Reading Ideas

I wrote an e-mail to one of the legends of computer science, John Holland, who revolutionized the concept of complex adaptive systems (better known as genetic algorithms) which serves as the basis for 'fuzzy logic.' If we are ever capable of generating artificial intelligence, this would be the path it would be down. The remarkable capability of these complex adaptive systems is that it eliminates the need for a human being to 'understand' it. In doing so, its ability to 'grow' a solution is only limited by the problem itself, rather than the limits of human capacity to think up a solution. I won't go into the details, but it models 'evolution' in concept, favoring solutions which perform better over those that don't.

Just to give you an idea of how brilliant John Holland is, I'll just say that he currently is a professor at the University of Michigan, teaching computer science, electrical engineering, AND psychology. If I receive a reply from him, I will be sure to post it. My e-mail was entitled "Complex Adaptive Systems and Other worthless topics" and is follows:

Of course, I do not believe complex adaptive systems to be a worthless topic. Quite the contrary in fact, I would be a fool for basing my thesis topic on it if I believed them to be worthless. In my humble opinion, modelling behavior without regard to "how" but rather with emphasis on "what" is quite genius.

Then why the subject? I tend to think the focus of modern computers is to create a faster and more "profitable" product for clients, and don't get me wrong, in the practical world of business, it is important in of itself to produce a faster computer. However, nowadays, computer science has been a rat race to produce a more efficient and dilligent programmer to reflect the business side of things.

Then, wasn't it Dijkstra who said that computer science is no more about computers than a telescope is like astronomy?

I feel that true genius models nature, which ironically has unwittingly produced the human mind that is so complex that not even we ourselves can begin to understand. We, computer scientists, have put so much emphasis in the "why", that we've forgotten that the ultimate goal is the "what." It means we've sacrificed what we are capable of in order to keep a firm grip on the understanding.

If an algorithm is a concise instruction list on how to do something, then by focusing on such conciseness, the most we can expect from a computer is what is in our capability of describing concisely.

Why can we not take another page from nature? The human mind consists of a left hemisphere and a right hemisphere. The left hemisphere follows instruction down to the letter, without abstraction. The focus resides in logical steps, typically represented by symbols (words, numbers). This is loosely like the ability of a computer now. The right hemisphere uses images and abstraction to solve problems. The type of problems solved by the right hemisphere is typically that which follows an "Eureka!" with abstract models that aids in understanding of complex problems. Contrary to the left hemisphere, its input nor its output is capable of being described concisely.

It is rumored to be true (however not proven unfortunately) that the distinctive quality homosapiens had over their neanderthal cousins was a 'split-function' brain, which left neanderthals (while intelligent) the inability to abstract. Was this the reason homosapiens thrived over the neanderthals?

Perhaps a computer should have such an engine. It would not be used to solve problems like 4*2, but rather image analysis or fuzzy logic computer intelligence (give a computer a 'mind' of its own). More importantly, it would solve the answers to questions we yet do not know how to ask accurately.

My question to you, from one computer scientist to another, would adaptive computer systems (possibly genetic algorithms) be a starting point for such an engine?

"If I die of a heart attack eating bacon, I'll be a happy man." -My father

John Holland's reply:

I largely agree with your outlook. Indeed I think that
metaphor is the heart of ALL creative work. Words have auras of meaning, context, and tools; metaphors extract interesting
intersections, suggesting new possibilities. Building blocks
underpin metaphors (see HIDDEN ORDER) and GAs are good at recombining building blocks in new plausible arrangements. So, yes, I think there is a possibility there. You may also find Melanie Mitchell's book on analogy of interest (MIT press).
If you should read HIDDEN ORDER and have questions, let me know.
John Holland

"If I die of a heart attack eating bacon, I'll be a happy man." -My father