Newcomb's problem

3,292 Views | 59 Replies | Last: 7 yr ago by BusterAg
Zobel
How long do you want to ignore this user?
AG
Aside from the basic injustice of the predictor being right about one person or the other and them getting disparate results based on them being them...the real bummer is the disparity of outcomes of the predictor is wrong.
Star Wars Memes Only
How long do you want to ignore this user?
MQB: There's a few ways this can go down.

If the predictor has knowledge of your choice then the choice and the prediction are causally connected and your strategy is optimal. If this is the case logically the prediction is dependent on your choice, whatever contrivance we you use to make this work. However, this is not the only way a predictor could make good predictions about your choice.

Consider an external set of variables that are highly correlated to your choice. Maybe a person who wears a blue shirt chooses A & B 80% of the time, whereas a person who wears a red shirt chooses A & B only 20% of the time. Maybe the shoes a person wears has a similar correlation. Keep going adding whatever things you can think of, and then let the predictor know the joint probability distribution given these variables. All of a sudden you have a predictor making highly accurate predictions without needing the predictions to be dependent on the choice at all. In this case, the optimal strategy is go A&B every time. When I originally read the problem I assumed something like this, because this doesn't come up against the philosophical problem of retrocausality, or other such things.

If you want to insist on the predictor having some form of foreknowledge of your choice then you are correct in your strategy, of course. But it's a very unnatural assumption because of the aforementioned problems, and if this is the case it seems kind of like a gotcha question.

BusterAg
How long do you want to ignore this user?
AG
dargscisyhp said:

MQB: There's a few ways this can go down.

If the predictor has knowledge of your choice then the choice and the prediction are causally connected and your strategy is optimal. If this is the case logically the prediction is dependent on your choice, whatever contrivance we you use to make this work. However, this is not the only way a predictor could make good predictions about your choice.

Consider an external set of variables that are highly correlated to your choice. Maybe a person who wears a blue shirt chooses A & B 80% of the time, whereas a person who wears a red shirt chooses A & B only 20% of the time. Maybe the shoes a person wears has a similar correlation. Keep going adding whatever things you can think of, and then let the predictor know the joint probability distribution given these variables. All of a sudden you have a predictor making highly accurate predictions without needing the predictions to be dependent on the choice at all. In this case, the optimal strategy is go A&B every time. When I originally read the problem I assumed something like this, because this doesn't come up against the philosophical problem of retrocausality, or other such things.

If you want to insist on the predictor having some form of foreknowledge of your choice then you are correct in your strategy, of course. But it's a very unnatural assumption because of the aforementioned problems, and if this is the case it seems kind of like a gotcha question.
It's designed to be a gotcha question. The big issue here is that we don't know for sure whether or not the predictor actually has foreknowledge or not.

Maybe you are right, and the predictor is just using a set of highly correlated values that a person with a personality who will take only the $1 million can be identified with extreme precision. The argument goes that making the decision to be a risk taker in the game gives the predictor some sort of tell that helps his prediction.

To make this more concrete, pretend that the predictor is a highly sophisticated brain scanner based on yet unknown technology that, while it can't read your mind, can detect the arrangements of ganglia in your brain to know how your brain will analyze this decision tree. Based on this scan, it will predict what you are going to do. If you decide you are going to take your chances before your scan, you will win. If you decide that you are going to change your mind after the scan, you will lose, as the brain scanner will pick that up to. The only way to win is to decide right now to take the gamble, putting your faith into the brain scan being right.

But, the point of the exercise is partially that you don't know how the predictor does it. You just have past performance to rely on, without an ability to understand how he is so good at it. You are either convinced that your brain is pre-programmed to make the choice, and that programming can be picked up, or the predictor has some unknown super-natural powers, and rely on past performance, OR, you are not convinced, you have no way to see how these events can be causally connected, and you take the safe choice.

Which set of evidence holds more weight with you? Past performance, or your ability to logically assess the situation?
BusterAg
How long do you want to ignore this user?
AG
k2aggie07 said:

Aside from the basic injustice of the predictor being right about one person or the other and them getting disparate results based on them being them...the real bummer is the disparity of outcomes of the predictor is wrong.
Well, if you really want that $1 million, just convince yourself that the predictor is smart enough to know what you are going to pick, and your outcome will improve!
Star Wars Memes Only
How long do you want to ignore this user?
I think mostly what you've said agrees, or at least doesn't disagree with what I've said.

