Surely it means: a god steps into the world beside you, and offers you the choice: “either take the A-or-C gamble, in which case I’ll guarantee to arrange for A or C; or else refuse it, in which case I’ll go away and leave your life uninterfered with”. Or, in other words, B means whatever future or lottery-between-futures you’d have been facing if you hadn’t been given this offer.
My problem with this definition of PLP is that it does weird things with determinism vs. free will (just like some people complain about Newcomb’s problem statement).
If I win A will I live to a healthy 120 no matter what I do, like quantum immortality? Could I, e.g., infect myself with AIDS, drink a deliberately randomized cocktail, and have it turn into a magic AIDS cure because I’m guaranteed health? Could I cause an atom bomb to malfunction by standing next to it and setting it off? Could I donate organs and limbs for transplants and then regrow them? (More serious examples can be provided. For instance, I suspect I could rig up an experiment that’s guaranteed to kill me unless a source of randomness just happens to produce code for a Friendly Seed AI, or a proof of P!=NP, or other useful things.)
Conversely, in option B I can’t know what my life will be like. I used to have a pretty good estimate, but I just realized I really have no idea. After all, my estimate didn’t include a god offering me limited immortality. Maybe tomorrow I’ll get an offer from Satan for my soul?
For instance, under your first suggestion—that A is an immediate prize of e.g. money—the PLP becomes simple risk aversion of death vs. money. The original description seemed to suggest more than that, what with choosing whole alternative futures.
Surely it means: a god steps into the world beside you, and offers you the choice: “either take the A-or-C gamble, in which case I’ll guarantee to arrange for A or C; or else refuse it, in which case I’ll go away and leave your life uninterfered with”. Or, in other words, B means whatever future or lottery-between-futures you’d have been facing if you hadn’t been given this offer.
My problem with this definition of PLP is that it does weird things with determinism vs. free will (just like some people complain about Newcomb’s problem statement).
If I win A will I live to a healthy 120 no matter what I do, like quantum immortality? Could I, e.g., infect myself with AIDS, drink a deliberately randomized cocktail, and have it turn into a magic AIDS cure because I’m guaranteed health? Could I cause an atom bomb to malfunction by standing next to it and setting it off? Could I donate organs and limbs for transplants and then regrow them? (More serious examples can be provided. For instance, I suspect I could rig up an experiment that’s guaranteed to kill me unless a source of randomness just happens to produce code for a Friendly Seed AI, or a proof of P!=NP, or other useful things.)
Conversely, in option B I can’t know what my life will be like. I used to have a pretty good estimate, but I just realized I really have no idea. After all, my estimate didn’t include a god offering me limited immortality. Maybe tomorrow I’ll get an offer from Satan for my soul?
Could you possibly take your prolific imagination for how this is misspecified or wrong, and turn to imagining ways to improve it or patch it up?
For example, suppose that winning A, instead of guarantees extending into the far future, merely gives you a very large gift in the near future.
Alternatively, suppose that all of your questions regarding A were answered “Yes”.
Alternatively, is it not possible to price a bet offered to you even if you are uncertain regarding exactly how much you would win, if you did win?
Well, I’m not sure what it’s supposed to specify.
For instance, under your first suggestion—that A is an immediate prize of e.g. money—the PLP becomes simple risk aversion of death vs. money. The original description seemed to suggest more than that, what with choosing whole alternative futures.