If Omega can’t predict dice, then he isn’t Omega. How can you spend so many words on this without getting it? If your ‘solution’ to the problem involves Omega being unable to predict whether you one or two box, because of dice or anything else, then you aren’t talking about the same problem as everyone else. I’ve pointed this out like 4 times. Oscar and Mr Mind pointed it out too.
you aren’t talking about the same problem as everyone else.
facepalm
Duh. You’re the one not talking about the games in the OP! I was talking about certain very specific modifications to Newcomb’s problem (Starting from the formulation I quoted from the wiki at the top of the OP, not from Nozick’s paper!). I even put “cheating” in the title, as in “breaking the rules” and also in the sense of “cheating on a deal”. I used both of these senses in the OP:
I just cheated Omega out of an expected $250 over one-boxing.
(Cheating a deal.)
The argument requires actual hardware. Hardware that some humans posses, but not innately. If you want to call that cheating, see the title
(Breaking the rules.)
I also spoke of several alternative ways the game could be modified, both in the OP and in this thread. Don’t pretend I said they’re all the same game! That’s logically rude.
Taboo “Newcomb’s Problem”. It’s clearly biasing your thinking if anyone is disrespecting your sacred cow. We’re talking about The Deal. While we’re at it, Taboo “Omega” too, in case he is also a cow.
###Game A
We’re talking about The Deal in a MWI universe. There’s a near-omniscient Being offering the deal. Wait, Taboo “omniscient”, it’s probably ill defined, like everything else about God. Said being knows the wavefunction of the entire universe of discourse (if that’s even consistent), including both the Chooser and the dice, and can predict how it will evolve. This means the Being knows the whole future. The being CAN predict the outcome of a quantum coin flip. Look! He even drew a picture for you:
prediction
|
set up box
|
flip!
|\
| \
H T
The Being correctly predicts both Everett branches. Both a Head and Tail happen. Remember, I said “MWI”. Both branches are equally real. But the Being can either fill the box, or not, before the flip, per The Deal. That any given Chooser copy only sees one “random” outcome is a subjective illusion. This is not random. MWI is deterministic.
The optimal strategy for the Chooser in Game A is to use a QRNG to one-box with just over 50% measure, and two-box with just under 50% measure.
The optimal strategy for the Being before the split (predicting this) to maximize the measure of accurately set-up boxes after the split, is to put the million in.
If you want to dispute Game A, because I missed some important logically valid reason, I’d like to know what that is, but it’s certainly nothing you’ve said so far. And call it “Game A” to be clear.
###Game B
We don’t know if we’re in MWI or not. There’s a different Being offering The Deal. This being has a track record of predicting significantly better than chance, so Newcomblike reasoning applies. But said Being certainly isn’t omniscient. You have access to a randomizer the Being can’t predict. (You can always do this in principle, for any physically realizable Being. For a sufficiently weak predictor, ordinary dice will suffice, but even vs a Being with a Jupiter Brain in his back pocket, a QRNG is sufficient.)
It’s logically rude to take the ordinary dice from Game B and the Being that knows the wavefunction from Game A, and pretend that’s the only game.
Game B is a weakened version useful to think about since it would also apply to superintelligent AGIs. We can use the strategies developed in Game A to help us think about Game B. We can use it to help develop and test a good decision theory. Does your decision theory handle Game B properly? If so, great! More evidence it would work in the real world. If not, you should update and rethink your decision theory.
Since “your decision theory” may also be a cow, note that I never defined what your decision theory is, not here, not in the OP. I also never said your decision theory necessarily fails this test. I said,
Having proven a strategy superior to one-boxing, I can claim that if your decision theory just one-boxes without pre-committing to use quantum dice, something is wrong with it. [added emphasis on if]
Of course, this claim is for Game A!
Now have we sufficiently established what we both think Pluto is, or is there more you wanted to say? Or do I have to Taboo more cows?
Look, surely in your diagram there is more than just the one fork, right? You could have a heart attack, get struck by a meteor, commit suicide, take a sudden vow of poverty or whatever. Point is, there’s a zero box fork, right? So what does the cow do when you will zero box?
See the trick? The whole one-box or two-box was a false binary all along. How has the everett fork where you had your heart attack play out? No quantum dice required, the cow has always been a fraud!
Except not, because it’s just a logic puzzle. It doesn’t need to consider the fork where you zero box. You are given as a profit maximizer. The cow is given as able to discern your future actions (including futile efforts at randomization). These are just parts of the question, same as the jail being inescapable in prisoner’s dilemma.
