Reductive and Playful Taxonomy: Would Duchamp Desire a Turing Machine?

With this article I aim to draw attention to some of the difficulties inherent in the production of taxonomies and definitions per se. In the process I hope to illustrate the special pertinence of this problem with respect to the classification of digital artwork. It sometimes seems as if the usual instabilities and uncertainties of taxonomical practice are amplified when discussing work that involves the use of a computer as a platform, medium or production tool. I would like to suggest that this is due largely to a tension between the fundamental nature of a computing device and the kinds of artefact that this nature facilitates. In order to grasp this notion fully we will need to consider computers under the aspects of output and process – which is to say that we will need to view them from the vantages of surface and depth.

In one sense the machine logic of a computer perfectly exemplifies a kind of formalism. Being an almost sculptural manifestation of a rationalist mindset it foregrounds concerns of clarity, efficiency and parsimony. It seems slightly absurd – and no doubt ironic – that this collection of strictly formal requirements should give rise to a tool ideally suited to postmodern practice and production, with its contrary aesthetic of plurality, interpretation and difference. In light of Lev Manovich's not too distant comments concerning the influence of Turing and Duchamp on digital art in general[1], I hope to look at the dichotomies in their thinking from a kind of taxonomical perspective.

Many of the texts which concern themselves with digital arts, multimedia and related disciplines set out in the first instance to ascertain what is unique about the artefacts associated with this territory. It is as if the authors hope to produce an essential set of defining characteristics that will serve to place their subject matter and to simultaneously capture it. This approach would seem to fixate upon the aspects of language that are primarily concerned with naming and labelling and in this sense they neglect some wider linguistic possibilities. In light of this, the attempt to define digital arts becomes an interesting vehicle for a consideration of Wittgenstein's comments concerning family resemblance[2].

Until Wittgenstein, much of the history of philosophy had been preoccupied with the process of definition. Many philosophical questions seemed to be of the form 'what is x?' The x in question is a variable that may have stood for concepts as seemingly diverse as self, goodness or, in the case of aesthetics, art. Much philosophical debate centred upon expounding definitions of a concept, showing the inadequacy of existing definitions or attempting to formulate more inclusive/exclusive alternatives. We see echoes of this strategy in the more contemporary debates attempting to ascertain the nature of intelligence, mental illness and perhaps more pertinent to this context, both information and digital arts.

The perennial problem of definition is the belief that in order to capture a specific class we must have a taxonomy that will include everything that is of that class and simultaneously exclude everything that is not. Relating this specifically to digital arts, a definition of this sort would require some criterion or set of criteria that could be evidenced throughout the entire spectrum of digital artwork. Every instance of valid work would exemplify these qualities and likewise there should be nothing that possessed these qualities that was not an example also. As we can see from the efforts of recent commentators, this can prove to be a somewhat difficult task.

In the field of philosophy, the influence of Wittgenstein encouraged us to become more pragmatic in our approach to questions of this form. Throughout the course of the last century philosophers inspired by Wittgenstein became less fixated on the ingredients and essences of concepts and more sensitive to studying the diversity of employment of a given term within a language. They also became more sensitive to the various social stage settings that served as a backdrop or context for the use of a word. Probably the most important insight with respect to our desire for a taxonomy of digital arts is the notion that our words need not point to or label some kind of entity called a 'meaning'; it is rather that in using a word as a social collective we bestow a meaning upon it. The slogan that became associated with the Wittgenstinean position states it clearly: ≥Meaning is use≤.

≥I can think of no better expression to characterise these similarities then 'family resemblances'; for the various resemblances between members of a family: build, features, colour of eyes, gait, temperament etc. etc. overlap and criss-cross in the same way. – And I shall say 'games' form a family.≤   (Ludwig Wittgenstein, Philosophical Investigations, 1953)

Wittgenstein's comments concerning family resemblance attempted to foreground the lack of essential qualities underpinning concepts that were nevertheless happily at work in the world. He famously drew attention to our ability to recognise a diverse set of extended family members without the requirement that there be a single feature (type of nose or colour of eyes for instance) that is common to them all. That is to say, there is a wide spectrum of similarity that exists between a set of related people and we need not necessary try to conceive of individuals as being strung together as beads upon a single thread. Wittgenstein's point is that there need be no single strand of commonality in order for us to employ a term – and this would seem to remain the case when the term in question is 'digital art'. It is perhaps instructive to notice that the qualities suggested by Wittgenstein with respect to recognising a family encompass a very broad set of criteria. The inclusion of gait and temperament as candidates is an attempt to prevent us from fixating too narrowly on one type of feature (i.e. physical features) alone.

