Part one. Creative Coding
“The machine makes the music, but I created the machine... I don’t know where responsibility lies in that situation.
Autechre’s Sean Booth, in interview, 2010
Chapter 1. Generative Art: In Theory and Practice
On first appraisal, the question “What is generative art?” may seem simple. But, as is the nature of all things generative, even this definition has an emergent complexity.
The elucidation most often cited in recent years is attributed to Philip Galanter, artist and professor at Texas A&M University, from his 2003 paper “What Is Generative Art? Complexity Theory as a Context for Art Theory”:[1] “Generative art refers to any art practice where the artist uses a system, such as a set of natural language rules, a computer program, a machine, or other procedural invention, which is set into motion with some degree of autonomy contributing to or resulting in a completed work of art.”
1 You can read Prof. Galanter’s paper in full at www.philipgalanter.com/downloads/ga2003_paper.pdf.
Although this is accurate and descriptive—and a long sentence with all the right words—a single phrase like this isn’t enough. I don’t think it quite captures the essence of generative art (GenArt), which is much more nebulous. In my mind, GenArt is just another byproduct of the eternal titanic battle between the forces of chaos and order trying to work out their natural harmony, as expressed in a ballet of light and pixels. But flowery crap like that isn’t going to get us anywhere either We have to be careful treading around this topic, because we want at all costs to avoid trying to define “What is art?” which is an argument best left alone. The concept of art can be so fragile and fuzzy that if we were to prod it too much, it would evaporate. So, instead of trying to carve up a subject that doesn’t want to be dissected in search of a pithy description, in the section that follows I’ll take a more delicate and obtuse approach. We’ll begin by examining what generative art isn’t.
1.1. Not your father’s art form
With more traditional art forms—sculpture, painting, or film, for example—an artist uses tools to fashion materials into a finished work. This is clearly doing it the hard way. With generative art, the autonomous system does all the heavy lifting; the artist only provides the instructions to the system and the initial conditions.
The artist’s role in the production process may be closer to that of a curator than a creator. You create a system, model it, nurture it, and refine it, but ultimately your ownership of the work produced may be no more than a parent’s pride in the work of their offspring.
This is hideously unfair, of course. We shouldn’t underestimate the human role in the collaboration. In addition to the programming, the human contributes one other important skill: aesthetic judgment. It’s feasible for computers to develop a sense of aesthetics (plenty of work toward this aim has been done within the field of evolutionary systems[2]), but it will never be the best division of labor in a human-machine creative partnership. If we need to calculate pi to a million digits, it would be a misappropriation of resources to set a human brain to this task. Similarly, it’s probably best not to leave it to machines to decide what’s pretty and what isn’t.
2 Richard Dawkins’ book The Blind Watchmaker (1986) is a good, accessible introduction to the topic of genetic algorithms and their application.
Although GenArt is almost always abstract in nature, it can’t be defined by the style of the work. The common factor of generative artworks is the methodology of its production, not the style of the end result.
Figure 1.1, for example, is an abstract work, and it’s a monochrome work, but we can only say that it’s a generative work if it happens to fit my claims as to how it was created. This particular arrangement of pixels may have been achieved in Adobe Illustrator (not by my cloddish paws, I would hasten to add); by photography; or by pencil, paper, and scanner; and in all cases, it would still be a monochrome abstract. It may also still be generative, depending on the way each of these tools is used, but this isn’t a given.
Figure 1.1. Tube Clock (2009)
The tools used aren’t the defining factor; it’s the way they’re used that provides the commonality. In this book, the programming language, specifically the Processing language, is the chosen tool. But that doesn’t mean everything you create with that tool is generative. Programming languages are just ways of making computers do as they’re told; there is nothing inherently generative about following orders.
To be able to call a methodology generative, our first hard-and-fast rule needs to be that autonomy must be involved. The artist creates ground rules and formulae, usually including random or semi-random elements, and then kicks off an autonomous process to create the artwork. The system can’t be entirely under the control of the artist, or the only generative element is the artist herself. The second hard-and-fast rule therefore is there must be a degree of unpredictability. It must be possible for the artist to be as surprised by the outcome as anyone else.
Creating a generative artwork is always a collaboration, even if the artist works alone. Part-authorship of any generative work must belong in part to the mechanisms the artist uses: the system that generates it. Fortunately, anonymous autonomous systems aren’t usually bothered if their unscrupulous artistic partners decide to steal all the credit.
To retain a necessary focus, this book discusses only one tool—the programming language—as a way of producing only one range of output: visuals. But generative methods may also be used to produce music, architecture, poetry, dance, storytelling or interactive experiences, and the autonomous systems behind their creation may also be mechanical, games of chance, natural phenomena, or subconscious human behavior. We’ll touch on some of these alternative approaches later in the book.
1.2. The history of a new idea
Generative art has a history measured in decades, not long compared to other arts, which is probably why it’s still on the periphery of the art world. Art colleges across the globe are churning out tens of thousands of painters, potters, fashion designers, and graphic designers every year, but the number of practicing generative artists in the world at present could probably fit comfortably onto a single Caribbean cruise liner (which would be a lovely idea if anyone fancies arranging it). This demographic is changing fast, though. As popular computing technology accelerates, more creative people are getting theirs hands on the tools and discovering this novel art form.
The term generative art has only been in general use since the 1960s, but the concept has been with us much longer. Generative forms of music, for example, have been around since Mozart. His Musikalisches Würfelspiel (Musical Dice Game) was an early example of a generative artistic system.
The idea was to create a minuet by cutting and pasting together prewritten sections, making selections according to the roll of a die. Even with a single six-sided die, the number of possible combinations rises quickly: by 5 rolls, there are 7,776 possible combinations; and with 6 rolls, 46,656. These types of artistic parlor games became popular in the eighteenth century.
In the last century, composers such as John Cage, Karlheinz Stockhausen, and Brian Eno expanded on the idea of generative music.[3] John Cage’s 4’ 33”, his controversial note-less piece defined only by its length, takes environmental ambient sounds as its only content, meaning no two performances of the work are ever the same.
3 Even though Processing can be used to create generative music, this book focuses solely on the creation of visuals. Generative music is a huge and fascinating topic that warrants a book of its own. David Toop’s Haunted Weather might be a good place to start.
Later, Stockhausen and Eno (...