Given what I know about our current state of technology, the most logical choice to maximize your gain is to pick A&B.
Martin Q. Blank
How long do you want to ignore this user?
dargscisyhp said:

MQB: There's a few ways this can go down.

If the predictor has knowledge of your choice then the choice and the prediction are causally connected and your strategy is optimal. If this is the case logically the prediction is dependent on your choice, whatever contrivance we you use to make this work. However, this is not the only way a predictor could make good predictions about your choice.

Consider an external set of variables that are highly correlated to your choice. Maybe a person who wears a blue shirt chooses A & B 80% of the time, whereas a person who wears a red shirt chooses A & B only 20% of the time. Maybe the shoes a person wears has a similar correlation. Keep going adding whatever things you can think of, and then let the predictor know the joint probability distribution given these variables. All of a sudden you have a predictor making highly accurate predictions without needing the predictions to be dependent on the choice at all. In this case, the optimal strategy is go A&B every time. When I originally read the problem I assumed something like this, because this doesn't come up against the philosophical problem of retrocausality, or other such things.

If you want to insist on the predictor having some form of foreknowledge of your choice then you are correct in your strategy, of course. But it's a very unnatural assumption because of the aforementioned problems, and if this is the case it seems kind of like a gotcha question.
You proved my point: we don't know how the predictor does it (foreknowledge, shirt color, reading this forum, reading your mind, etc.), we just know he's good at it. We could come up with theories all day long and try to outsmart him, but in the end, he is exceptionally skilled. I take that as 90, 95, 99% of the time he will predict correctly. So I choose Box A because more likely than not he would have predicted that and not both boxes because he would have predicted that too.
kurt vonnegut
How long do you want to ignore this user?
AG
I still say 'exceptionally skilled' is nebulous.
BusterAg
How long do you want to ignore this user?
AG
kurt vonnegut said:

I still say 'exceptionally skilled' is nebulous.
It's meant to be nebulous.
kurt vonnegut
How long do you want to ignore this user?
AG
BusterAg said:

kurt vonnegut said:

I still say 'exceptionally skilled' is nebulous.
It's meant to be nebulous.


Is the OP supposed to be one of those questions where there isn't really a right or wrong answer, but how you answer it is supposed to reveal something about yourself?

Like what would you do if you had a million dollars.

BusterAg
How long do you want to ignore this user?
AG
kurt vonnegut said:

BusterAg said:

kurt vonnegut said:

I still say 'exceptionally skilled' is nebulous.
It's meant to be nebulous.


Is the OP supposed to be one of those questions where there isn't really a right or wrong answer, but how you answer it is supposed to reveal something about yourself?

Like what would you do if you had a million dollars.
Pretty much. It gets talked about in philosophy circles because most people believe that there is a correct answer, and have a high conviction that their answer is the right answer, and that people that disagree with them are wrong. What makes that interesting is that the number of people who would pick both boxes is close to 50%. So, almost half of the world is very sure that almost half of the rest of the world is wrong, and vice versa.

But, if you break it down, your decision comes down to how you weigh the evidence. Is past results or your ability to understand the mechanics more important?

It is purposefully set up that way.
BusterAg
How long do you want to ignore this user?
AG
AstroAg17 said:

Two boxes at the same time
This is funny. Well done.
Drum5343
How long do you want to ignore this user?
AG
AstroAg17 said:

Let's replace the predictor with an omniscient god, let's replace the million dollars with everlasting life, and let's replace the thousand dollars with sinful pleasures.

If you take the earthly pleasures, god will know and you won't get box A. If you plan on box A and put aside earthly pleasures, you'll get it.

We've just replaced the uncertainty of whether the predictor will correctly guess your choice with the uncertainty of whether such a God exists. The OP was secretly Pascal's wager.


I think this is a false dichotomy.

I don't have to give up joy in order to follow God. I do, however, have to give up porn, adultery and other such things which bring no joy.
kurt vonnegut
How long do you want to ignore this user?
AG
Porn and adultery are the first and only examples you can think of earthly pleasures that might give someone joy?
Drum5343
How long do you want to ignore this user?
AG
kurt vonnegut said:

Porn and adultery are the first and only examples you can think of earthly pleasures that might give someone joy?


