Dionysian Rationality (part I)

In which I try to answer UrbanElf's Challenge: Exhibit a game in which the optimal winning strategy necessitates the use of random behavior!

I give an example of solving the traveling salesman problem--a deterministic solution can be beaten by a randomized algorithm.

With this example I'm aiming to prove that "Randomness in the Brain" can be more than just randomness in the brain. I want to point out that there is an inherent random component to rational behavior, which I call "Dionysian Rationality"

Category:

Education

Tags:

• rational
• irrational
• free
• will
• determinism
• nondeterminism
• causality
• randy
• helzerman
• UrbanElf
• traveling
• salesman
• simulated

[smotviddy]

randy i think you have convinced me!

[randyhelzerman]

:-) awesome smotviddy! Now that you know you have free will, please use your superpower for good, not evil! :-)

[smotviddy]

actually i can't agree with you. i have a few major problems. perhaps ill make a vid.

[randyhelzerman]

So what did I convince you of?

[smotviddy]

i thought you made an excellent point that we are not simply stimulus response machines. however, i didn't see any evidence of random behaviour. i am opposed to treating the subject as if its floating in a bubble outside the greater system it is a part of. i think such a premise is responsible for your conclusion there's non-determinism in play.

[MikeSarno]

Excellent!

[naphra2]

I know I probably should, long time no video make. :)

[randyhelzerman]

Well naphra2, there's a good research project for you :-) why don't you investigate it and make a vid! :-)

[naphra2]

Rather, I'd look for perhaps some sort of a vestigial use for neurons that would have benefited from neurons that are closer and closer to neurons that satisfy the conditions for nondeterminism, perhaps under a selective process not even selecting for better randomness. But there's still the problem that the nondeterminism itself would, I think, then emerge abrubtly and not gradually, and so it could ruin the achieved selective advantage, and the achieved nondeterminism would likely not survive.

[naphra2]

Yes, that's true, I can see that. But I think I can see some problems, too: good pseudorandom-number generators are always more and more complex, the proposed nondeterministic generator is simple. In order to get a good deterministic generator, you need more and more neurons; in order to get a nondeterministic generator, you need just one neuron. So evolving a better deterministic generator wouldn't necessarily aid, or even relate to, evolving a nondeterministic generator.