RA3 — Generative Art Theory by Philip Galanter

1. Generative ‘Art’?

Prerequisite:

Galanter in his overarching preface, states that there are some pre requisites for an Artwork to constitute ‘Generative’. On a superficial level, it involves Art created by ‘Non-Human’ Systems as opposed to Art created by Humans. It is important to note that ‘Non-Human Systems’ do not necessarily mean the use of technology but rather some basic underlying algorithm (many of which are numerical and analogue). Rather, Generative Art in its ‘primitive’ beginnings, paved the way for Computers.

The Jacquard Loom Machine can be considered one of the earliest Generative System where loom manufacturing was automated using cards with holes punched in. Islamic Patterns that followed geometric rules and mathematical algorithms were also early explorations of Generativity. The precision in the way a single modular pattern was laid out and then repeated, lead to a wide variation of designs and motifs.

Sheikh Lotfollah Mosque

Source: https://mymodernmet.com/islamic-architecture/

Autonomy

It is emphasised by the author that most contemporary is not to be associated with generative works, as the artist rarely relinquishes control of the work. Galanter then further argues that there are issues that exclusively impact generative art as a practice. And that generative art is a methodology to making art rather than a subset of ‘art’ in itself which he explains is ambiguous due to the many nuances surrounding what constitutes art — for which he raises the provocative question “if it is art, is it ‘good’ art?

Hence there is a need to come up with a broader schema as to what Generative Art is as a theory rather than prescribing it with a strict and stringent definition. It is important to note however that the  fundamental manner in which generative art operates is very strict and has to satisfy the following —

1) there must exist a designed system (within the work)  with some sort of functional non-human operating system involved

2) the choices and decisions being made by the system has to be specific.

2. Using Randomness Effectively (‘Disorder’)

When approaching the conception of a Generative Work, one should utilise randomness as a complement but not as the fundamental function/operative. Randomness, Chance produce ‘disorder’ (which is key to achieving ‘effective complexity’ which will be explained in the later portion of this essay,) but are meaningless without any sort of contrasting order or framework for the system in which it is being used.

For example in Noll’s Gaussian Quadratic, horizontal positions are visualised using a quadratic function while the vertical positions are visualised using Gaussian distribution of random numbers. The Gaussian distribution in itself is some sort of skeletal system whose functional output differs depending on its input numbers. The numbers inserted here by randomness is ‘random’ yet when visualised with a highly familiar mathematical graphical function, gives the work some sort of arbitrary visualisation to compare to, allowing viewers to appreciate the deviations from the piece relative to one of normalcy.

‘Normal Gaussian Distribution’

‘Noll’s Gaussian Distribution (taken from PDF)’

A completely random function in contrast would be unintelligible and exist pointlessly — though generative, if the underlying system is one that lacks any sort of cultural, emotional or scientific association we can intuitively relate to or identify with, then the generative results produced by that system will not constitute ‘art’ but rather just gibberish. We hence have to be mindful that ‘Generative Art’ has to satisfy both ‘Generative’ and ‘Art’.

This notion of ‘disorder’ brings us back to Galanter’s argument that the peak complexity occurs when there is a mix of both order & disorder as opposed to Shannon’s notion that there is an indefinite positive correlation between the increase in disorder and complexity. As illustrated by Gallant, in his analogy of pixels which by Shannon’s theory would constitute ‘complex’, according to him are modular and easily identifiable in their discrete elements (making them the opposite of ‘complex). Her theory is one that is empirical and personally I feel when applied to the nuances of human condition and cognition does not apply as ‘accurately’ or intuitively.

While understanding the ‘machine system’ and its unique modus operandi is key, as artists we have to also be highly aware of how this notion of ‘machine intelligence’ is perceived and processed by the human mind — this then beckons me to recall the idea of the ‘Uncanny’ discussed in the previous assignment where ‘peak uncanny’ is at a specific point that is half recognisable yet half foreign. Similarly, peak complexity is achieved when the Generative Piece is ordered enough conceptually/algorithmically to be processed, yet its structure and visualisation can be highly disordered, seducing the mind to put in effort in consuming the work.

We do not associate disorder with complexity if we are able to make sense of break down the disorder in an orderly fashion. Rather the disorder is manifested in some sort of superficial sense — sound, visual. But if we are able to identify and associate this ‘disorder’, that in itself prescribes some sort of psychological order over the work — we know that white noise are just pixels, in contrast to a long strict of characters in sentences that make no sense. We naturally will be inclined due to conditioning to identify some sort of pattern or word formations, requiring innate effort to process- this in itself then makes the process of consuming the work complex. Though letters are simply digits the same way pixels are individual elements.

Hence as an artist we have to be highly aware of the ways in which we as humans perceive and process the entity of the work as a whole rather than just extrapolating its potential effectiveness by just scoping into one aspect of it – it can blindside us from achieving a much more effective complexity.

3. Complex Systems as Framework for Generativity:

A highly complex system that operates as a whole with multiple processes that are able to synergise and function within itself without any external intervention. Non Linear. (Small continuous changes resulting in macro level phase changes)

Complexity Science 

Complexity Science as a bottom up process instead of top down reductionism — the emergent whole is greater than the simple summation of the same of its parts. This idea of the ‘whole’ being greater than the sum is not exclusive to Generative Art and I personally believe can be seen by both Bodies in the 2D and 3D planes

2D plane — individual elements are placed precisely, relative to each other on a 2D plane, keeping in mind ‘invisible’ yet important concepts of Space and Balance — these are not measured elements that form the summation of the elements yet they are part of the whole.