Certainly not the only. But what earthly pleasures, that bring lasting joy, do I have to give up in order to follow God?
Post removed:
by user
Drum5343
How long do you want to ignore this user?
AG
All I was pointing out was that you presented a false dichotomy.

Take nothing now and receive everything later is not the Christian life.
kurt vonnegut
How long do you want to ignore this user?
AG
Drum5343 said:

kurt vonnegut said:

Porn and adultery are the first and only examples you can think of earthly pleasures that might give someone joy?

Certainly not the only. But what earthly pleasures, that bring lasting joy, do I have to give up in order to follow God?

I won't pretend to know the specifics of the particular version of God you believe in, but likely the list of earthly pleasures you are expected to give up would include wealth, power (in some senses), self righteousness, pride, gluttony, greed. . . . there are a lot of examples outside of sex. Again, I only find it curious that only the sexual pleasures are the ones that come to your mind.
Post removed:
by user
Texaggie7nine
How long do you want to ignore this user?
I think the biggest variable to need to know for me would be what the sample size of the being "always" getting it right and how many chose both. It would need to be enough that would demand that my current logic cannot come up with a satisfactory explanation other than the predictor having some ability that defies logic.

If that is the case, I would also take the choice that defies logic which would be to only pick B.
7nine
BusterAg
How long do you want to ignore this user?
AG
Texaggie7nine said:

I think the biggest variable to need to know for me would be what the sample size of the being "always" getting it right and how many chose both. It would need to be enough that would demand that my current logic cannot come up with a satisfactory explanation other than the predictor having some ability that defies logic.

If that is the case, I would also take the choice that defies logic which would be to only pick B.
Yeah, you are a B guy.

The point of the game is to see if you would defy logic.

I guess you could probably come up with an expectation / utility curve, as the number of guesses and accuracy of the predictor increases, where is your indifference point.

For me, it is probably pretty low. If like 10 people had gone before, half picking both boxes, and the predictor was right 90% of those times, I'm close to even-steven on which way to go. We get to 20, with 90% accuracy, and I'm likely going box B. Less than 10 attempts, I am probably taking both boxes.
Post removed:
by user
Texaggie7nine
How long do you want to ignore this user?
I think if there were anything less than 100% accuracy I probably would pick both. But the article says ALWAYS. If it was any times that someone picked both and the predictor was wrong, then I would probably rule out any type of outside of time perception. That would mean that what I pick after the boxes are already set would have no affect on what the predictor decided in the past. I would need to be convinced that my choice could actually affect the past decision.

Anything less than 100% means that the predictor is probably using some really reliable trait or identifier of a person. So my decision would have absolutely no bearing on what's in the box, therefore the best odds are to pick both.
7nine
Drum5343
How long do you want to ignore this user?
AG
kurt vonnegut said:

Drum5343 said:

kurt vonnegut said:

Porn and adultery are the first and only examples you can think of earthly pleasures that might give someone joy?

Certainly not the only. But what earthly pleasures, that bring lasting joy, do I have to give up in order to follow God?

I won't pretend to know the specifics of the particular version of God you believe in, but likely the list of earthly pleasures you are expected to give up would include wealth, power (in some senses), self righteousness, pride, gluttony, greed. . . . there are a lot of examples outside of sex. Again, I only find it curious that only the sexual pleasures are the ones that come to your mind.


Well, to be honest it's the first that came to mind because that's what I personally have to "give up". I don't particularly struggle with wealth, power, gluttony or greed, at least at the current time. Maybe a touch of self-righteousness.
Drum5343
How long do you want to ignore this user?
AG
AstroAg17 said:

In certain settings that is the choice Christianity poses. Take a gay person again as an example. He may have to choose between obeying God and the person he loves.


He certainly might.

But that doesn't mean he has to gamble between "no joy now and all joy later" or "some joy now and nothing later".

Pascal's wager, I think, is bunk. No one has ever chosen faith because of it or anything like it.
BusterAg
How long do you want to ignore this user?
AG
logic meaning your ability to understand how it works. See my blue diamond post on page 1.
Refresh
Page 2 of 2
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.