A Problem You'll Never Solve 

Here's Robert Nozick's paper if you'd like to read more about this problem. faculty.arts.ubc.ca/rjohns/nozick_newcomb.pdf

Me: Yes, obviously both Kevin: A magical genie predicted that and makes B worth 0 Me: That seems like it was an important part of the setup

*Chooses only box A* Genie: Wait, that's illegal

Kevin: Will you choose box B or both boxes? Me: Both Kevin: Now let me introduce the genie Me: well fu too then

it's kind of a flawed question though, right? It really depends on how smart the genie is, if he is omniscient then choosing the second box is always the best choice.

I choose box A. I'm not interested in the Genie's bullsh1t.

vsauce: or is it? vsauce 2: WRONG

The real question is, why would I really need 1,000,000 candies.

The issue with strategic dominance in this scenario is that it has a fixed view of time, whereas in this situation, your choice has some effect on what is in the mystery box despite the contents already being decided.

Kevin: you will take both boxes right? Me: *knowing kevin* WRONG Kevin: RIGHT Me: right? Kevin: Wrong

Kevin the type of guy to actually count out 1000 candies.

The problem is also that it’s never really made clear if the participant *knows* about the genie’s powers, or if the scenario is presents as it was in the first bit of the video. And if you know, can you outsmart the genie by thinking really hard about choosing one option and then switching suddenly? Can you mind-battle the genie? Or are you unaware of the genie the whole time?

It always comes down to the accuracy of the prediction, if we know how often he is right we can do the math and figure out the best choice (box b method). If we dont know how often he is right we might as well pick both of them, since he could always be wrong. The aperant contradiction lies in the question that omited this crucial information to solve the problem. Therefore both explanations are correct given their assumptions and there is no contradiction

I choose only box A. There is clearly not enough room for 1,000,000 candy in box B!

The problem is that original question contains no information about a genie.

I’m mad there wasn’t a poll

It's a perfectly all-knowing genie at first, right? So the whole "but he's already put the candy in or he hasn't" premise doesn't hold. It breaks causality, therefore is no longer a game between the genie and me, but just a choice for me.

Take only box A. Confuse and disturb the Basilisk.

11:00 Genie's a freakin' liar, he put 18 candies inside the mystery box

I think the expected value equation can be used to prove either case. If the genie is right most of the time, it is better to take just box B. If the genie is wrong most of the time, it is better to pick both. So I don't think this is so much an unanswerable question as it is a question that does not provide enough information to come to a provably better solution.

Take only Box B and hope it has no candy in it. (if you don't like jelly beans)

Host: "For tonight's show, we have two boxes, one contains $1000, and the other one containing either nothing or $1.000.000! Now, guest, would you like the second box or both of them? " Guest/Me: "Hold on a sec, I need something to write on. Also, how psychic are you? Like, what percent exactly?"

The true problem: the candy may have been in a box for a week or more

The paradox stems from a temporal self reference paradox which invalidates strategic dominance. The genie's foresight is impossible, but if taken as a given, it makes picking just B obviously correct

I saw this problem elsewhere and a good portion of my grade in school was debating this question. I am on team both boxes: because if box b was glass, you would take it either way; and it doesn’t matter if it is opaque because of this.

How can we make the right choice when you keep adding new conditions

Both. Choosing only box B has a possible outcome of $0. Taking both has a minimum outcome of $1000. I'm not gonna deal with a Genie for free.

I’m seeing this problem as having three different results: loss(choosing box B when it’s empty), small win(choosing both and box B is empty), big win(choosing both and box B has a million dollars). Choosing only box B is the only way to actually lose, so to me it makes sense to choose both as you’re guaranteed a win(even if it might be small, at least you’re not losing).

I think the real solution to this is realizing that when you don't specificy how the predictions work; the math will get confused

The only reason there is a question of which to take is because the presenter keeps changing the rule for how the contents of the mystery box are determined. At first, it is implied to just be independent of the player. Then it is change to be a genie that bases the contents off of what he believes the player will pick, which is no longer truly independent. But when the viewer gets pushed towards the idea that it is best to take only the mystery box, the presenter starts pushing that the mystery box was determined long ago. Then out of nowhere the presenter introduces the idea that the genie is only right 90% of the time. If you keep changing the rules to suit your purpose, you can make any problem into "a problem you'll never solve".

"B only" seems based on trusting the genie can make an accurate prediction while "both" seems based on trusting the genie can't actually predict the future.

I am on the team Expected Utility. I feel like the only reason people choose Strategic Dominance is that they are too arrogant to believe their actions can be predicted with near-perfect accuracy.

I'll just take the Genie.

"This question is actually a lot less simple than it seems... because here are completely game-changing additional parameters I didn't mention before." Eye roll.

THE GENIE IS FOOLING US!! that mystery box cant physically contain a million skittles...

It always comes down to the accuracy of the prediction, if we know how often he is right we can do the math and figure out the best choice (box b method). If we dont know how often he is right we might as well pick both of them, since he could always be wrong. The aperant contradiction lies in the question that omited this crucial information to solve the problem. Therefore both explanations are correct given their assumptions and there is no actual contradiction.

The interesting part of this experiment is that... Really, the best option for each individual is their gut instinct. These predictive factors have to be based on who we are as people for them to be remotely accurate... So someone like me, who would take both as a loss aversion strategy, should double down on that decision because the prediction would LIKELY be accurate and screw me over for deviating. Same is true if you're the gambling type: odds are that you're walking away with a million, with a 10% chance to just get unlucky. Of course if you're the type to get discouraged and take the safe bet of guaranteed 1000 dollars, then the genie would have likely bet on you taking both anyways.

Me: I am choosing box B Genie: I knew you say that- Me: So I'll take both Genie, whispering: what now

Did half the people watching this really think the best valued answer would be to choose just the mystery box instead of both before he brought the genie into it?

I think this just comes down to how accurate the genie is. If the problem states almost always right, then b is the way to go.

This problem works better when you say "super computer" rather than Genie. Also if the super computer has just 51% accuracy, just Box B is the simple choice.

This doesn't seem like a paradox to me, it's more like a problem without enough information. It's really just guessing how well can the genie predict your anwser.

5:03 who else thought he was going to say popcorn

Two contingencies here and the conclusion that one who is free to choose should choose both. Genie right: Practical to assume in the context of this contingency that you were predestined to get $1000 or $1M after his decision w.r.t. the mystery box's content, and the consequence of this predestination is that "reasoning" between 1 option and the other doesn't matter. If you are "tempted" mentally to go against what Mr. Genie predicted, you will somehow end up changing your mind. Genie wrong: You go against the prediction; in this admittedly less likely (but nonetheless relevant) case, it is clear that what is best is this: going for both boxes after a prediction of your going for just the one box. (The alternative is going for the one and feeling a sense of loss because of the content that was mistakenly omitted from it.) "Choosing correctly" only is a real issue in the event that it will turn out that the genie will be exposed as fallible, after a choice by you that goes against the prediction. The chance of the genie's being wrong slightly tips the scales of superiority in favor of choosing both boxes:) --- QED.

here my take: theres a 100 percent chance that if you pick both, you will get one thousand candies or one million plus a thousand candies, ethier way is good so you should pick both. the worst possible outcome if from picking ONLY box b and getting 0. but if you choose both theres a 0 percent chance you will get 0 candies. so yeah pick both

I'm on team "how accurate does the genie have to be to make the expected value of both choices equal?!" *leaves to go do the math*

Answer: The mystery box is too small to have 1 000 000 candies so it has 0

Plot twist: Box B contains the same amount of candy, but each piece is broken into 1000 smaller pieces.

well, tbh its not a paradox. a) the genie prediction is always 90% accurate then u chose only box b. b) its random what he chooses then u choose both boxes bcs the genie is nothing more than a 50/50 coin. u just have to specify the genie in the first place

Choose only Box A and the genie will be so impressed he'll just give you the million.

1,000,000 candies has 1,000 x the volume of 1,000 candies. This won't fit in the mystery box. Therefore the genie is lying. Problem solved

Since the genie predicts with “near perfect accuracy”, we can assume that the genie predicts correctly more than 50.5% of the time (and likely much more). Since 50.5% is the threshold at which picking box B becomes advantageous, and the genie exceeds that threshold, it makes no sense as to why you wouldn’t pick solely box B. The alternative argument makes no sense. Yes, what’s there is already there and won’t be altered in the moment due to your choice, but the entire premise is built upon the fact that the genie predicts your choice and fills (or doesn’t fill) the box accordingly. That “solution” completely ignores the condition that the genie actively plays a role in the outcome. It would only make sense under an assumption that the genie has a 50/50 chance of guessing right.

