My theory produces the exact same predictions as a theory containing the realism postulate. It shouldn’t affect any of your decisions unless your decisions hinge on incoherent claims that are irrelevant to predictions. I think such decision theories are wrong in some sense. Regardless, I don’t think that my arguments have any real relevance to what anyone should or should not do.
Matter in what context? I didn’t discuss decision theory in the post (other than indirectly w.r.t. altruism), Gordon is the first to bring it up, I’m simply saying that I believe that decision theory shouldn’t depend on incoherent claims. My point about predictions is because Gordon asked why something happened, and I’m making the point that both theories predict the same thing happening and explain it just as well.
You don’t have any clear criteria for saying that things are coherent or incoherent.
The general idea of of realism adds value in that explains how science works, where observations come from, and so on. Thats EYs defense of it, and it has been referenced several times in the comments.
There may be specific issues in figuring out which specific theory to believe in, but that’s another matter. There is a stable position where you accept realism, but don’t invest in specific theories.
Given my intuitions about (in)coherence , having to take it in faith that science works, without having any idea why, is less coherent than the alternative!
You don’t have any clear criteria for saying that things exist or don’t exist.
An explanation that depends on incoherent claims isn’t much of an explanation. Which specific post of EY’s are you saying had this defense? I’ve already responded to all of his posts I could find that bear on the issue.
Take it on faith that science works
This is not required, and you’re in exactly the same position whether or not you accept realism. Realism doesn’t imply Occam’s razor, which is required for induction and science more generally. EY has a post justifying Occam that does not require realism. I don’t see what realism adds to the argument. See https://www.lesswrong.com/posts/C8nEXTcjZb9oauTCW/where-recursive-justification-hits-bottom
You can’t do it with occam alone. You need a source of data, and that data needs to have some discoverable consistencies. You can’t perform induction on entropy .
Occam applied to the only data we have, which is our direct observations, says that the scientific method has been useful in the past and should be presumed to continue to be useful. EY laid out the argument in the post I linked to.
“The universe exists and is inherently simple, therefore induction tends to work” and “Observations are well predicted by inductive formulas, therefore induction tends to work” are of the same form. The first is incoherent and the second is meaningful, but the conclusions are the same.
What is the exact argument and conclusion for which you’re saying my view cannot reach?
“The universe exists and is inherently simple, therefore induction tends to work” and “Observations are well predicted by inductive formulas, therefore induction tends to work” are of the same form.
But not if the same content. The second doesn’t tell you how induction works.
My theory produces the exact same predictions as a theory containing the realism postulate. It shouldn’t affect any of your decisions unless your decisions hinge on incoherent claims that are irrelevant to predictions. I think such decision theories are wrong in some sense. Regardless, I don’t think that my arguments have any real relevance to what anyone should or should not do.
You haven’t explained why predictions are the only thing that matter.
Matter in what context? I didn’t discuss decision theory in the post (other than indirectly w.r.t. altruism), Gordon is the first to bring it up, I’m simply saying that I believe that decision theory shouldn’t depend on incoherent claims. My point about predictions is because Gordon asked why something happened, and I’m making the point that both theories predict the same thing happening and explain it just as well.
You don’t have any clear criteria for saying that things are coherent or incoherent.
The general idea of of realism adds value in that explains how science works, where observations come from, and so on. Thats EYs defense of it, and it has been referenced several times in the comments.
There may be specific issues in figuring out which specific theory to believe in, but that’s another matter. There is a stable position where you accept realism, but don’t invest in specific theories.
Given my intuitions about (in)coherence , having to take it in faith that science works, without having any idea why, is less coherent than the alternative!
You don’t have any clear criteria for saying that things exist or don’t exist.
An explanation that depends on incoherent claims isn’t much of an explanation. Which specific post of EY’s are you saying had this defense? I’ve already responded to all of his posts I could find that bear on the issue.
This is not required, and you’re in exactly the same position whether or not you accept realism. Realism doesn’t imply Occam’s razor, which is required for induction and science more generally. EY has a post justifying Occam that does not require realism. I don’t see what realism adds to the argument. See https://www.lesswrong.com/posts/C8nEXTcjZb9oauTCW/where-recursive-justification-hits-bottom
So how does science work?
You can’t do it with occam alone. You need a source of data, and that data needs to have some discoverable consistencies. You can’t perform induction on entropy .
Occam applied to the only data we have, which is our direct observations, says that the scientific method has been useful in the past and should be presumed to continue to be useful. EY laid out the argument in the post I linked to.
It doesn’t tell you how it works.
Neither does realism.
“The universe exists and is inherently simple, therefore induction tends to work” and “Observations are well predicted by inductive formulas, therefore induction tends to work” are of the same form. The first is incoherent and the second is meaningful, but the conclusions are the same.
What is the exact argument and conclusion for which you’re saying my view cannot reach?
But not if the same content. The second doesn’t tell you how induction works.
Neither does the first.