Coca Cola Poster

 

3D plane — negative space etc (architectural bodies) cannot be simply deconstructed using the positive elements because the flow of space and the use of negative space is as equally significant in the function of the final body.

Tianjin Ecocity Ecology and Planning Museum

Non Linearity of Chaotic Systems leads to amplification of small differences  —’the Butterfly Effect’. A Complex Adaptive System is akin to the process of Evolution and Natural Selection—Adaptation over time in reaction to the environment and other external factors causing certain genetic mutations to be favoured and others to be ‘phased out’ — speciation

Domesticated breeding of Foxes as a Generative System: 

When internalising Galanter’s argument that Complexity Science forms a strong foundation for Generative Methodology, I immediately recalled an experiment detailed in one of my favourite writings — Richard Dawkins’ ‘The Greatest Show On Earth’. 

This phenomena can be observed in the Experiment Conducted by Russian Geneticist Dimitri Belyaev in the 1950s (detailed in Richard Dawkins’ ‘The Greatest Show On Earth’). This phenomena is one that has both order (external restrictions imposed) and disorder (innate DNA or behaviour of the animal). A process that happened over multiple generations, over time it’s outcome becomes less abstract and more crystallised – the difference/separation in species becomes increasingly apparent and we end up with two or more divergent breeds from 1 ‘System’. Belyaev as the ‘Artist’ of this Generative Process, utilised the Fox’s innate flight or fight response to select Fox’s that were more receptive and calm to the intrusion of his hand. This is an example of how one harnesses the system’s idiosyncracies without directly manipulating it.

Short Video on the Experiment

What is interesting to note here, in Belyaev’s Breeding Process, in terms of unpredictability was that certain behavioural patterns were linked: as one line of his breed became more behaviourally domesticated, they also started to change physically. They were linked (some genetic mutations or some genes were linked and influenced each other) they did not fully occur individually as scientists previously hypothesised :

“These dog-like features were side-effects. Belyaev and his team did not deliberately breed for them, only for tameness. Those other dog-like characteristics seemingly rode on the evolutionary coat-tails of the genes for tameness” — The Greatest Show on Earth, Dawkins.

Breeding not just isolated to Foxes but to that of Dogs for a wide variety of purposes -—Daschunds hunting Badgers, Borzoi for Guarding, Whippets for Racing etc. These were all possible due to the ‘Artists’ (in this context Breeders and Scientists) who observed the system ‘DNA/Genetic Mutation’ and over a period of time favoured certain unpredictabilities to form an ordered production line of the same ‘unpredictabilities’. As this process became more crystallised, other nuanced attributes became increasingly apparent (further inherent mutations that were carried forth by the more ‘major’ mutations’).

Hence observation of the system we are working with is key. We need to make sense of the highly profound Chaos of the Chaotic System and harness it in a way that allows us to establish intuitive order (without forcing it directly). This then utilises the core of the system and makes the operative of the system itself a component of the Artwork.

This to me is ideal in Generative Art, the outcome should not be so far removed from the initial starting point that we see no correlation between the system’s underlying function and it’s output. Rather we want to be able to see a divergent set of outputs that are fascinating yet unique to a specific system.

4. Complexity Science as the trailblazer for Generative Art:

New Models of Complexity science that form the basis for Contemporary Generative Art – Fractals and L systems. These models are somewhat ordered systems that are still able to simulate certain processes in Nature such as the branching of plants.

It is significant to note that nature in itself to some extent is ‘ordered’ in the way the golden ratio or Fibonacci sequence ratios are manifested. Hence order does not correlate to a lack of complexity. We have to be mindful of the type of ordered system that is being implemented and if the ordered system is ‘diverse’ enough to produce different and divergent results depending on the information being applied to it.

The process may undergo a very ordered procedure but if the function produces a unique outcome each time, depending on the type of input, then ‘order’ in this context still constitutes a valid and intelligible generative exploration. Mindless Order that reproduces the same mundane result repeatedly, would be an example of a non generative ordered framework/system.

5. Problems

Authorship

Galanter discusses many problems involving the methodology and approach of ‘Generative Art’. Amongst which the problem of ‘Authorship’ resonated with my own dilemma when approaching or designing a Generative Work. Is the Production of Meaning to be borne by Artist, Machine or Viewer? Is it truly possible to strictly confer authorship to only one of the entities or is the role of Generative Art as suggested by Galanter, to destabilise the fundamental idea of authorship?

The importance is upholding Ambiguity, as long as there is some sort of shared authorship instead of it being restricted to just one entity, the outcome and interaction with the system automatically becomes non predictable and ambiguous. The control given to the reader to some extent allows the work to be manipulated depending on their choice while at the same time, the readers are limited by the options offered to them by the computer (which ultimately is further limited by the parameters we as designers set for it to function within).

Personally, I would then view authorship as a sort of collective and feedback loop process in itself rather than the structuralist theory put forth by. Authorship is in a constant and non definite feedback loop between all three entities and each entity needs some aspect of authorship autonomy to sustain the whole.

 

Leave a Reply