Tuesday, June 12, 2007

Science, the Evolutionary Algorithm.

I do a bit of work with evolutionary algorithms. They are exceedingly powerful, and in fact are the only way we know to increase the complexity of a system without direct action. As evolution does not require any intellect to perform.

Evolution needs three things, replication, modification and filtering. You simply need to filter out those things which don't work, create more things, and change them a little. And the most important step "filtering out those things things which don't work" is the most subjective and thus the most powerful. It doesn't matter what you say doesn't work. If you are modifying a text string and the closer the letters are to the end of the alphabet the more likely they are to be filtered out... then you'll end up with large amounts of early letter text strings. Basically you need to be able to figure out what isn't the answer, and that's pretty easy to do.

That's even assuming you have a goal for the evolution to accomplish, which is certainly not needed nor existing biological evolution. The modification also need not be random, though random is probably better than non-random in truth.

This brings us to science. We test ideas scientifically by looking at the evidence. We create a hypothesis (genotype) about how the world works (phenotype) and find a method for testing this hypothesis (filter). We accept and improve upon those ideas which pass this filter creating more hypotheses (duplication) with slightly altered (modification) ideas in order to produce a more exact theory (fitter).

We tend to think that we are awesome in our thinking ability. We are this fantastic thinking individual. We can set our minds to something and through our amazing mental powers determine something truly amazing. In reality, some of the most major discoveries are mistakes. This isn't because we are just lucky with our mistakes, rather our ability to think is holding us back. We do not create non-random modification. We create predictable, often flawed modifications, over and over again, we reject brilliant (accidental) ideas out of hand without evaluation. And tend to only get a little entropy into our science when we fuck up. Most genius is really in not rejecting an idea. In providing that extra flash of insight to induce us *not* to reject an idea.

We aren't that bad ass. We are naked apes, the third known species of chimpanzees. Sure science is great, but really we are just latching on to the power of evolution and performing the operations (often unwittingly) on ideas faster than ever before (most species are unable to adopt ideas at all stunting the process as replication). We are taking evolution to the next level, and frankly, species just can't adapt to us fast enough, save perhaps some bacteria.

No comments: