Reading Assignment 3 | System Stories and Model Worlds: A critical approach to generative art by Mitchell Whitelaw


Text Selected: System Stories and Model Worlds: A critical approach to generative art by Mitchell Whitelaw


A Critical Approach to Generative Art

Mitchell Whitelaw discusses the true attraction of Generative Art, and what it represents other than the technicality it is composed of. He suggests that there are two fundamental distinctions in software generative art; “Software formalism”, where it is prospective and exploratory, and “Software culturalism”, where it is local, situated, concrete and interventionist. Software formalism focuses on the “generativity of code, typically visually abstract, focuses on the processual relations of coding and aesthetic output” whereas software culturalism focuses on the “cultural text, critical, discursive and reflexive, deconstructing the “mind control” techniques of software”. In other words, Whitelaw separates generative art into the code meant to create the aesthetics of an art piece and the narrative behind that has deeper significance in terms of providing meaning to the art.

Whitelaw mentions that generative art is usually fundamentally created with complex systems, unlike the Paradigm of reduction which introduced abstraction to the visual arts. According to Lev Manovich, it oscillates between “order and disorder- always vulnerable ready to change with a single click of the user.” Over here, Whitelaw describes the inconsistency and unpredictability of generative art, compared to the pre-determined aesthetics of visual arts. I resonate with his claim that generative art can be a guess-and-check game, and that is what makes it so different and unique from the usual visual arts. It is an art that can only be fully appreciated by watching it grow during a set period. However, it makes me wonder if the art is considered to have died after the growth is completed.

Whitelaw mentions that “the systems, not their outputs or residues, are the core of the work”. The important thing of generative art is not the results of the art piece, but the process of the creation and the system that produces it. It is also brought up that code is the language-specific text that implements the abstract, formal structure that is called system. Even though code might seem very technical due to the numbers and calculations that are involved, but it is the formula that creates a system that gives birth to a whole world of unexpected possibilities through concrete collections of objects, relations, actions and processes. This is why the process itself is the generative art, not the end results.

Software generative art can be considered “agents” in an “environment”, where subjects in an artwork act as subjects that react with each other in a formally constructed “environment” which provided them the circumstance for reaction. The “environment” here refers to the systems that produces the outcome. In a way, we can also refer to these systems as a form of narration. This is because they structure the artwork, acting as a sort of umbrella framework which houses its growth. I enjoy Whitelaw’s interpretation of generative art as a story with a structure here because it gives greater depth and less complication in how software code works when we view it as a storybook. As an artist, this is important motivation.

TIERRA BY THOMAS RAY - ADA | Archive of Digital Art

Tierra by Thomas Ray

Therefore, I agree with Whitelaw when he states that “systems are literally texts, involving specific figurations, relations, decisions, values and ideologies”. If we see systems as storybooks, and storybooks comprises text, then we can say that the story in the storybooks are the specific figurations. Tom Ray’s Tierra system is brought in, where Ray’s biological and theological analogies are spelled out in the narration and the construction of the visualisation.

John Holland, Echo and agent-based models in biology – Biosystems Analytics

John Holland’s ECHO

Whitelaw also brings up a-life science, which is a concept brought up by Stefan Helmreich and Katherine Hayles. A reference he makes is to John Holland’s Echo, a platform for creating agent-based a-life simulations; Stefan Helmreich analyses it based on conversations with a programmer and inspection of the code; the observations come as much from the defined formal structures of the software, as they do from the discourse around those structures. A-life science is seen as a deconstructive approach, its systems being fundamentally narrative in their operation. They are the embodiment of subjectivity, gender, family and theology, decoded in part from the discourse around the software system. This is just like Whitelaw’s interpretation about how systems are like texts; we can read systems as stories. However, these stories are quite subtle, since the narration is embedded in the code that is hidden from public view. The “system story” translates or narrates the processual structure, ontology, entities and relations in a software system.

Building Blocks GIF - Building Blocks Minecraft - Discover & Share GIFs

Minecraft Building Blocks

Hence, results and final appearances mean nothing without the systems that created them. These systems tell us a story or narrative. Just like Brad Borevitz says, simulations are merely to abstract them and play with them according to the demands of an aesthetic production. The work is entirely shaped by the construction of its underlying system, its configuration of entities and relations. Generative art is essentially about creating a whole artificial world like Minecraft and showing people how it works. With Minecraft as reference, it is not the giant Swedish meatball nor the mansion you have built in the game that matters; it is the process of gathering materials and resources to build the meatball or mansion that is to be appreciated as generative art.

