Natural evolution in the digital medium is a new technology, about which we know very little. The hope is to evolve software with sophisticated functionality far beyond anything that has been designed by humans. But how long might this take? Evolution in the organic medium is known to be a slow process. Certainly there remains the possibility that evolution in the digital medium will be too slow to be a practical tool for software generation, but several observations can be made that provide encouragement.
First, computational processes occur at electronic speeds, and are in fact relatively fast. Second, as was noted above, during the Cambrian, evolution produced such a rapid inflation of complexity and diversity, that it has come to be known as an ``explosion''. Figure one illustrates conceptually, that the bulk of the complexity of living systems on Earth appeared suddenly at the time of the Cambrian explosion. If complexity had developed gradually, at a steady pace through the history of life, then it would probably be hopeless to attempt to use evolution as a methodology for generating complexity. However, if the Cambrian explosion phenomenon is a general property of evolving systems, then it may be practical to use evolution to generate complexity in evolving digital systems.
A third point remains to be made. Let us consider a thought experiment. Imagine that we are robots. We are made out of metal, and our brains are composed of large scale integrated circuits made of silicon or some other semi-conductor. Imagine further, that we have no experience of carbon based life. We have never seen it, never heard of it, nor ever contemplated it. Now suppose a robot enters the scene with a flask containing methane, ammonia, hydrogen, water and a few dissolved minerals. This robot asks our academic gathering: ``Do you suppose we could build a computer out of this material.'' The theoreticians in the group would surely say yes, and propose some approaches to the problem. But the engineers in the group would say: ``Why bother when silicon is so much better suited to information processing than carbon.''
From our organo-centric perspective the robot engineers might seem naive, but in fact I think they are correct. Carbon chemistry is a lousy medium for information processing. Yet the evolutionary process embodies such a powerful drive to generate information processing systems, that it was able to rig up carbon based contraptions for processing information, capable of generating the beauty and complexity of the human mind. What might such a powerful force for information processing do in a medium designed for that purpose in the first place? It is likely to arrive more quickly at sophisticated information processes than evolution in carbon chemistry, and would likely achieve comparable functionality with a greater economy of form and process. Evolution is a process that explores the possibilities inherent in the medium.