Ultimately Wittgenstein recommended description of the use of a term in its broadest context of employment as being more constructive then attempting to reduce anything to essentials. He would probably have urged us to recognise that digital arts, like games, form a kind of family. That is to say, in order to grasp digital arts more fully we might chart the various strands and overlaps of the term as it is employed in day to day living - from discipline to discipline, context to context. In short we would need to discover how it fits broadly into a culture – and 'culture' here means 'form of life'.

There are two insights relating to the Wittgenstinean approach that seem to me to be particularly valuable. The first is the lack of ultimate closure of definition. This allows a concept to continue to evolve over time along with new ways of speaking - and in this case new ways of making and producing work. This is important when we consider the perception of the avant-garde as a fringe activity that is typically regarded as being concerned with pushing at boundaries and radical innovation. There is always uncertainty as to whether a given practice will be legitimated by art institutions and absorbed into the officially sanctioned vocabularies of criticism - and even then there is a question as to which of its aspects will be prioritised.

The second consequence of adopting a Wittgenstinean position is the fostering of a certain kind of pluralism. This is interesting as it allows for a number of vocabularies to coexist. As such it is important for the development of specific schools and traditions within a single discipline. Linguistic plurality is further complicated in the context of digital arts because of the number of disciplines – with their related communities - that have an impact upon the work. Each of these communities has its own specific vocabulary and terminology and in some cases they have more than one. Nelson Goodman, taking a neo-Wittgenstinean stance has suggested that some of the most interesting conceptual territory lies in the borderlands and overlaps of different linguistic sets[3]. In retaining a lexical pluralism whilst mapping degrees of semantic overlap we enrich a subject area. That is to say we might cross index a number of vocabularies whilst retaining the richness that is provided by linguistic independence and subtlety of nuance. If we give this approach the respect it deserves then perhaps what we require is a more relational form of taxonomy.

Any taxonomy might be seen as a product or extension of a working language. With this in mind we might avoid the pitfalls of reification that may at first seem liberating but can equally become an obstacle to the flourishing of both language and practice. We might do well to remember that the rulers that frame a 3D modelling environment can in one sense be said to measure the objects that exist within the environment - but that in another sense they can be said to construct the space itself. In a similar fashion, despite it being possible to describe a phenomenon in terms of a specific schema it does not follow that the resulting classification has revealed something fundamentally intrinsic to its subject matter – or that it has clearly and finally delineated the boundaries of its terrain. It may be that it has simply framed a particular way of speaking. There is always scope for the extension of a concept, and a creative impulse - whether artistic or linguistic - tends towards this. Perhaps the key point to entertain here is that any taxonomy can be seen as a kind of reading and as such we should be wary of treating any in an overly prescriptive fashion.

  In this context the work of Marcel Duchamp becomes particularly interesting. A peculiarity of Duchamp's work that I would like to foreground is his unusual sensitivity to taxonomy. I would go so far so to argue that Duchamp was the first artist to begin making and exhibiting work that traded upon the playful production of taxonomies. This is evident both in his strategy of signing/annotating objects and also in the exhibiting of found objects within the formal institutional space of a gallery. The former practice embeds an object within a new network of classification. This is rather akin to an artist taking fragments of found information and placing them within a database of their devising. Duchamp's annotations complicate the reading of his objects in the same way that the field names of a database might serve to complicate the reading of informational fragments. Conversely, the placing of non-art objects in a gallery space toys with the existing taxonomical structure of the gallery and the related forms of social behaviour. This is perhaps analogous to the practice of inserting mischievous or anomalous information into an existing data structure. Such information might challenge or compromise a database's formal structure. In this sense I believe that we might make a case for citing Duchamp as the first of the information artists. The main point that I want to make with respect to Duchamp is that his strategies are essentially playful with respect to taxonomy. That is to say, there is a sense in which Duchamp uses taxonomic practices against themselves. In using categorisation to broaden the possibilities for reading of an object it becomes a tool for complicating our perceptions of the world: In Duchamp's hands taxonomy becomes a tool for multiplying complexity and difference.

When something is in a constant state of process then it requires a highly flexible form of classification. I suggested at the outset that problems of classification are of particular pertinence to digital media and that this is largely a consequence of the dual aspect of a computing device. An examination of one of these aspects - the notion of process - draws our attention to a very different kind of taxonomist and taxonomy. Alan Turing is probably the strongest representative of this position. The concept of the Universal Turing Machine[4] (essentially a computer with the additional allowance of an infinite memory tape – though Turning's machine is abstract, algorithmic and non-physical) was primarily concerned with simulation/replication of functional systems. Turing had attempted to create the minimal specification for a procedure that would be able to simulate any other kind of mechanistic system. In the strong sense this can be read as his having outlined a specification for a machine that could be programmed to mimic the behaviour of anything that is mechanistic. In a weaker sense – and in accordance with the thrust of this article - Turing can be said to have outlined a specification for a machine that could be programmed to mimic anything that could be read as being mechanistic. The upshot of this is that in reality the Universal Turing Machine is capable of simulating the behaviour of anything that fits a taxonomy of formal mechanism. The debates concerning the possibility of machine intelligence and more recently the value of research and experimentation concerning artificial life are testament to the way in which a taxonomy occupies an ambiguous area between mirroring and projecting states of affairs. With this in mind, there is an important sense in which we can see Turing and those who he has influenced as being reductive taxonomists. That is to say Turing's concept of simulation employs a taxonomy of phenomena that attempts to reduce the complex to the simple, In this sense it is fundamentally a process of abstraction/subtraction. As a use of taxonomy, then, it represents the antithesis of Duchamp's.