Whitelaw proceeds to talk about combinatorics or the playing out of permutations to make generative art, which is the opposite of Generative art’s complex systems. This is the questioning of the shape of the system, the question of interpreting, or responding to the configuration of entities and relationship in the code. The results from these are visually complex, but the underlying system is quite simple. These forms and patterns generated were deterministic, in the sense that they were sure to provide a certain sort of result. However, they were impossible to predict exactly despite having fixed systems. This unpredictability is what gives rise to more stories.

Casey Reas | { Software } Structures (2004) | Artsy

Casey Reas’ Software {Structures}

An example of this is Casey Reas’ Software {Structures}: Reas’ #002 and #003, Tarbell’s #003A and #003B, and Ngan’s #003B. In this project the artist’s focus was reflexive and processual: considering the “natural language” specification of a structure, and its varied implementation. Removed from that context, however, we are faced once again with the shape of the system, and the question of interpreting, or responding to that configuration of entities and relations. The model worlds in these instances are pure machines, clockwork constellations.

Physical simulation is then brought in the text to compare against software art, where the systems were usually digital in effect. Physical simulations were pragmatic and effective, creating complex, dynamic interactions between elements at low computational cost. For physical simulations, there is immediate physical resonance, inherently narrative and metaphorical. Mark Napier and Scott Snibbe’s works in the CODeDOC project were brought in for comparison. This is to prove that generative art does not necessarily succeed only in the digital realm; it is also important in the physical sphere as long as there is a process and interaction between different elements of the art piece.

Further in chapter 3 of the text, Whitelaw talks about how results are “organic” and “cellular”, because entities get to “experience” and “choose” which way to go at intersections. This is a symbolism of progression of a narrative. Generative art makes use of multi-agent systems, where entities are explicitly defined and visualised to encode an ontology, a structure of entities and relations, which must be read as the core of the work. These entities have static properties and behaviour over time, which means that history is often absent. “History” will only exist when these entities are introduced to each other to create a catalyst which will elicit the gears of a system to move. In a sense, these can be termed as agent-based systems, where it presents specific attributes and modes of being and relationships between individuals, groups, and the environment.

On this kind of dependent system, where agents meet the environment to run the system, power relations never emerge because there is no “leader” in the system. I think that this is key in generative art because the idea of unpredictability and randomness is that we cannot figure out what will happen, hence having no “leader” is important. There should not be a leading feature that signals the next happenstance, or it will ruin the whole idea of generative art being a work that is special because of how it happens.

Chapter 4 talks about generative software art, as Whitelaw explains, as a potential platform for telling system stories that are more sophisticated, critical, or experimental. It has potential to provide a visual of the systems we live in. In this case, there is no agents-meet-agents story. There is a disconnection between agent and the environment, where agents interact with each other but have no functional impact on their world. These agents lose their individuality when they intersect with each other. In other words, the agents do not have links to each other until they are able to interact with each other, which will spur on a catalyst for the system to go into operation, which is why interaction is more important than the by-product or the pre-product of the experimental art. is a database for Casey REAS

Casey Reas’ Microimage is a database for Casey REAS

Casey Reas’ Tissue

Whitelaw brings up Casey Reas’ works in Tissue, Microimage, Articulate, TI and Cells. Tissue and Microimage in particular begin to develop the homogenous swarm, creating distinct “species” of agent with distinctive (but again fixed) relationships. The added complexity of the interaction within the system is revealed in the images, as tangled clouds resolve into dark loops and braids. This is the case where agents themselves are individual until they meet each other, where in this case begin to become tangled clouds that creates a new aesthetic for the artwork.

Ichitaro Masuda’s Haohao

Ichitaro Masuda’s Haohao has multiple species of agent, differentiated in size and colour, and attracted to and repelled from each other to varying (randomised) degrees. There is no agent-meets-agent story. When the forces of attraction and repulsion meets equilibrium, the agents form clusters of give or more which might orbit other groups and might change unpredictably if disrupted by an invading agent.

Towards Artificial Societies

Mauro Annunziato’s Artificial Societies

Mauro Annunziato’s Artificial Societies drawings is the final example of agents interacting with each other but have no functional impact on their world. Their character arises from a simple feature of his system in which agents’ paths are drawn into the environment, and where agents also “die” once they interact with each other. Equipped with a simple genetic/ evolutionary mechanism, the agents progressively divide their environment into isolated “habitats”, each applying a particular selection pressure to the agents within it.

Ultimately, Whitelaw wants us to understand that an environment is not the blank space we usually visualise, which can only bear meaning once there are subjects defining what the space is for. Environment in generative art is instead something that encompasses a system’s history to create results. The environment in this sense, is the system of the code.

Finally, we can decompose a system and analyse the modes of being and relation that it encodes in a generative art system. However, we have little say in how the encodings play out and how they function during the operation. Whitelaw terms this as “computational sublime”, where it is an emergent generativity; it generates art as it emerges. Moving back to the previous concept of a-life, it does not provide increasingly accurate simulations of an authorised “life”, but instead provides experimental and reflexive performances of possible lives.

