Self-Improving Systems
Roland Olsson writes "A relatively easy way to construct "intelligent" systems that improve
themselves practically ad infinitum is described at
http://www-ia.hiof.no/~rolando/SIG/
Maybe Steven Spielberg's AI film is closer to reality than the general
public knows *smile*?"
The issue worth thinking about is the implication that there *can't* be any "silver bullets" (not just that there aren't any right now). Especially since, if we find something that dramatically improves productivity, we crank up the level of complexity of the systems we design.
As the scope of the systems keeps increasing, it's hard to see how a single tool or technique could solve the compelexity problem for more than a few years. If we want to get out of the tar-pit permanently, we're going to need some really radical (and currently unfeasible) like replacing digital, sequential processors with something more suited to good engineering or reinventing ourselves as as less falible people (or genuinely intelligent, self-improving expert systems that make like less falible people).
However, I don't think that the scope _is_ going to increase indefinitely. Back in 1980, any system that could be produced at low risk was probably to trivial for general use and a useful system was probably too hard. Any improvement in process had immediately to be submerged by bigger systems because the existing systems were so inadequate. Nowadays, there are many useful systems (especially the smaller embedded ones) that are quite managable with existing tools. The complexity of useful systems is limited by the complexity of the human affairs that they support, and we aren't evolving our society fast enough to raise that complexity limit much.
Over the next ten years, I predict that process, tools, and techniques will gradually catch up with the sort of jobs we routinely need to undertake. So, no silver bullets, but we may yet get the werewolf house-trained and turn him vegetarian.
Imagine a Beowulf cluster of these.
"What are you doing, Dave?"
This space intentionally left blank.