It is this plasticity of computers in the context of simulation that complicates the definition of computer-mediated artwork. The commonality between any two pieces of work may be in no way obvious given the diversity and flexibility of output (or surface) that the Turing process facilitates. The same flexibility plays havoc with the concept of 'style'. It is the sense in which computer mediated work is based upon a highly amorphous machine (or material) that it becomes so difficult to categorise. It is when we turn our eyes towards the surface that these problems ensue. A computer mediated durational piece of work might adopt a form that we recognise from another tradition - film for instance – but it might just as easily break radically with expectation or tradition. A formal simulation is effectively capable of becoming something entirely different at a specific point in its execution. Likewise, drawing the boundary between a piece of work and its setting is problematic when we consider that a system may respond to external sensor system or data feeds. The network aspect of computing devices is of particular importance in the sense that it is antagonistic to the atomistic 'white cube' mode of exhibition. That is to say, if a piece of work is networked – either to a series of sensors or to another machine positioned elsewhere in the space - then there is an important sense in which the architecture of the space changes. With a network we take a step towards collapsing a space. We are familiar with this in relation to our working practices – it is embraced as a means of facilitating speeded up production and collapsing the space of an office – but we have yet to make the same leap within the context of exhibition.

I'd like to end with some comments about a degree of fit between Turing and Duchamp. There is a sense in which their differences can be exaggerated in a way that obscures points of similarity. It is worth considering that there is a sense in which their interest in taxonomy and the workings of systems – though they be expressed in different ways and directed towards different ends - brings them into similar territory. With this in mind I think that it might be interesting to look at some of the tendencies in Duchamp's work that might have positively disposed if not towards Turing then at least towards a Turing Machine.

Duchamp was one of the earliest exhibitors of machines in an artistic context – albeit the case that he was sometimes exhibiting mechanistic devices with some form of annotation (whether literal or symbolic). Similarly, allusion to mechanism features prominently as a form of symbolism throughout his work. It has to be said that despite his use of the machine as material and motif his stance towards technology seemed to be for the most part distopic. Naturally he was highly critical of machines being used as any kind of creative agency in the production of artwork. Automated processes typically feature as a force of dehumanisation in the symbolic aspects of his work. Despite this hostility to machinic art (art produced by machines), Duchamp was very intrigued by the artistic possibilities presented by the manipulation of light. In 'Beyond Modern Sculpture'[5] Jack Burnham suggests that Duchamp saw the use of light as a medium as being a probable direction for the future of art. This is particularly interesting as Duchamp was speculating about the manipulation of light from a largely optical/lense based perspective in a time when computers and light sources had yet to become so intimately connected.. It is perhaps interesting to note that a contemporary Turing Machine is the most flexible means of manipulating/coordinating light (be it by way of a screen, projector, laser or any of the plethora of automated lighting devices). Furthermore, the aspects of the mechanism that most disturbed Duchamp appear to be the repetitive drudgery associated with the physical/mechanical mechanism. This seems also to be an aspect of the machinic process that is specifically associated with an earlier technological form.

My final comments concerning Duchamp are related to his interest in randomness and unpredictability. One of the tendencies of computer-based artwork that may have served as a point of intrigue is the apparent fracturing of otherwise deterministic processes.  Lev Manovich speaks about related processes as a kind of 'high level automation'[6]. Rather as Duchamp's taxonomic strategies served to open up or loosen systems that were otherwise formal and closed, much of his physical artwork toyed with elements of instability and uncertainty (in terms of both reading and physical process). I believe that this deserves to be foregrounded as it more than hints at an interest in the notion of the anomaly – which in turn implies an interest in formal systems.

The field of digital arts is distinctive in that there is a sense in which it needs both Duchamps and Turings. In addition to this it needs some means by which they might converse without doing damage to one another's epistemology or ontology. Despite differences in general approach and outlook there is often a shared concern with taxonomies and systems – be they playful or reductive. If my suspicions are correct then it is unlikely that Turing would have taken to Duchamp – though a likeness in territory may have disposed him towards his work. Likewise, Duchamp may have had little time for Turing – but he may have been more excited by one of his machines.

Spencer Roberts, 2004