Three Reasons Why Computational Art May Be Another Kind of Software

The predominance of these three big ideas in art production is based on the view of art practice as the manifestation of skill and craft, which has evolved throughout history, and which, for digital culture, is pre-computational. The specialisation idea assumes that all individual creators worth their salt have to master and hone their skill in one particular technique. Flexibility implies that the work of art ends once the outcome is produced, that is, it has a moment at which it stops and is not changing anymore — such as a finished painting, or a site-specific work which has occurred once in a particular time and place. The authenticity of the work implies that it has one original instance, which is perceived as the ‘original’, as opposed to its widely circulated representations, which are seen as copies.

'Chronic Time Pressure', @ALagrecaAI

These three art production assumptions may not be the only ones related to computation, however, these appear front and centre if computation is to be considered the cause of a radical shift. What needs clarification is that computational art in such a situation does not only rely on software, but is itself software, and therefore, like any other digitally produced product or service, is based on the production and maintenance of its infrastructure. Let’s first clarify why computation could be seen as a turning point, and then summarise what the three assumptions above may mean to how we think about computational art practice.

Process and complexity in computational art

In the definition of economist and complexity scientist Brian Arthur, computation is a process of executing a set of instructions or operations, or an algorithm, within a system. (Arthur, 2023: 8.) Two important moments that this definition reveals are that the work becomes more of a process and that it opens up to a possibility of infinite folding and modification.

1. The process moment. As we saw previously in the discussion of Artist as a Formal System (AFS) proposition, artistic practice is present as a performative act of mediation. Computational artists are largely concerned with the topology of the epistemic infrastructure of their practice, where the work emerges as the outcome of a specific set of negotiations between the artists and the broader communities they operate. In other words, the artistic process is never fully complete, and the final piece may appear differently depending on the topological relations within its underlying infrastructure. For example, the artist may compose the prompt for the software that generates the visuals, yet, since the software in itself acts as a queer and performative technical object, its responses to the prompt may vary at every moment depending on its technical specifications and the condition of its databases, training and so forth. All of these factors may suggest that artistic discourse and technology become the key concerns of production, making the seemingly immutable AI-generated visuals closer to performance art than to traditional illustration or painting.

Artistic discourse and technology become the key concerns of production, making the seemingly immutable AI-generated visuals closer to performance art than to traditional illustration or painting.

2. The complexity moment. Computation is a technical relation based on code and is capable of creating software complexity through layering of abstractions - for example, repeatable code patterns could be encapsulated as separate components, which may in turn, have their components, and so forth. Such abstracting capability is not present in non-computational technology, which deals with the physical means of creating an output, which have spatial dimensions and other tangible properties. The physical manufacture therefore inevitably has a limit to how complex it can become given the available resources. Computational complexity is usually treated as a risk, for example, as the probability of infinite loops which can use the execution algorithm to freeze if it encounters abstraction layers that cross-reference one another. (Cf. Arthur, 2009:18.) Importantly, the fact that the infinite loop is possible means that calculative resources are unbounded by their nature. In the situation when calculation costs tend to approach zero, the result of any change is likely to be on the side of increasing complexity by creating more calculations.

In this context, computational art becomes predominantly reliant on infrastructure, rather than any specific events or objects, with the core reason being its unlimited access to computation as its key production resource Computational art is performative, because it deals with space and time limits throughout the process of its production, and is extinguished when the process is stopped.

@niceaunties, Placing blue center stage

Summary of assumptions

It is evident that the traditional artistic discourse is generally assumed to apply in the sphere of computational art. Yet, some of the assumptions, such as those related to the differences between the production of physical objects and digital products, may be misleading when we try to understand the products of computational art praxis. The comparative study of art production and production in general is possible because computational art can be understood as software, and therefore is most likely to relate to non-computational production in a fashion similar to other kinds of software. In this context, it is valuable to look at the foundational argument posed by production analysts Stefan Thomke and Donald Reinertsen, (CF. Thomke and Reinertsen, 2012.) who compare mass manufacture with software production to point out the broader implication of the differences in other spheres of software operations, including artistic production. The axes for comparison, as we saw in the beginning, would be:

1. Specialisation. Many tasks of art production are unique. In non-computational art, repeatability is made visible through artistic style. In computation, the notion of style adopts the interpretation discussed by the historian of technology Thomas Parke Hughes, as a way particular technology adapts to its environment. (Hughes in Bijker, 2012: 61.)
2. Flexibility. Project requirements constantly change. Computational work of art in this sense is a service, rather than a product because it is implicated with software, which is essentially queer and performative. (As my earlier post (https://artculturedigital.substack.com/p/software-as-queer-boundary-making) explains).
3. Authenticity. The output of computational art is information, which can reside in multiple places at the same time, due to its properties of nonrivalry and nonexludibility, pointed out by media scholar Yochai Benkler, (Benkler, 2002:35.) which calls for a re-understanding of what the authenticity of artwork means.

I promise to address these three points in one of the upcoming letters!

To conclude this week’s intervention, computational work of art belongs to the history of software and requires different categories of assessment, which cannot be unproblematically adopted from pre-computational art. Another quite important corollary is that software is ubiquitous, which means that any work of art, even those which are created without explicit use of software, can still inherit some of the software complexity that is compounded in the production of its means of production or distribution.

References

Arthur, W. Brian. 2009. The Nature of Technology: What It Is and How It Evolves. London: Free Press.

Benkler, Yochai. 2002. “Coase’s Penguin, or, Linux and ‘The Nature of the Firm.’” The Yale Law Journal 112 (3): 369. https://doi.org/10.2307/1562247.

Thomke, Stefan, and Donald Reinertsen. 2012. “Six Myths of Product Development.” Harvard Business Review 90 (5): 84–94. https://hbr.org/2012/05/six-myths-of-product-development.

Vanushkin, Alexey. 2022. “Artist as a Formal System: Towards a General Theory of Art.” &&&, 26.