Axis - Interactive Art by Golan Levin and Collaborators

Axis – Interactive Art by Golan Levin and Collaborators

The important thing about generative art is its structures and properties. These factors are the things that can be tinkered with to change a system and the way it operates, hence leading to the creation of different worlds. Generative art or abstract software art is a potential alternative of stories, and also toying with complex, dynamic systems to generate new model worlds. An example is Golan Levin’s Axis applet, which abstracts political rhetoric into a database-driven combinatoric. It draws its algorithms from the same source, extrapolating, diverting or visualising rhetorical entity and relation structures. This is the proof that we should reconsider the difference between critical, reflexive cultural software art, versus the utilitarian, un-reflexive, result-oriented generative art.

Overall, I really like the idea Whitelaw establishes in his research, where software code is seen as narrative, and where the process matters more than the results. I am someone who really love reading and creating stories, and I think looking at the technicality of software coding is really draining at times. It is mathematical, dry, and formulative. It does not appeal to me as an artist because of the precision it requires. However, when we look at the code as a narrative that gives birth to endless unexpected derivatives, it does not seem as strict and static anymore. An artwork is shrouded with an air of mystery and curiosity, even if it is technical.

Furthermore, it relates to my final project because of the unpredictability of the event, despite being coded beforehand. As my project is about having unpredictable follow-up stories from options chosen by the reader to proceed the story, I think it gels very well with the idea that the code is a text that forms a narrative. In my case, there is a parallel between the software code I am using (HTML) and the story I am crafting. They are both narratives; one is a technical story that decides how the algorithm of the story will appear, and the other an actual unpredictable storyline. Ultimately, I feel strongly for the condition that narratives are the basis of generative art, where it crafts the essence of the artwork for public appreciation. It is not like normal arts where the results are what matters, be it a graphic design or a film. Generative art in interactive media is probably one of the only mediums where the process is the art piece.

Reading 2: Amplifying The Uncanny | Response


Amplifying the Uncanny

by Terence Broad, Frederic Fol Leymarie, Mick Grierson

Pushing the limits of Perfection:

The relationship between Generative Art and the uncanny is their ability to create something unexpected from a given set of data, and then generating further distortion at greater emphasis; kind of like evolution. In deepfakes, “machines [are] optimised to make representations that are more realistic, they also generate information on whether or not a generated image is fake.” However, the cool part of this is that even though these machines can be so precise, they can end up unexpectedly making the generated image less precise; it becomes distorted because the machine is constantly improving and sharpening the accuracy of any “out of place” element on an art piece or photo. As the machine does not have human limitations imposed on it, it does not know when to stop. This unpredictability of how far the machine can push the limits of distortion can be seen as a form of generative art.

Evolution into Unpredictability through “Over-learning”:

In the text, “Machine Learning” is also introduced, where it “generates patterns in data… to predict future data or other outcomes of interest”, with the mindset of “solving” a pre-defined objective. In the text, Mackenzie also reminds us that machine learning is about the knowledge-practice and not knowledge-consciousness. In other words, it is the repeated process of the same thing over and over again by the machine which provides greater accuracy and precision of a desired outcome (as they have no brain to think and be conscious about the details they are producing). This is a form of growing, compared to the usual programming and coding of specific instructions for machines to execute. I think that this is significant in explaining how vast generative art can be; when a machine “grows”, it does so by self-improvement through previous data created but when there is no “stop button”, it becomes unpredictable and beyond our physical calculations since it is also endless. The unpredictable results become an art form that we could not have foreseen.

The Elimination of “Zombie Art” through the Unknown:

However, it is mentioned that there is public perception that GAN generated artwork is also considered “zombie art”. It is said that there is a reliance on deep neural networks to produce neverending variations based on one sample of information fed into the system, but this system of construction is not meaningful. Even though it can seem mesmerising at first, it can eventually become monotonous after a while because viewers will start to feel overwhelmed by the “sublime of algorithmic productivity”. I agree with this standpoint because we have such short attention span. When something is too repetitive, we can “predict” what is generally going to happen next and end up not chasing the information further; since it is no longer an “unknown” that we are curious about. As people are usually lazy and will not want to do meaningless things which is why we leave it to the machines instead. However, if the growth is something that is unpredictable, it will not be meaningless and instead be intriguing. However, we are then left in puzzlement with regards to how to make something precise (since most digital artworks with auto-generation relies on pre-calculated equations) and yet not monotone.

