Entropy and freedom
Mar. 1st, 2012 09:53 am![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
davvHere's an interesting thing I just thought of. I hope I'm right, because I'm not completely familiar with the concept of entropy yet.
First, say you have a goal-seeking mechanism aiming to maximize entropy. Then, if it can find a libertarian incompatibilist ("free-willed in the sense of breaking causality") component, it will implement it, because that will increase entropy more quickly.
If I'm right in that entropy is basically probabilistically weighted variety, that seems reasonable enough. A component not bound to causality will introduce different output where there was identical input (from the causal perspective, i.e. from the universe). Furthermore, since the component isn't random, either, its output can't be reduced to a simple probability distribution with little "effective" variety.
Second, I have the suspicion evolution is such an entropy maximizer. What does organization, like the self-organization of evolution, do? It makes the "interior" more orderly (or perhaps one should say, more controllable) at the cost of making the "exterior" less so. As the complexity of organization rises, the system as a whole becomes better at that sort of control.
(And now that I think about it, I have a dim outline of an idea suggesting that entropy maximization is not particular of evolution. Consider any goal-seeking mechanism. To get to its goal, it likely needs to control the external environment. And the way it can do so, in the manner of requisite variety (once all the options to work with nature instead of canceling it out have been exhausted), is to increase its own variety.)
But even if I'm right, there are two very big ifs here. The first one is that libertarian incompatibilism is true. If not, there are no free-willed agents in this sense, and so they can't be acquired. The second is that a sufficiently advanced entropy-maximizer will find them. If free-willed agents exist but divining the form of the interface required to pull them into your system is as difficult as breaking 256-bit AES, then we're out of luck.
Yet they are conditional. If you want a Pascal's-wager sort of conclusion, just make a better goal-seeking entropy maximizer. If either assumption above is false, then you lose little (as you're just doing, more directedly, what evolution would eventually have got around to doing); but if they are true, then you can import truly free beings into the universe even if you yourself aren't one of them.
Can it be that easy? That's what's making me doubt.
(And to think all of this came about, in a roundabout fashion, from reading the story I talked about in my previous post.)
First, say you have a goal-seeking mechanism aiming to maximize entropy. Then, if it can find a libertarian incompatibilist ("free-willed in the sense of breaking causality") component, it will implement it, because that will increase entropy more quickly.
If I'm right in that entropy is basically probabilistically weighted variety, that seems reasonable enough. A component not bound to causality will introduce different output where there was identical input (from the causal perspective, i.e. from the universe). Furthermore, since the component isn't random, either, its output can't be reduced to a simple probability distribution with little "effective" variety.
Second, I have the suspicion evolution is such an entropy maximizer. What does organization, like the self-organization of evolution, do? It makes the "interior" more orderly (or perhaps one should say, more controllable) at the cost of making the "exterior" less so. As the complexity of organization rises, the system as a whole becomes better at that sort of control.
(And now that I think about it, I have a dim outline of an idea suggesting that entropy maximization is not particular of evolution. Consider any goal-seeking mechanism. To get to its goal, it likely needs to control the external environment. And the way it can do so, in the manner of requisite variety (once all the options to work with nature instead of canceling it out have been exhausted), is to increase its own variety.)
But even if I'm right, there are two very big ifs here. The first one is that libertarian incompatibilism is true. If not, there are no free-willed agents in this sense, and so they can't be acquired. The second is that a sufficiently advanced entropy-maximizer will find them. If free-willed agents exist but divining the form of the interface required to pull them into your system is as difficult as breaking 256-bit AES, then we're out of luck.
Yet they are conditional. If you want a Pascal's-wager sort of conclusion, just make a better goal-seeking entropy maximizer. If either assumption above is false, then you lose little (as you're just doing, more directedly, what evolution would eventually have got around to doing); but if they are true, then you can import truly free beings into the universe even if you yourself aren't one of them.
Can it be that easy? That's what's making me doubt.
(And to think all of this came about, in a roundabout fashion, from reading the story I talked about in my previous post.)
no subject
Date: 2012-03-04 09:45 pm (UTC)no subject
Date: 2012-03-05 07:30 pm (UTC)no subject
Date: 2012-03-07 02:10 am (UTC)no subject
Date: 2012-03-09 06:05 pm (UTC)no subject
Date: 2012-03-10 08:43 am (UTC)