[grey_whiskers]

I then went on to explain that if we use those random mutations to actually produce new programs using a mechanism similar to those found in nature (reproduction of the most fit plus small random mutations, or recombining best programs and picking bits from each one), those programs can display unbelievable complexity and fitness for the solution very rapidly.

That's probably where you lost him. In the genetic algorithm/ simulated evolutionary work, you have a known, constant goal the entire time; ecosystems do not suddenly appear and vanish; there is no predation nor disease; and there is never any extinction.

Not to mention that it begs the questions about "origins" etc. The sample code from which you start, the computer, the kernel of the algorithm which keeps the "best" solutions and discards the others, the very parameters used in the scoring function, did not spontaneously appear.

Perhaps a better illustration would be to start with a blank slate of memory, or random 1's and 0's, and not even a compiled program to start as the basis for your mutations. [Pre-amino acids, you know. ;-) ]

So as an illustrative example to point the capabilities, or a "thought experiment" put into practice, it's just dandy.

But the degree of similarity to biological systems might not be as close as you want for a well-formed simulation...

Cheers!

[js1138]

Chemistry is not a blank slate. The trick in building computer simulations of evolution is to model the crucial features of chemistry with available computer resources.

That's one of the things the Santa Fe Institute was established to study. If they had "the" solution we'd have heard about it.