Quote:The mutation rate is interesting - isn't it far too low to allow for some evolution of interesting features with scripts?

It has to be big enough so that a lineage of a certain "species" can maintain its genotype to be able to form large, complex structures reliably.

Otherwise, the mutations will turn it all into chaos, and a script with complex and interesting properties will not sustain its form long enough to make use of its long-term advantages, and the small, dumb, stable scripts will dominate.

Quote:Are you actually running these things as parallel processes? The nondeterminism introduced by the operating system's scheduling would prevent you from being able to deterministically repeat a trial.

No, they won't be parallel-processing. A script will have to wait on a chain before the next instruction can take over.

And I have no idea what you were saying about repeating trials.

Quote:Original post by polyfrag

Quote:Are you actually running these things as parallel processes? The nondeterminism introduced by the operating system's scheduling would prevent you from being able to deterministically repeat a trial.

No, they won't be parallel-processing. A script will have to wait on a chain before the next instruction can take over.

And I have no idea what you were saying about repeating trials.

I think what he means is that usually by seeding the random number generator with the same seed and using the same input data, you can run something twice and get the same results. But since you are running 6 processes with the OS taking over the process and thread timing, you might not be able to reproduce the same results with the same random number seed and input data.

But it seems that with your new implementation, you've moved into genetic programming territory.