Artistic Rendering

Artists use external systems, anything from simple tools to complex human interactions, to bring their ideas into the world. When using a system we provide inputs, turn a crank of some kind, look at what comes out the other end, and then try to intuit its actual behavior. When the behavior is desirable we want to control it to some purpose.  In the Arts this purpose is usually to produce -- render -- a tangible product such as a print, film scene, or musical interlude.

At it's Latin roots the word render means to give or put. This made its way into English meaning as:

1. Transmit to another (render a verdict);
2. Create an image (render the model);
3. Cover a surface (render stucco);
4. Extract by melting (render lard).

The purpose of most (all?) human activity is to transmit ideas or goods and services, and artistic activity brings the first definition to the fore. In the Media Arts the second definition applies directly to the activity of using a computer to make an end product. However, I submit that this usage, more often than not, devolves to one or both of the last two senses, either covering, resulting in a thin facade, or melting, resulting in schmaltz.

There was an Alternative

In the late 1960s Jack Burnham hypothesized something he called Systems Art (for a start see my Cybernetic Serendipity entry). His ideas were quickly absorbed and diluted in the Conceptual Art wave leaving us with two divergent online definitions of Systems Art.

Wikipedia says:

In systems art the concept and ideas of process related systems and systems theory [c.f Cybernetics] are involved in the work [and] take precedence over traditional aesthetic object related and material concerns.

ArtNet.com (now a defunct link, but from the Grove Dictionary of Art via the Wayback Machine) says:

[Systems art is a] Term loosely applied to art produced by means of a systematic or highly organized approach to an image or concept.

The latter refers to what are called Generative Systems, into which the products of artistic rendering usually fall. The system produces a product and we don't really much care what happens inside the black box. It's all about the Form of what is produced.

The former is more interested in the system's behavior and provenance. What is interesting is the Function of the system itself. In the 1960s interesting behavior was not easy to produce, but in the 2010s we have the capability to create much more complex systems. Systems which have lives of their own.

An Aesthetics of Function rather than Form

With an Aesthetics of Function we consider the qualities of the system's behavior, and by extension into second-order-cybernetics, the quality of the relationship of the observer to the observed -- the artist-viewer's relationship with the art-system.

While there are systems which have no inputs, the entire Universe might be an example, we are more interested in those with the structure described above, where by providing appropriate inputs we can mold the outputs in certain ways:

Input - Process - Output

For artificial systems this requires input sensors, internal cross connections, and output actuators. This usually invokes Art and Technology, by which we usually mean Electronic Technology, which in this century usually means Computer Technology. And this is the medium at the heart of "interactive" New Media art.

I put "interactive" in quotes because it is usually a mis-applied description. Most "interactive" art is better described, at a somewhat lower level of function, as "responsive". In order to clarify this I propose the following continuum of system function and behavior.

Responsive:

A doorbell responds by ringing when we push the button. The old fashioned ding-dong type might even be described as interactive because the button push causes the ding and the release causes the dong, giving us some modicum of control over the proceedings. But the same inputs always produce the same outputs.

Interactive:

Learning to use a tool or play a musical instrument is interactive in the sense that we have to experiment to find the capabilities and interfaces that allow us to use the system. While the ding-dong-bell may fit this description, its state-space -- the number of different conditions it might be in -- is very small and easily explored.  Playing a piano requires the manipulation of a much larger set of states with varying inputs and outputs. The user and the system form a feedback loop which ultimately produces the output, but only the user changes his/her behavior.

Adaptive:

If the system changes its behavior as we use it -- generally we like it better when the changes benefit our intentions but it could also be an obstinate SOB -- it is adapting. To do this it needs a large state-space which changes over time and this requires memory. Pushing the interactive tool analogy rather harder than it should be, tuning a guitar while learning to play it might be considered to be adaptive on the part of the guitar. The unfortunate thing is that there are very few examples of adaptive behavior in the arts. Some video games or those films in which one can vote for various outcomes might fit the bill.

Collaborative:

If both the system and the user adapt to each other in order to render a result we have the start of collaboration.  I know of no complete examples of this in the artificial art-world.

This set of way-points is ordered by increasing autonomy and independence of control. Responsive systems have very little control over their behavior whereas a collaborative system ideally shares control equally.  Another way to put it is that they are increasingly lifelike. Or Artificial Life Like.

The Musical Analog

Musical production provides better examples of my categories, and in general, has made more progress with both humans and their instruments. Gordon Mumma's Hornpipe (1967), for waldhorn, valvehorn & cybersonics, is an early example of an interactive and adaptive system of performer, instrument, and space.  George Lewis's player algorithms, e.g., "Rainbow Family" (1984), for soloists with multiple interactive computer systems, which he described in such terms as (from my memory of a talk he gave at Mills College in October, 1984), "This guy is sort of a backup player where this other guy really likes to play lead," ventured into the collaborative.

In music we might consider a symphony orchestra to be, ideally, responsive to exactly the requirements of the score and conductor. In reality of course the conductor and players interact and adapt to each other. But in the extreme, consider Stockhausen's use of computers to render his compositions such that he had complete (well, almost) control over all the parameters of pitch, timbre, and time.

A string quartet provides a better example of interaction. Traditionally there is a detailed score under which each player has some autonomy of interpretation, and the ensemble as a whole must interact to produce the result.

A group composition, for instance a popular band developing a song, may cover the ground from interaction to collaboration but probably displays more of the features of adaptation. Each player makes a 'riff' off of the suggested material and all of these inputs are adapted to each other resulting in a more-or-less fixed end product.

A free jazz ensemble -- when they actually listen to each other -- is an example of a collaborative system. Each player makes an equal contribution while interacting with and adapting to the other players.

Conclusion

Learning to use any system is interactive in the sense that we need to probe it and learn from its responses. In this process we are adaptive, so the system as a whole exhibits that property.  However the external system may not learn anything about us. When the system passively adapts in some form we have a master-slave relationship, but also the beginning of a dialog. When the system experiments with us -- hopefully benignly -- and we adapt in turn, then we have at last the beginnings of a true collaboration.

This should be the goal of Artificial Life in the Arts.