A question came up in response to EY’s recent sequence posts that I’d like someone to take a shot at: EY seems to me at least to be saying that the universe is a ‘fabric of causal relations’ or is ‘made of cause and effect’ or something like that.
He’s also said that probability (and so causal relations, given how he understands them) are ‘subjectively objective’.
The first claim implies that casual relations are fundamental to the universe, the second implies that they’re ways in which limited observers and agents deal with what is fundamental. As such, the two claims seem to be inconsistent. What’s going on here?
The solution is that causal relations are a map, reality is the territory. You and I could very possibly have different causal structures in mind when we’re talking about, e.g., moving billiard balls, and we can both be correct if we have different sets of information. There is only one reality, but there are many correct maps of reality, each one corresponding to a different set of previous information.
If I understand you, you’re saying that causal relations are a (perhaps necessary) feature of the map but are not features of the territory. Is that correct? If so, it seems like the claim “the universe is a fabric of causal relations’ is strictly speaking false, or at least it’s only true if by ‘the universe’ we mean the map rather than the territory, which would be weird.
I made a mistake, but I think fixing the first sentence is all that I need to do. (Maybe I merely misspoke, but I’m not sure what I was thinking, even only a couple hours ago).
The first sentence should read something like: Reality is a particular causal web, but the correct model of that causal web depends on your state of information. In other words, the subjectively objective component only comes in when we try to infer something about the causal web that is reality.
Some of it might be that no two agents will have the same experiences, and so they will not have the same probabilities assigned to particular propositions even if they started with the same priors, has identical sense-receptors, and are both perfect Bayesians.
But it seems misleading to use the label “subjectively objective” for that phenomena. And I might be totally off track, in which case I am totally confused about what “subjectively objective” is supposed to be about.
But it seems misleading to use the label “subjectively objective” for that phenomena. And I might be totally off track, in which case I am totally confused about what “subjectively objective” is supposed to be about.
Probability is subjective in one sense and objective in another sense. It’s subjective in that the correct answer to “What’s the probability of A?” depends on who is asking the question. It’s objective in that the answer depends on who is asking the question only through the information she has and not, e.g., who she is. Part of the reason to call it subjectively objective is to acknowledge that critics of Bayesian epistemology/probability/statistics are correct, in part, when they complain that it’s subjective. The objective part answers the criticism by pointing out that probability is subjective in a very benign sense and in precisely the sense we intuitively expect it to be. E.g. “Mary didn’t know Jack had pocket aces, so in her situation thinking that she was highly likely to have the winning hand was correct.”
A question came up in response to EY’s recent sequence posts that I’d like someone to take a shot at: EY seems to me at least to be saying that the universe is a ‘fabric of causal relations’ or is ‘made of cause and effect’ or something like that.
He’s also said that probability (and so causal relations, given how he understands them) are ‘subjectively objective’.
The first claim implies that casual relations are fundamental to the universe, the second implies that they’re ways in which limited observers and agents deal with what is fundamental. As such, the two claims seem to be inconsistent. What’s going on here?
The solution is that causal relations are a map, reality is the territory. You and I could very possibly have different causal structures in mind when we’re talking about, e.g., moving billiard balls, and we can both be correct if we have different sets of information. There is only one reality, but there are many correct maps of reality, each one corresponding to a different set of previous information.
If I understand you, you’re saying that causal relations are a (perhaps necessary) feature of the map but are not features of the territory. Is that correct? If so, it seems like the claim “the universe is a fabric of causal relations’ is strictly speaking false, or at least it’s only true if by ‘the universe’ we mean the map rather than the territory, which would be weird.
I made a mistake, but I think fixing the first sentence is all that I need to do. (Maybe I merely misspoke, but I’m not sure what I was thinking, even only a couple hours ago).
The first sentence should read something like: Reality is a particular causal web, but the correct model of that causal web depends on your state of information. In other words, the subjectively objective component only comes in when we try to infer something about the causal web that is reality.
I’ve wondered the same thing.
Some of it might be that no two agents will have the same experiences, and so they will not have the same probabilities assigned to particular propositions even if they started with the same priors, has identical sense-receptors, and are both perfect Bayesians.
But it seems misleading to use the label “subjectively objective” for that phenomena. And I might be totally off track, in which case I am totally confused about what “subjectively objective” is supposed to be about.
Probability is subjective in one sense and objective in another sense. It’s subjective in that the correct answer to “What’s the probability of A?” depends on who is asking the question. It’s objective in that the answer depends on who is asking the question only through the information she has and not, e.g., who she is. Part of the reason to call it subjectively objective is to acknowledge that critics of Bayesian epistemology/probability/statistics are correct, in part, when they complain that it’s subjective. The objective part answers the criticism by pointing out that probability is subjective in a very benign sense and in precisely the sense we intuitively expect it to be. E.g. “Mary didn’t know Jack had pocket aces, so in her situation thinking that she was highly likely to have the winning hand was correct.”
Edit: clarified example