkenntnis] . Like Foucault’s notion of theépistéme, or Kuhns notion of the paradigm, they are historically contingent, intersubjective constituents of all public processes such as knowledge, politic, economy etc. To borrow a phrasing from Deleuze & Guattari, abstract machines consist in the intersection of human organs, technical apparatuses, and social mechanisms. They rule the dynamic between a multitude of qualitatively different agencies and are―at the same time―subject to transformations caused by the interplay of these agencies. An illustrative example is the evolution of tax codes: a tax code rules the monetary evaluation of economic deeds and at the same time has to adapt constantly to the change of economical behavior which it itself causes by the criterions it establishes for the monetary evaluation.

Current mainstream architectural and urban planning is prone to center around planning instruments. Here, the naming all too often becomes program: their recipe to deal with the heterogeneous socio-material conditions of contemporary cities is the deferral of architectonical agency to mechanic machines. Planners and city administrations dream of a Turing-machine for urban planning and hide behind politically correct empty words such as objectification [Versachlichung] ; but it is a too seldom recognized phenomenon that only the party that looses an argument will call for an objectification of the debate: theVersachlichung of architecture and the urban debate is precisely its end. There are no objective properties of architecture or of the city waiting to be analytically lifted by Boston Consulting or McKinsey. Not surprisingly, current architecture and urban design only seldomly and incidentally arrive at architecture.

Mechanistic machines aim at the representation of an assumed universal order, whereas abstract machines allow for the construction of new regional orders; they are “ways of world-making” (Nelson Goodman) which establish and maintain regional ontologies (Frederik Stjernfelt)―and eventually dissolve other conflicting regional ontologies, as there can’t be any such thing as an objective abstract machine; in fact, objectivity is itself an abstract machine which prompts different mechanic machines and representational regimes in the course of history, as Daston & Galison have pointed out. Abstract machines are precarious media to escaping reified representational orders. Unlike mechanic machines, they can’t be defined by their telos; instead, they must be characterized by the operational procedures they consist in: they open up possibilities to act gemeinschaftlich and to deal with all kinds of events. These events are not completely independent from any representational mechanics. On the contrary the point is, that the form of representation always is relative to that which it represents. Representation is no unidirectional event, but an epistemic “revolving door” (Günter Abel): on the one hand, what there is is dependent on how it is represented and on the other, representation has to deal with the epistemic obstacles (Gaston Bachelard) the to-be-represented imposes. Accordingly, abstract machines do not represent something that would also exist independent from its representation: they stratify (Deleuze and Guattari) reality.

Architecture as a reality is produced within abstract machines. It is not the material, the form, the history, the function, the symbolicity [Zeichenhaftigkeit] , or the context, but their concatenation qua abstract machine. As per linguistic simplicity I shall address this architectural abstract machine as architecture machina.

It is a question of scale whether one sees cities, single buildings, or urban furniture as architectural machinae. It depends on the scope of the interpretation―both in case of analytically or projectively motivated interpretations and interventions. But on every scale the determining moment of the built environment is―an indeterminate architecture machina. They determinate the material manifestations which eventually amount to the city; the city consists, so to say, of the material traces of the continuously skinning of competing, indeterminate architectural machinae. These ‘virtual-real possibility machines’ (Raunig) are the possibility conditions for decisions in architectural und urban design―which can only be pragmatic, never final. Thus, architectural machinae can never produce a collective identity without rupture; they always deal with regional urban transformations which potentially conflict with other architecture machinae.

In the history of architecture the question frequently emerged whether there is a universal or unified theory of architecture. The question has muted after the collapse of modernism, but the pluralism of developments in architecture after modernism has led some to try to reintroduce the idea of a unified theory of architectural theory, as for example in Patrik Schumacher’s conception of “parametricism”. Transposed to our machinistic narrative, this would come down to the question if there is an universal architecture machina. The answer to that question is already implicitly in the above said and thus not spectacular: as architectural machinae produce and operate within regional ontologies, there can’t be such a thing as a universal architectural machina. Never the less there is something universal about architecture machinae, as all architecture on all scales is determined by architectural machinae. But this universal aspect does not amount to a somewhat Newtonian picture of a totality of all architecture which, in fact, could be produced resp. interpreted by way of one universal architecture machina, but to the necessity of architecture machinae for the emergence of any architecture. Architecture machinae do not represent a structure, idea, or relation independently existing of themselves, but it is them through which architecture for particular situations, ends, and users in heterogeneous spatio-temporal contexts is created. These contexts are of course not given in a naïve realist way―be it in an empirical or idealistic fashion―, but made within the constant flux of a multitude of architecture machinae which intersects and intermingles with other abstract machines of other realms of reality.

Architecture machinae are the practical mean to deal with the architectural uncertainty principle. They―not platonic architectural types or Neufertian architectonical encyclopedias―are the medium of establishing relative certainty of architectonical formations by destining their perception and use. They do not reflect or represent a given universal architectural ontology but produce regional ontologies (Frederik Stjernfelt). Thus it is no use to turn to the especially by Alberti promoted figure of the superior knowing architect-as-author. Instead, we need to turn to architecture machinae as concatenations of technical, social, semiotic, historical, material dispositifs of architectural production.

On the first glance, architecture machines resemble Leibniz’ monads, especially as he has painted the picture of a city seen from different angles and perspectives in order to illustrate his concept of the monad. But there are two distinct differences between architecture machinae and monads. Monads have no windows, i.e. they do not communicate among each other. This is the reason that they must be organized in a prestabilized harmony, i.e. they must be harmonized by some sort of an unmoved mover (Aristotle) which for Leibniz as a disciple of his time was the Christian God. If the monads can’t communicate or influence each other directly, they must be externally harmonized because if they wouldn’t be, there could be no consistent, continuous world at all. Architecture machinae, on the contrary, are not harmonized and do have windows―and here I’m not talking about oriel windows and the like. They are bellicose versions of monads struggling for survival, thus more resembling Deleuzian intensities in the course of whose explication differentiation goes along with de-differentiation of other intensities[2. Which in Deleuze, by the way, do have an “implizierten Charakter” [p. 320, German translation] . Despite Deleuz’ argument, that this character is not an ‘I’ but on the contrary ‘the system of the dissolving I’ [ibid.] , I can’t really see why this character is not an essence, because also the ruled dissolving of something can be seen as an essence/Wesenszug. To me, the crucial point lies not in an anti-essentialism, but in the distinction between process and object: the difference is not an object, but a process; but nothing hinders from saying that the difference’s being a process is its essence. The whole world (or, a number smaller, everything in it such as e.g. architecture) essentially is a process. This is a simple claim (historically going back at least to Heraklitsπάντα εpanta rhei); the problem emerges through its linguistic formulation: language as a structure, i.e. linguistic grammar, is atemporal. It can address the panta rhei-constituion of the world only semantically, not syntactically. The same holds true for all digital notational systems (Goodman), as they all are based on discontinuity. So if something continuous such as architecture is the subject, analogue systems might be the better choice to deal with them; hence my focus on drawing as epistemic medium in architecture. But ya, that’s for another current of thought..]. Architecture machinae do communicate among each other―not because they would speak in a lingua universalis but because they do the same thing: processing architectonical reality. They do so by guiding the construction, use, and interpretation of the build, material environment. Within the historical development of societies, architecture machinae deploy a―temporary limited―sufficient cohesion to inscribe material events in the urban fabric. This cohesion is not a question of mechanical consistency as in the case of Turing-machines, but of socio-cultural construction. Therefore, as with every abstract machine, architecture machinae are unstable and precarious, anarchistic devices.

The practical functioning of architecture machinae[3. Another by-the-way: Rainer Noennig at the TU Dresden has developed the concept of “architectonical synthesizers” (cf. Jörg Rainer Noennig, ARCHITEKTUR, SPRACHE, KOMPLEXITÄT: Acht Essays zur Architekturepistemologie (Bauhausuniversität Weimar, 2007).) out of an epistemological analysis of architectural design. That might be of interest for the topic of abstract machines.] is based on a two-fold iconicity: they resemble the material (in the sense of Aristotle: ληhylē―, latinizised materia) they process on an operational and on a mereological level. Even though they are abstractions of what they process, architecture machinaecan be used operationally similar than that which they process. They function like diagrams in the sense of Peirce or what Husserl calls ‘categorial intuitions’ [kategoriale Anschauungen] : By way of “contraction” (Deleuze) they isolate aspects of the urban manifold, bring them into a form or under a scheme within which they can be worked withgemeinschaftlich, thus making possible the production of material architecture. The urban manifold is contracted by way of architecture machinae , which, in turn, rule the principally contingent dynamic of the material city. Thus, prior to such a contraction it does not make sense to speak of architecture or a city at all; architecture machinae produce the city in the first place and hence they are the ‘real’ (Günter Abel) objects of architectural theory.


Alberti, Leon Battista. De re aedificatoria. München: Prestel, 1975.

Bachelard, Gaston. Essai sur la connaissance approchée. Paris: Vrin, 1987.

Carpo, Mario. The Alphabet and the Algorithm. 1. Aufl. Writing Architecture. Massachusetts: The MIT Press, 2011.

Daston, Lorrain, und Peter Galison. Objectivity. New York: Zone Books, 2007.

Deleuze, Gilles. Differenz und Wiederholung. Übersetzt von Joseph Vogl. München: Wilhelm Fink, 1992.

Deleuze, Gilles, und Félix Guattari. A Thousand Plateus: Capitalism and Schizophrenia. Übersetzt von Brain Masumi. Minneapolis: University of Minnesota Press, 1987.

Husserl, Edmund. Logische Untersuchungen. Prolegomena zu einer reinen Logik. Bd. 1. Reprint. Niemeyer, Tübingen, 1993.

Kant, Immanuel. Kritik der reinen Vernunft. Stuttgart: Reclam, 1966.

Luhmann, Niklas. Die Wissenschaft der Gesellschaft. Frankfurt/Main: Suhrkamp, 1992.

Neufert, Ernst, und Johannes Kister. Bauentwurfslehre. 38. Aufl. Wiesbaden: Vieweg, 2005.

Noennig, Jörg Rainer. ARCHITEKTUR, SPRACHE, KOMPLEXITÄT: Acht Essays zur Architekturepistemologie. Bauhausuniversität Weimar, 2007.

Peirce, Charles. Collected papers of Charles Sanders Peirce. Herausgegeben von Charles Hartshorne, Paul Weiss, und Arthur W. Burke. Cambridge: Harvard University Press, 1931.

Raunig, Gerald. Tausend Maschinen. Bd. 7. Es kommt darauf an. Wien: Tauria+Kant, 2008.

Stjernfelt, Frederik. Diagrammatology: an investigation on the borderlines of phenomenology, ontology, and semiotics. Dordrecht/London: Springer, 2007.

Turing, Alan. „On computable numbers, with an application to the Entscheidungsproblem“. Proceedings of the London Mathematical Society.

vor 11 Jahren

3 Kommentare

Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert.