>A model compatible with the known laws of physics is that what we think of as modeling, predicting and making choices is actually learning which one of the possible worlds we live in.
(I would state this somewhat differently, but let’s go with it for now for the sake of argument.)
Do you consider “which one of the possible worlds we live in” to be synonymous with “reality” or “territory”? If so, would you agree that this model is useful anytime we make decisions (i.e., there’s not really an alternative model that we can use to serve the same purpose)? If so, it seems like the concept of territory isn’t just a “sometimes useful” model but at least one of the most useful models we have, and in fact pretty much indispensable? How does this differ in practice from what Eliezer thinks? I think you were complaining that Eliezer asks whether wavefunctions are real, but couldn’t you ask a similar question, namely, does the possible world that you live in contain wavefunctions?
Do you consider “which one of the possible worlds we live in” to be synonymous with “reality” or “territory”?
I consider the map/territory model to be useful in this case, yes. I don’t promote the idea of the territory into anything other than a useful model in this case.
If so, would you agree that this model is useful anytime we make decisions
I wouldn’t make a sweeping statement like that, no. But it is definitely useful to consider the person making decisions as a part of the physical world, and not having magical free will, the way the usual decision theories go, while paying lip service to the idea of reality.
If so, it seems like the concept of territory isn’t just a “sometimes useful” model but at least one of the most useful models we have, and in fact pretty much indispensable? How does this differ in practice from what Eliezer thinks?
I don’t know what he thinks exactly, but my impression is what I had described above, talking about territory while still thinking that the intentional stance is anything more than an occasionally useful approximation. That “occasionally” part does not include decision theories.
I think you were complaining that Eliezer asks whether wavefunctions are real, but couldn’t you ask a similar question, namely, does the possible world that you live in contain wavefunctions?
I don’t recall complaining about it, but wavefunctions are a mathematical abstraction, obviously. Not a lot of use in asking whether they are really real or only seem real and what not. As for “does the possible world that you live in contain wavefunctions?” question, my answer would be that at the level of coarseness that corresponds to observing someone’s actions, “wavefunction” is not a useful abstraction, just like quarks are not a useful abstraction when talking about, as in Eliezer’s example, a Boeing 747. The only residue that I expect to find useful from quantum mechanics in the macroscopic world of agents is the inherent unpredictability and randomness at the level of the ion channels opening and closing, which, when combined, result in the appearance of conscious decisions.
Not sure if this makes sense, but thank you very much for being patient and engaging in this discussion, and not just shrugging it off.
In your decision making post, you wrote:
>A model compatible with the known laws of physics is that what we think of as modeling, predicting and making choices is actually learning which one of the possible worlds we live in.
(I would state this somewhat differently, but let’s go with it for now for the sake of argument.)
Do you consider “which one of the possible worlds we live in” to be synonymous with “reality” or “territory”? If so, would you agree that this model is useful anytime we make decisions (i.e., there’s not really an alternative model that we can use to serve the same purpose)? If so, it seems like the concept of territory isn’t just a “sometimes useful” model but at least one of the most useful models we have, and in fact pretty much indispensable? How does this differ in practice from what Eliezer thinks? I think you were complaining that Eliezer asks whether wavefunctions are real, but couldn’t you ask a similar question, namely, does the possible world that you live in contain wavefunctions?
I consider the map/territory model to be useful in this case, yes. I don’t promote the idea of the territory into anything other than a useful model in this case.
I wouldn’t make a sweeping statement like that, no. But it is definitely useful to consider the person making decisions as a part of the physical world, and not having magical free will, the way the usual decision theories go, while paying lip service to the idea of reality.
I don’t know what he thinks exactly, but my impression is what I had described above, talking about territory while still thinking that the intentional stance is anything more than an occasionally useful approximation. That “occasionally” part does not include decision theories.
I don’t recall complaining about it, but wavefunctions are a mathematical abstraction, obviously. Not a lot of use in asking whether they are really real or only seem real and what not. As for “does the possible world that you live in contain wavefunctions?” question, my answer would be that at the level of coarseness that corresponds to observing someone’s actions, “wavefunction” is not a useful abstraction, just like quarks are not a useful abstraction when talking about, as in Eliezer’s example, a Boeing 747. The only residue that I expect to find useful from quantum mechanics in the macroscopic world of agents is the inherent unpredictability and randomness at the level of the ion channels opening and closing, which, when combined, result in the appearance of conscious decisions.
Not sure if this makes sense, but thank you very much for being patient and engaging in this discussion, and not just shrugging it off.