After reading a recent post then having a late night conversation about quantum physics I realized that I didn’t have any reason to believe in MWI any more, since I only think it’s better as Truth than Copenhagen, but if I’m only concerned with description then Copenhagen describes the fact that I end up experiencing one timeline better.
I want a description of my expected future experiences; if that means that I have expected variables in it rather than forks in a road, that actually makes it better because the “fork in the road” metaphor is agenty whereas the “random variable” metaphor is uncontrollable.
I can imagine purposes for which envisioning multiple different hypotheticals is useful for decision-making, so I will concede this point. My original opinion was simply that I have different criteria for what makes me sleep better at night than I thought I did, anyway.
Decision-theoretically, what matters is consequences, not experiences.
I’m confused by this distinction. Can you give me an example of an experience that is not a consequence and therefore doesn’t matter decision-theoretically? Can you give me an example of a consequence that is not an experience and therefore matters decision-theoretically?
For example, if you make a decision and then die, there will be consequences, but no future experiences. While future experiences are part of consequences, they don’t paint a balanced picture, as (predictable) things outside experiences are going to happen as well. You can send $X to charity, and expected consequences will predictably depend on specific (moderate) value of X, but you won’t expect differing future experiences depending on X.
If you come to a fork in the road, and go left, you shouldn’t update your priors to assume that the right-hand fork didn’t really exist. Equally, if you end up in WorldA, you shouldn’t update your priors to assume that WorldB doesn’t exist.
Of course, we don’t have any evidence for WorldB existing, unlike the right-hand fork, since there’s no experimental difference between MWI and Copenhagen right now. Given that, I’d assume the proper belief is “Either MWI or Copenhagen, with even odds between the two”?
If there is no experimental difference between MWI and Copenhagen, the “proper” belief is the one that serves your purposes other than seeking universal truths.
I’m realizing we’re probably both confused by the word “belief.”
Newtonian physics has been proven wrong, but it’s still a perfectly reasonable system to use for, say, landing a man on the moon. I thus have the belief “Newtonian physics is inaccurate under certain extreme conditions” and “Newtonian physics is an accurate-enough model for a specific domain of problems, which includes basically every practical problem I am likely to run in to in my daily life”.
Given that both MWI and Copenhagen are accurate models of the data, I would think that it is accurate to have the belief “both of these are accurate models of the data; I can use either one and will get equally accurate predictions”. One could then have additional beliefs, such as “I think MWI is closer to Universal Truth” and “I think Copenhagen does a better job of telling a story that is compatible with my conscious narrative”.
It seems important to me not to get these multiple beliefs confused in to a single simplified belief such as “MWI is true” or “Copenhagen is true” or even “Newtonian physics is true”.
So, when I said “I’d assume THE proper belief is...”, what I really should have said was probably “In addition to the belief you just stated, you can still ALSO believe that they’re both accurate descriptions of experimental evidence.”
tl;dr: Good point on Copenhagen vs MWI, but don’t let yourself forget that both of them are accurate models of our experimental evidence so far. If there’s ever a test between the two, you should probably only be ~50% confident in either one turning out true :)
After reading a recent post then having a late night conversation about quantum physics I realized that I didn’t have any reason to believe in MWI any more, since I only think it’s better as Truth than Copenhagen, but if I’m only concerned with description then Copenhagen describes the fact that I end up experiencing one timeline better.
It should be description of the world, not description of your experience.
Whence this should? That is my point.
I want a description of my expected future experiences; if that means that I have expected variables in it rather than forks in a road, that actually makes it better because the “fork in the road” metaphor is agenty whereas the “random variable” metaphor is uncontrollable.
For what purpose? Decision-theoretically, what matters is consequences, not experiences.
I can imagine purposes for which envisioning multiple different hypotheticals is useful for decision-making, so I will concede this point. My original opinion was simply that I have different criteria for what makes me sleep better at night than I thought I did, anyway.
I’m confused by this distinction. Can you give me an example of an experience that is not a consequence and therefore doesn’t matter decision-theoretically? Can you give me an example of a consequence that is not an experience and therefore matters decision-theoretically?
For example, if you make a decision and then die, there will be consequences, but no future experiences. While future experiences are part of consequences, they don’t paint a balanced picture, as (predictable) things outside experiences are going to happen as well. You can send $X to charity, and expected consequences will predictably depend on specific (moderate) value of X, but you won’t expect differing future experiences depending on X.
Gotcha! Sure, that makes sense. Thanks.
If you come to a fork in the road, and go left, you shouldn’t update your priors to assume that the right-hand fork didn’t really exist. Equally, if you end up in WorldA, you shouldn’t update your priors to assume that WorldB doesn’t exist.
Of course, we don’t have any evidence for WorldB existing, unlike the right-hand fork, since there’s no experimental difference between MWI and Copenhagen right now. Given that, I’d assume the proper belief is “Either MWI or Copenhagen, with even odds between the two”?
If there is no experimental difference between MWI and Copenhagen, the “proper” belief is the one that serves your purposes other than seeking universal truths.
I’m realizing we’re probably both confused by the word “belief.”
Newtonian physics has been proven wrong, but it’s still a perfectly reasonable system to use for, say, landing a man on the moon. I thus have the belief “Newtonian physics is inaccurate under certain extreme conditions” and “Newtonian physics is an accurate-enough model for a specific domain of problems, which includes basically every practical problem I am likely to run in to in my daily life”.
Given that both MWI and Copenhagen are accurate models of the data, I would think that it is accurate to have the belief “both of these are accurate models of the data; I can use either one and will get equally accurate predictions”. One could then have additional beliefs, such as “I think MWI is closer to Universal Truth” and “I think Copenhagen does a better job of telling a story that is compatible with my conscious narrative”.
It seems important to me not to get these multiple beliefs confused in to a single simplified belief such as “MWI is true” or “Copenhagen is true” or even “Newtonian physics is true”.
So, when I said “I’d assume THE proper belief is...”, what I really should have said was probably “In addition to the belief you just stated, you can still ALSO believe that they’re both accurate descriptions of experimental evidence.”
tl;dr: Good point on Copenhagen vs MWI, but don’t let yourself forget that both of them are accurate models of our experimental evidence so far. If there’s ever a test between the two, you should probably only be ~50% confident in either one turning out true :)