The real issue here is with the genie. The paradox arises from the issues with multiple agents. It’s somewhat related to the prisoner’s dilemma

Now there are two problems I'll never solve. This and getting Karen to let me see the kids again.

Seeing him about to eat the candy but then not is the most aggrivating thing.

that last rhyme at the end made me chuckle!

this whole thing basically boils down to whether you believe the genie or not. the logical dominance solution is basically thinking the genie is bullsht, while expected utility is basically thinking the genie will predict your choice

Me: I studied for this test I’ll be fine. **opens test**

Wait but who says that the genre's accurate's is 90%?

it depends totally on wether the genies guess is unchangably accurate or not, if so then you essentially get to decide what litteral reality is in terms of what is inside the box, which is not determined until you open the box since there is no way it could be if the genie will always guess right no matter what

I don't see how this is a paradox if you just understand the definition of omniscient. It isn't a prediction. He KNOWS which one you'll pick. The prediction will always line up with the real outcome under the premise of an omniscient being.

So you ask a question first and then add extra rules after everyone answered?

Me : Has lots of work to do. Also me : Let's watch a problem I can never solve.

Nice little brain teaser there ! Enjoyed this video, thank you Kevin !!!

"one half of you is certain the obvious answer is to take both boxes" "The other half just as sure to take just the mystery box" ... No. I'm sure to not be sure what to take at all xDD

But who decided the genie has a 90% chance of being right? That's the thing that's skewing the results

4:41 "What exactly is going on here?" Schroedinger's candy... no, wait.

One thing I think is interesting is that the expected value calculation is specifically calculated before the boxes are filled. The higher expected value is not in taking only Box B, but in *intending* to only take Box B, such that you're likely predicted to do so. When you get to the actual event of deciding, the expected value is clear; you get X, or 1000 + X, where X is 1,000,000z, for some probability z of Box B having a million. It falls in line with the dominance strategy. The crux of the matter is that, obviously, whatever you do in the present can't affect something that happened in the past. But I'd only take Box B. Because while what occurs in the present can't change the past, I think it's wrong to conclude that there's no causal relationship. There is one, there's a shared cause. What you decide to do in the future isn't an instantaneous choice. It's a culmination of your thought about the scenario, the strategy you adopted, the desires you've had, all of your thinking and existing since becoming aware of the scenario, even more broadly just a consequence of you being you and then eventually finding yourself in that scenario. If the genie or AI or what have you predicts your actions by acquainting itself with you, your personality and thoughts, in some magical way, and then predicts your future decision through simulation... Well then, whatever intention you have will affect the outcome. But whatever intention you have will also affect your decision; that's, y'know, how intentions work. What we were intending to do affects what we actually do, in the event. Of course, it is possible to act differently than how you previously intended. Intentions have an influence, but they aren't the only influence. So the question becomes, do you think you can honestly intend to take only box B, and then after a point change your mind, in a way where this change of mind isn't predictable? I don't think I could. I think my decisions are culminations of longer processes which are (hypothetically) predictable, and if I planned to change my intentions that would obviously contradict that I had those intentions. So, I'd honestly intend to take only Box B. And in the event, I think I'd take only Box B, because that's what I'd do if I was really intending it, not just thinking I'd intend it. So, even though obviously the contents of the boxes are already set by the time I actually take one or both, and I'd be better off taking both, I've already committed to my decision process to take one. If I take both, it means I didn't actually have a decision process to take both to begin with. With that said, with a 90% accuracy rate I might take the certain $1,000 and chance for a failed prediction with $1,001,000, rather than a 90% chance for $1,000,000 but one in ten chance I get nothing, but that's like, a risk aversion thing. Which is also cool but not so much something to be philosophically pretentious about in a comments section.

If the genie is 100% correct, you can choose both and get an extra cool looking box.

Those are not even the same problem. Why is there suddenly an omniscient genie in the problem? If there's a 50/50 chance of $0 or $1,000,000 then taking both would always be the most beneficial.

of course it has 0 candies in it the box isnt big enough

You were right in that one video the more you think about something the more it gets harder to choose

Setting up an equation: 10⁶X = 10³ + 10⁶(1 - X) where X is the probability of the genie being correct we get that the genie has to have an accuracy greater than 50,05% for it to be worth choosing only box B. If you do not beleive in genies then you should assume that he will be correct 50% of the time and therefore choose both boxes. In contrast, if you do beleive in the genie, i think it would be fair to assume that he has a greater accuracy than 50,05% and therefore, if you beleive in him then you should pick only box B.

This problem is flawed. Normally without the genie, picking box a and box b is the clear logical answer. However, factoring in that the genie has already predicted what you will choose *with near perfect accuracy*, taking only box b is the correct choice. The two answers that "contradict" each other are really just answering two different questions.

Hmmm, Box A or Box B? Take the genie

As the person sitting in front of the boxes ready to make my choice am I aware that a robot/genie/whatever has made a prediction with 90% accuracy of my choice?

The trick is to be confident you will pick just B. Then pick both

The expected utility is only this high because they set the genie's accuracy to 90%. If it was 50%, both boxes would be the best choice.

Genie's don't exist, always pick both. How's that for some simple math

The genie always knew which box you would choose.

I’d only choose box A I don’t need that many candies

Plot twist, you cant fit 1 million candies in box B

I took only Box B and it contained Schrödinger's cat.

My question is how does this genie know what my choice is before I make it? I like the math behind the expected utility but that’s assuming that the genie is right 90 percent of the time. How do we know that’s true?

i get bored i open youtube Kevin gives me an existential crisis no longer bored i exit youtube

The moment you said "genie" my D&D instincts kicked in and said neither box run

the question really comes down to, do you trust the "genie" to actually be clairvoyant. If you do, then pick Box B, If you don't believe the genie has the ability to know the future, then pick both. In this strictly hypothetical situation, i'll assume the Genie, being a super natural being that exists in this situation, is in fact omniscient and i would pick only Box B. If this was some guy on the street claiming to be a genie in real life, i'd pick both.

Kevin: Simple right? Me: Wrong Kevin: Right Me: Wtf Kevin: wrong Me: huh Kevin: maybe Me: do u know the answer? Kevin: I honestly don't know Me: ok then

0:18 I’ll take both because the mystery box definetely can’t fit a million candies so it has 0, so really might as well take a 1000

I'm taking both boxes literally every time

"Right" "Wrong!" "Maybe?" "Honestly. I don't know"

Kevin: "blah blah blah among us blah blah blah... right?" Me: "Oh boy, here we go again"

At 1:30 the problem states that the genie can predict what we'll choose with near-perfect accuracy. It's easy to forget this by the end of the video. This changes the question from, "which strategy should I use if the genie can be right or wrong?" to "what strategy should I use if I assume the genie is right?" I wouldn't say 90% is near perfect; let's up it to 99.99%. If the genie is essentially always correct, there's no point in trying to outsmart it by declaring that the genie has already made his choice and so you should collect the extra 1,000, because that would have already been part of the genie's prediction. Essentially, the genie can go back in time.

The answer is flip a coin before your brain explodes

A problem I can never solve? Why Micheal never uploads

This math problem is pretty cool and accurate, but I do see one supposed flaw (i might be wrong): you can’t know the accuracy of the genie, making it to where depending on the percentage you choose, you choose the reasoning in the expected utility calculation and therefore change which choice is actually better. I am just a kid who is trying to comorehend this problem, so if anyone sees any flaw in my reasoning please point it out :)

The problem with this question is that it cannot be predicted what I will choose. So in that case the options are either just box B or box B + 1000, which is an obvious choice, as box B + 1000 is obviously more. The Genie however, can predict the future, therefore if you choose to take both boxes he will HAVE to have predicted that and put 0 candies in box B. Yet again if you choose only box B, the Genie will HAVE to have predicted that as well and put 0 candies in box B. Regardless of what you think, you can only have one final answer, which, in this case, is the same as repeating an answer. It is like you have already told the Genie what you will choose and no matter what happens you cannot change the answer this time around. Since the Genie already knows your answer he will have either put 0 or a million candies in box B, depending on your answer. From that, you know, taht if you choose both boxes, there will be no candies in box B whatsoever. And if you choose only box B, there will be a million candies in the box. The answer to this question has already been provided to you and you just have to think logically.

You say the guy has already decided what's in the box so it makes zero sense to me to not choose both.

Seems obvious to me: Box B if it's a Genie. Both boxes if it's a person

"The boat's a boat, but the mystery box could be anything! It could even be a boat!" -Peter Griffin