The Next Big Leap in "Programming"

Tonights Portland Nerd Dinner was a barn-burner. Not only did we have the typical geek banter and laughter, but we spend some time on the question I posed the other day about what should replace character stream-based programming.

The problem is that all of the hand-crafting of solutions doesn't scale. The answer was posed by a fellow nerd who, who pointed out that we've already got a model for self-organizing, self-evolving, self-maintaining systems: biology. God didn't hand code humans and zebras and platypi -- he defined a class and instance encoding method (DNA) and an environment to serve as the runtime (Earth) and let us interact with and improve ourselves and our environment over time to solve the problem. What's the problem? What species survives best on this planet. So far, we're the winners, but we'll see what happens at the next ice age or when the next asteroid hits.

.NET and Longhorn have provided us a big step forward in terms of hand-crafting computer-based solutions to our problems, but only self-organizing, self-evolving, self-repairing systems can really scale. I know that IBM has done some work in this area with their automonic computing, but my personal favorite work on this topic is Genetic Programming III by John Koza. Anyone for a Genetic Algorithm Runtime (GAR)?