It feels like you are circling (grazing?) around to being right. Like, earlier when you got to “In that case, you one-box.”, you were there. ‘That case’ is the base case, the case that we all mean when we say ″Cow’s problem’.
“They just have to be unpredictably random.”
eyeroll
If Omega can’t predict dice, then he isn’t Omega. How can you spend so many words on this without getting it? If your ‘solution’ to the problem involves Omega being unable to predict whether you one or two box, because of dice or anything else, then you aren’t talking about the same problem as everyone else. I’ve pointed this out like 4 times. Oscar and Mr Mind pointed it out too.
More logical rudeness!
facepalm
Duh. You’re the one not talking about the games in the OP! I was talking about certain very specific modifications to Newcomb’s problem (Starting from the formulation I quoted from the wiki at the top of the OP, not from Nozick’s paper!). I even put “cheating” in the title, as in “breaking the rules” and also in the sense of “cheating on a deal”. I used both of these senses in the OP:
(Cheating a deal.)
(Breaking the rules.)
I also spoke of several alternative ways the game could be modified, both in the OP and in this thread. Don’t pretend I said they’re all the same game! That’s logically rude.
Taboo “Newcomb’s Problem”. It’s clearly biasing your thinking if anyone is disrespecting your sacred cow. We’re talking about The Deal. While we’re at it, Taboo “Omega” too, in case he is also a cow.
###Game A We’re talking about The Deal in a MWI universe. There’s a near-omniscient Being offering the deal. Wait, Taboo “omniscient”, it’s probably ill defined, like everything else about God. Said being knows the wavefunction of the entire universe of discourse (if that’s even consistent), including both the Chooser and the dice, and can predict how it will evolve. This means the Being knows the whole future. The being CAN predict the outcome of a quantum coin flip. Look! He even drew a picture for you:
The Being correctly predicts both Everett branches. Both a Head and Tail happen. Remember, I said “MWI”. Both branches are equally real. But the Being can either fill the box, or not, before the flip, per The Deal. That any given Chooser copy only sees one “random” outcome is a subjective illusion. This is not random. MWI is deterministic.
The optimal strategy for the Chooser in Game A is to use a QRNG to one-box with just over 50% measure, and two-box with just under 50% measure.
The optimal strategy for the Being before the split (predicting this) to maximize the measure of accurately set-up boxes after the split, is to put the million in.
If you want to dispute Game A, because I missed some important logically valid reason, I’d like to know what that is, but it’s certainly nothing you’ve said so far. And call it “Game A” to be clear.
###Game B We don’t know if we’re in MWI or not. There’s a different Being offering The Deal. This being has a track record of predicting significantly better than chance, so Newcomblike reasoning applies. But said Being certainly isn’t omniscient. You have access to a randomizer the Being can’t predict. (You can always do this in principle, for any physically realizable Being. For a sufficiently weak predictor, ordinary dice will suffice, but even vs a Being with a Jupiter Brain in his back pocket, a QRNG is sufficient.)
It’s logically rude to take the ordinary dice from Game B and the Being that knows the wavefunction from Game A, and pretend that’s the only game.
Game B is a weakened version useful to think about since it would also apply to superintelligent AGIs. We can use the strategies developed in Game A to help us think about Game B. We can use it to help develop and test a good decision theory. Does your decision theory handle Game B properly? If so, great! More evidence it would work in the real world. If not, you should update and rethink your decision theory.
Since “your decision theory” may also be a cow, note that I never defined what your decision theory is, not here, not in the OP. I also never said your decision theory necessarily fails this test. I said,
Of course, this claim is for Game A!
Now have we sufficiently established what we both think Pluto is, or is there more you wanted to say? Or do I have to Taboo more cows?
Look, surely in your diagram there is more than just the one fork, right? You could have a heart attack, get struck by a meteor, commit suicide, take a sudden vow of poverty or whatever. Point is, there’s a zero box fork, right? So what does the cow do when you will zero box?
See the trick? The whole one-box or two-box was a false binary all along. How has the everett fork where you had your heart attack play out? No quantum dice required, the cow has always been a fraud!
Except not, because it’s just a logic puzzle. It doesn’t need to consider the fork where you zero box. You are given as a profit maximizer. The cow is given as able to discern your future actions (including futile efforts at randomization). These are just parts of the question, same as the jail being inescapable in prisoner’s dilemma.
It feels like you are circling (grazing?) around to being right. Like, earlier when you got to “In that case, you one-box.”, you were there. ‘That case’ is the base case, the case that we all mean when we say ″Cow’s problem’.
It’s not a cow, it’s a bull :-D