What is the Fine Line for Humanity?:

I think that it is also interesting that the authors of this text explored what it meant to be at unease due to uncanniness; they defined it as the state where the fine line is just crossed between what is living and what is machine. When something does not resemble human to a large extreme, it is easy to regard it as fake and objective, but when something lies in between, we start to become confused about the identity and origin of the object and become uncertain about how we should classify the subject. It is familiar, yet unfamiliar and expected, yet unexpected. In the text, the Uncanny Valley is also mentioned, where “increasing human likeness increases feelings of familiarity up to a point, before suddenly decreasing”. As a robot’s similarity to human form increases, it is proportional to the unsettling feeling we will feel until the similarity becomes so much it is determined to be fake.

What Is the Uncanny Valley? - IEEE Spectrum

The Uncanny Valley

This makes me ponder about the threshold of human consciousness and comfort. It feels like we seek comfort from things that looks “average” or “just right” and anything too extreme on either ends of the spectrum is deemed unacceptable to our comfort zone. Is this possibly also why people are uncomfortable staring at physically handicapped people or people with eating disorders for a long period of time; because they fall outside the threshold of “normal” and what is “expected”? Can we then conclude that in generative art, things become art when they fall outside the threshold of what our expectations are? Does this mean that since different people have different expectations as we are all individuals, our perception of what is considered generative art is then altered according to that logic?

In the end, is Generative Art a matter of perception, and could it be calculated yet intriguing (not monotonous)?


Closed systems: Generative Art and Software Abstraction by Marius Watz | Reflection Essay



Closed systems: Generative Art and Software Abstraction

by Marius Watz

While generative art is usually associated with pixels, it is not a compulsory criterion. It is also not necessary to be interactive, even though it is commonly open sourced. The most important factor of Generative art is its unpredictability, and its system of “growing” from given information.

Watz states that “Forms produced by generative systems often take on a complex nature, exploiting principles of emergence to produce structures that could not be made by human hands.” I agree with this statement; to me, the beauty of generative art is how precise, random and unpredictable it can be, and this is usually associated with numbers. This is very different from creating something from scratch, such as with clay or paintings. There is a lot more calculation and precision work involved. Then again, it should not be misunderstood that generative art can only come in the form of digital work.

Watz also mentions that it is difficult to simulate organic behaviour through computation, yet at the same time it helps in the virtual simulation due to the computer’s ability to generate and replicate the same sequence of code over and over again in an array without manual input. In this sense I think this is interesting because while it sounds like generative art is a double-edged sword with irony to top it off; it can be so precise but unnatural if you are not precise enough.

To add on, without the existence of computation, people would not even be open to the idea of generative art, as it is too tedious to execute while incorporating subtle changes to signify the growth of the art piece.

A common conception brought up is how the birth of generative art signifies the death of interactive art, as it in a way dampens the experience of open sourced art because the art can generate by itself when it is computational.

However, I feel that this conception is a misunderstanding because generative art is not necessarily digital, and even if it is digital it can also be open sourced. There are just many different types of generative art that is not limited to automated digitalization.

Ultimately generative art, while being heavily linked to software and digitalization, it is not about what the computer can do. Generative art is more about how real-time and self-contained art is still a form of art even if it is digital. Digitalization does not represent a loss in art; it is an addition to the scope and variety of how art can be portrayed. We should ultimately not see art done with software as an evil but as a different art form.

Generative Art Example: Smoke Water Fire 2016


Smoke Water Fire (2016)

by Mark J. Stock

Smoke Water Fire is a digital video that represents a virtual simulation of fluid flow. This piece can also be viewed in VR. It is rendered in 360-3D format, and rendered using Radiance,  synthetic lighting simulation software. This artwork highlights the dynamics of fluid flow and its universality. With no name plastered on the visual simulation and its changes, it gives us freedom to decide how we perceive fluid flow in this virtual simulation. The conventional perception is stripped, and we are given the autonomy to understand how fluidity actually works- through dynamic, ever-changing flow.

This artwork is interesting to me because it plays around with the perception of depth, as well as generative and degenerative elements in its digital simulation. It is a generative piece because we cannot really predict where the next change of motion of the “fluid” will take place at. Simply by subtracting or adding elements to its code, a fairly accurate depiction of fluid flow can be created and I think this is pretty cool. Just because we are dealing with numbers and code does not mean that we cannot create an actual visual representation of something very real and natural to us. Additionally, I think that it is interesting that that is an irony that an accurate depiction of something so natural is done using something so calculated (albeit random).

I feel that it relates to my own creative interests because I am someone who likes to deal with surprises and unpredictability when it comes to creation. I also like telling stories. I know they contradict each other because stories usually follow a linear storyline whereas unpredictability does not exactly fall into that line of thought. However, having an unpredictable motion that gels a whole story together in the most unexpected but pleasant way is something I like to work with, which I also feel is present in this work. In a way, this art work can be depicted as the story of fluid; where it tells us the journey of how a fluid develops and moves through its existence.