presented at Artificial Life 7, Reed College, Oregon, August 2000; To be published in LEONARDO magazine
Sound: Allowing Beauty to Create Itself
by Aaron Wolf Baum, Ph.D.
EternalNovelty.com
Can a man-made computer produce something that is not "artificial"?
Modern microprocessors are by far the most complex mass-produced objects humankind has ever made. A Pentium III chip contains over 30 million individual transistors switching up to 1 billion times a second, and is capable of computations of staggering complexity. In spite of the fact that their complexity seems at least potentially comparable to that of living things, they are almost always locked into repetitive algorithms, performing to exact and rigid specifications, creating a world of crystalline, over-perfect sterility. Humans assume that their role in the computer’s world is to exert unconditional control over its behavior. More powerful computers can create more "lifelike" virtual worlds and textures (witness Toy Story and Bug’s Life), but if their behavior is restricted to what a human can provide through what is almost always a highly limited interface (keyboard and mouse data rates of a few bytes per second, compared to the hundreds of millions that can pass through the processor at the same time), this world will always be a very pale shadow of our own.
Regard two images, one of a 3D-rendered cube, the other of a flower. Why do humans find the flower, even without the color and the scent, so much more attractive? The immediate explanations - the intricate interfolding of the whorl, calyx, and corolla, the luminescent partial translucence of the petals, the subtle balance between symmetry and chance, the memory of lush color, scent, and pleasant outdoor areas - all seem to beg the question, why do we like these things, and why do we see so much in the flower? Certainly there is more to see, in a purely informational sense - the cube can be fully described by a single number, the length of one side, a few geometrical relationships, and some information about the surface color and texture. To generate a realistic-looking flower, vastly more parameters and geometric relationships would have to be specified. if the simulation could capture every nuance of the flowers appearance and behavior, the existence of the model would still presuppose the existence of the flower. Such an approach could never give us a novel sort of flower, since the process by which new sorts of flowers come about - evolution - does not exist in the computer world.
this happen is, of course, what A-Life is all about. The field of artificial life is an outgrowth of the recent advances in computing technology that make possible the simulation of virtual organisms and environments of sufficient complexity that they are experienced as having life-like qualities. In the scientific realm, this work provides an opportunity to better understand and quantify the properties of life itself. In art, the possibility of collaboration with a new sort of Nature emerges.
, achieving this requires a total rethinking of how humans and computers interact. The barriers between the world of the human and the internal world of the computer are substantial compared to the freedom with which information flows within either world. Even though better interfaces (virtual/immersive) are being developed, the sheer amount of information being processed in a typical processor increases far faster. In order to realize the full potential of this other world (which under normal circumstances is not much of a world at all, only a low-grade mirror of the "real" world), the human operator must surrender control of the world to the occupants of that world. In most AL applications, the existence of organisms is pre-supposed, and their overall characteristics are pre-determined. The freedom to evolve and adapt is given, but not the freedom to create themselves from "scratch". But this is exactly how life created itself on this planet, and how all complex structures create themselves - they cannot be planned, the best thing that one can do to create living structures - societies, communities, etc. is to see what is already there and happening and to allow it to happen. Freedom, in terms of a lack of interference from sources that the organism cannot affect itself, is a perquisite for the emergence of life.
In my work I have tried to create worlds in which anything is possible. In my work the role of the artist is to play God in the Einsteinian sense, by setting the laws of physics and then giving the resulting world the freedom to find its own patterns and paths of evolution. My first works using this approach have been in the area of audio, allowing audio textures to arise spontaneously and play off of each other, in a system in which all the aspects of the sound, in various representations, are determined by other aspects of the sound at difference times, frequencies, and representations. The mathematical expressions governing these relationships are the "laws of physics" for each sound universe. There is no preconceived concept of what a creature is in the system; patterns emerge and interact through the non-linearities which are rife in the system, the most important being the portions of the system which keep the sound within bounds that can be expressed within the system (i.e. volume, length, etc.). An intial seed is specified, but changing this generally does not affect the output in the long run, as the patterns that can reproduce themselves in the "laws of physics" as specified tend to take over, as they do in our world, in petri dishes and on the surface of our planet. However, small changes can sometimes push the system from one recognizable mode into another, suggesting the differentiation of tissues in an organism or speciation in a gene pool. Although external sounds can be used as part of the input, experience shows that this tends to "muddle" the sound, which otherwise becomes extremely distinct and environmental, resembling the sounds of an alien ecosystem. Both people and animals seem to find the output attractive and interesting, as it contains correlations and structure of a fractal nature. I have implemented systems of this type on a Linux system, as well as MAX/MSP, a Nord virtual modular synthesizer, and a Kyma system. Future work will apply similar approaches to the creation of video universes. A description of the details of the systems will appear in a future article. Streams and samples of audio from this system can be accessed at www.eternalnovelty.com. Thank you.