You didn’t properly understand the scenario. I didn’t screw it up.
The downside to taking both boxes is that if the computer predicted that you would take both boxes it will have put $0 in the second box. And since the computer successfully predicted whether the last 1,000 people took one box or two boxes, I think it is safe to say it probably predicted right for me as well and if I take the single box I will find $1,000,000 but if I take both I will find $0 inside the second box.
That seems completely clear and obvious so I can’t imagine why anyone would take two boxes.
I can only conclude that two boxers are rejecting the premise. If, in fact, the supercomputer can predict accurately, then the only correct answer is to take the mystery box alone. If you doubt the hypothetical, then you believe it won’t predict correctly, and you feel free to take both boxes. You think you can fool the computer into predicting incorrectly, and get an extra $1k out of it.
Whether or not your action changes what is in the boxes is irrelevant. The question is whether or not you trust the predictive ability. The computer can scan your past behavior and predict whether you will take one box or two. As time goes on this hypothetical becomes more and more believable.
Oh. You meant that people would take the one opaque box. And leave the thousand dollars on the table?
Yeah I, and I think most of us, are reading it as take the opaque box in addition to the thousand bucks in the hand.
The computer has and therefore highly likely will continue to predict correctly; it knows each individual players’ decision processes better than they do. So based on that it will likely predict your actions correctly. When you take the opaque box and only the opaque box you are spending a thousand dollars on a lottery ticket. Nah.
Yes, those are obviously the only two options. There’s no reason to take the $1,000 and leave the other box.
Agreed with you so far.
Spending a thousand dollars on a lottery ticket implies spending a thousand dollars on something with a very low chance of paying off. If the computer is so good at predicting that it has gotten the last 1,000 people right, it’s not much of a gamble for me to take the single opaque box. It’s spending a thousand dollars to be nearly certain that I’ll win a million.
And I would say, you are giving up $1m for an exercise in arrogance. You believe you have fooled the computer into giving you the million, without needing to take the inherent risk. And you think you are superior enough to achieve this with zero effort. Just take the two boxes and somehow your inherent superiority will make you the one person who fools the computer and gets the $1,001,000.
I already explained. The “dilemma” is faulty. It’s a cheat. The computer effectively possesses supernatural predictive powers. “Ah!” you might say, “But the hypothetical offers as a given that there is nothing supernatural, just a really good computer.”
Bullshit. The computer has precognition based on its track record, and saying it does not have supernatural powers is ridiculous. It’s like a thought experiment that includes a machine that works forever without any external fuel—“but it’s not a perpetual motion machine!” Uh, yes, it is.
So, I choose one box, but again point out that this scenario is impossible, even assuming a tremendously advanced computer.
If I play I’m taking the thousand for sure. And the lottery ticket because why not? I think there is a lottery ticket chance that anyone who takes the opaque box, either by itself or as part of the pair, has, in this hypothetical, been the one person the computer failed to predict. I’m not so special.
The outcome is not 50:50, but the choice is from the perspective of the person doing the choosing. Anyway, we’re splitting hairs here. I think the computer is a real distraction, because the game then becomes do you accept that the computer is always right or are you really just trying to fool the computer. Such is always the weakness of a thought experiment - reality too often intrudes.
What if my past has never included taking part in logic puzzles like this? What basis does the computer have for predicting how I will respond in a situation that I have never encountered before?
Has the mere reading of this thread jinxed me? Has the magic computer got my number now? Dammit!
Ah. Now I understand what you had meant your hypothetical. The “second opaque box” phrasing is the part that confuses. Got some of us to prematurely anchor on it as the added variable of taking it as well.
Understanding the intended set up now - yeah. The choices are: take the opaque box alone, a single box, and the computer knew you would and put the big payoff in it; take both and get a thousand if the computer was right and $1,001,000 if it is wrong. Yeah that’s just a dumb question.
The only interesting part is how easy it was to misunderstand it.
Presumably, the computer could predict that you will try to beat the computer by going random, but not the results of the coin flip. Since the computer won’t be able to guess the results, the answer it gives will have a 50% shot of being wrong.
So you could take a 50% shot at ruining the computer’s perfect record.
But now there are four possible outcomes, and since you are picking at random and you sabotaged the predictor, all four are equally likely:
You pick one box, computer thinks you picked one box - you get $1,000,000
You pick one box, computer thinks you picked two boxes - you get $0
You pick two boxes, computer thinks you picked one box - you get $1,1000,000
You pick two boxes, computer thinks you picked two boxes - you get $1,000
This seems to be all around worse than just picking one box and counting on the computer to keep being right netting you $1,000,000
This is my take too. If there had been 1000 people testing this before me, half of them one boxers and half of them two boxers and the computer picked accurately for all of them, it doesn’t matter if it possesses supernatural abilities, or if it’s just really good at data mining and/or scanning our brains, or if there is some other thing going on like sleight of hand behind the scenes. The logical thing to do is to take the one box.
It’s like the hypothetical scenario of betting on a roll of a die that came up 6 100 times in a row.
Joe schmoe: It will probably be 6, since it came up 6 before. It’s just a lucky die. (Or conversely, it probably will not be 6 since it’s “luck ran out”.)
Statistician: its still only a 1 in 6 chance of being a 6.
Canny Tony: Something’s up with the die. It’ll probably be 6.
Now, the calculus changes if they want us to pay money for a shot at the die being 6 or the million dollar box. Then, the most likely scenario was that the die rolls and/or box picking was performative and rigged to make us think that it worked in order to get us to bet on it.
Heh, yeah. The options are poorly explained. I just watched the vid in question. Your options:
Only take the mystery box.
Take both boxes.
I figured you could just opt to take the 1K, the mystery box, or both boxes. I still don’t know how the computer has arrived at its conclusion of whether I would take both boxes. So, in order to avoid leaving empty handed, I’d grab both. My choice doesn’t affect the outcome, and it is the only one where I’m assured the $1K. If the computer is accurate or it is inaccurate, I’m still getting my maximum payout.
I am still missing the advantage of taking only the mystery box. It’s already filled or empty before I walk into the room, and my choice makes no difference. Take 'em both, and see if the computer’s accuracy holds up this time.
I’m surprised at how many people appear to feel they’re a unique exception to an otherwise universal phenomena.
I don’t see that in myself. If a computer has demonstrated that it can correctly deduce the behavior of several hundred pervious people without error, I’m going to assume - based on this evidence - that it will correctly deduce my behavior.
So I have two choices: take one box and get a million dollars or take two boxes and get a thousand dollars. I’m going to take the million dollars.