- #26

- 3,097

- 144

I was just making a small remark since we have some partial agreements before, but maybe I just misinterpreted you.

If you start with an infinity of possibilities, that you put constraints on, you somehow end up with a gigantic landscape that is problematic. I can't figure out how that's computable or managable to a limited processing agent.

I just think the idea of an infinite sea of degrees of freedom is doubtful, as it fails to match any reasonable expectation I have on an inside view.

I like to make a clear distinction between a defined uncertainty and just undecidability, maybe this is the source of our confusion.

If you consider a probability space; ie. the missing information is still constrained to an event space and comes with some equiprobable states or prior probabiltiy distribution. This way we can quantify the missing information, since our uncertainty is constrained by the context (microstructure of probability space and prior). To actually decide, and quantify and measure uncertainty actually requires alot of information!

Sometimes you can't even do that, and the state space itself is uncertain. If this is what you mean then I agree. I just wouldn't call that infinite possibilities. To me the set of possibilities are a physical in the sense that they determine the action.

Edit: This is also in a sense the essense of evolution, and evolutionary learning in the sense that variation must be there but it must be small, or we loose stability. Ie. we need options, possibilities but not too many of them so that we get lost.

/Fredrik

Indeed, it's easier to remove or trace out information, and to create it. I think this is a common argument and it's why this is a common approach.If you have everything, then it is easy to constrain its variety to be left with something more limited.

If you start with an infinity of possibilities, that you put constraints on, you somehow end up with a gigantic landscape that is problematic. I can't figure out how that's computable or managable to a limited processing agent.

I just think the idea of an infinite sea of degrees of freedom is doubtful, as it fails to match any reasonable expectation I have on an inside view.

So maybe you even meant the opposite of what I thought?This "infinity of degrees of freedom" does not actuallyexist. It is just a vagueness, a nakedpotential. So don't get too hung up on the idea of some actual realm of everythingness. That would be unreasonable!

I like to make a clear distinction between a defined uncertainty and just undecidability, maybe this is the source of our confusion.

If you consider a probability space; ie. the missing information is still constrained to an event space and comes with some equiprobable states or prior probabiltiy distribution. This way we can quantify the missing information, since our uncertainty is constrained by the context (microstructure of probability space and prior). To actually decide, and quantify and measure uncertainty actually requires alot of information!

Sometimes you can't even do that, and the state space itself is uncertain. If this is what you mean then I agree. I just wouldn't call that infinite possibilities. To me the set of possibilities are a physical in the sense that they determine the action.

Edit: This is also in a sense the essense of evolution, and evolutionary learning in the sense that variation must be there but it must be small, or we loose stability. Ie. we need options, possibilities but not too many of them so that we get lost.

/Fredrik

Last edited: