Тёмный

QI - Game Theory 

FSMGauss
Подписаться 1,4 тыс.
Просмотров 407 тыс.
50% 1

Stephen was wrong about the strategy in the big brother game, saying you will take the money is weakly dominant. If you want to find out more about the weird wonderful and throroughly counterintuitive world or game theory then
Stephen Fry gently guides guests Liza Tarbuck, Phill Jupitus, Sean Lock and Alan Davies through the world of games, taken from series 7 episode 3, (Games).
ALL HAIL STEPHEN FRY!

Опубликовано:

 

9 дек 2009

Поделиться:

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист
Посмотреть позже
Комментарии : 415   
@MacSchluffen
@MacSchluffen 8 лет назад
When Alan tries to shoot at nothing he has a 90% chance on not hitting nothing wich should result in hitting one of his two opponents. Trust me I've done the math.
@Solarn40
@Solarn40 8 лет назад
Or Liza.
@DrDeFrE
@DrDeFrE 7 лет назад
so try to shoot the space which is everything except both of your enemies? I like it!
@Dhundahl
@Dhundahl 7 лет назад
Exactly what I thought! He has a 90% chance of missing if he aims for one of them, so he should just take the risk :D
@revmeister007
@revmeister007 6 лет назад
Mathematically that's right, but in reality you can miss if you want 100% of the time
@cogtroper
@cogtroper 6 лет назад
I've done the meth too.
@Richard_is_cool
@Richard_is_cool 7 лет назад
Alan is so intelligent. You can see it every now and then.
@hammeredshitsteak
@hammeredshitsteak 8 лет назад
I've heard an example of two criminals being interrogated by police, and both are offered a deal for a very short prison sentence, if they rat the other one out. If neither talk, there's no evidence and they walk free.
@andrew7taylor
@andrew7taylor 8 лет назад
+hammeredshitsteak - That's called "prisoner's dilemma", and it's a classic in game theory. Look it up, it's really interesting, though the options are different (in the example you have, they're nothing to be gained for them by singing). They are: If A and B each betray the other, each of them serves 2 years in prisonIf A betrays B but B remains silent, A will be set free and B will serve 3 years in prison (and vice versa)If A and B both remain silent, both of them will only serve 1 year in prison (on the lesser charge)
@CyPhaSaRin
@CyPhaSaRin 12 лет назад
thanks for putting the details in, was wondering what ep it was.
@Ligerpride
@Ligerpride 7 лет назад
It depends on the amounts involved, or should I say relative splits.
@cwpo1973
@cwpo1973 10 лет назад
The Big Brother scenario is a version of the "prisoner's dilemma" scenario. While sharing the pot is the optimal scenario, game theory teaches us that the standard outcome is non-optimal, namely with both competitors claiming the entire pot. A key assumption is that the competitors can't communicate with each other (don't know if that happened on Big Brother), and that sort of cooperation is essential for reaching an optimal, coordinated result.
@ktscongakikoi
@ktscongakikoi 12 лет назад
upload more of such videos that a great help
@andysm
@andysm 12 лет назад
@MrTerra5005 No. If you aim to miss there are 3 possible outcomes: 1=you hit the target (thing you aimed at) 2=Hit the man (who you were trying to miss) 3=hit something else (i.e. you hit neither the thing you aimed at nor the man). You can not determine the chance of hitting the man (outcome #2) since we do not have sufficient information, all we can say is that there is a 90% chance we will have either outcome #2 or outcome #3.
@mdtalley
@mdtalley 12 лет назад
@EebstertheGreat There's a very old PC game for the Atari called M.U.L.E. It is a four-way competitive resource game where, since the object is to evolve the colony, cut-throat strategy almost always leads to an individual victory, but ultimate defeat for the colony.
@diurdi
@diurdi 12 лет назад
As you see in the description of the video, it also says Fry is wrong. And QI corrected themselves that they were wrong. Game theory plans out what is the best choice for you. And in this case you have to options: 1. Choose to share, and you get £25,000, unless the other person chooses to "take the lot" in which case you get £0. 2. Choose to take the lot, and get £50,000, unless the other person chooses to "take the lot" in which case you get £0. Not hard to see that 2. is best.
@GORDO151515
@GORDO151515 12 лет назад
The individual's best response is as follows: if the other chooses to share then payoff is maximised by taking the lot, if the other chooses the lot you are indifferent, your payoff is 0 either way. Therefore the Nash equilibrium is both players taking the lot and receiving a payoff of 0. Without repeated games or communication, this is clearly the optimal strategy.
@TomTheDrummer
@TomTheDrummer 11 лет назад
Stephen is actually right at 4:50 the 'best'/safest strategy for both is to share but the result is more likely to be that both steal as 'Typhoontoby' correctly makes clear why.
@EebstertheGreat
@EebstertheGreat 12 лет назад
This assumes that if I change my reasoning, for some reason you will also change yours. But in game theory, if I decide to cooperate, that doesn't make you any more likely to cooperate. It is true that in real simulations of the iterated dilemma, machines which sometimes cooperate do much better than machines which never do, but this can be explained by the fact that they are playing against other machines using rules that do indeed depend on whether or not they cooperated in previous rounds.
@juuonse
@juuonse 12 лет назад
and it just occurred to me that we might actually be using the same words (betray/defect vs. co-operate) in two different ways, hence causing confusion.
@slytown
@slytown 12 лет назад
I think he was describing The Prisoner's Dilemma within Game Theory, which is the applied normative situation for the theory. Not sure what happened with Big Brother but he got the just of Game Theory right.
@848harris
@848harris 11 лет назад
absolutely correct typhoontoby, in a format where there is no past history between the 2 people and there is little chance of the scenario existing again, then steeling is the best format every single time, however, as im sure you are aware, in an iterated form of the prisoners dilemma (the scenario exists over and over again) then learning history comes into the game, so once one guy realises the
@grahambutler734
@grahambutler734 6 лет назад
Isn't the John Nash advertising example called equilibrium theory or something
@Jakester88
@Jakester88 13 лет назад
The two above are correct, as you either win all the money (best option), or prevent the other person from stealing it all.
@MyMeichan
@MyMeichan 12 лет назад
That awesome moment when you understand something on QI. All my university tuition in economics was worth it :'D
@vidfreak56
@vidfreak56 12 лет назад
Yes I agree. And game theory proves (assuming everyone follows it) that sharing (in this case) is the ONLY solution as both parties will always pick "take the lot" and both will lose everything. There's a great deal of emotion (greed for one) that also goes into this and is why its never always what you'd expect.
@Tiboroun
@Tiboroun 12 лет назад
The point is that even knowing how the operation works, and what the end result is, there is still a randomness to it. The absolute was introduced to eliminate the information you needed to have 100% certainty. Since it doesn't care about references to position you lose information by doing it; which is why I said it introduces randomness. Absolutes, by nature, eliminate referential information and by losing information you lose certainty and gain probability.
@nsg666
@nsg666 12 лет назад
True, it is not the best "overall" outcome i.e. not Pareto optimal. Yet given the others player's choice Betray is the only stable equilibrium. Furthermore, every dominant strategy is also the strategy that constitutes part of the Nash equilibrium, whilst NOT every nash equilibrium constitutes a dominant strategy. The nash equilibrium is thus more powerful than dominance. I guess you all knew this but just wanted to weed out any misunderstandings in the terminology ;)
@bizoppglobal
@bizoppglobal 12 лет назад
Really great video!
@Tomwithnonumbers
@Tomwithnonumbers 11 лет назад
If we assume the two people are thinking identically, (as in are both game theorists) then its sensible to assume that they will always choose the same response as each other, unless both have just decided to toss a coin
@OneWayTrip1
@OneWayTrip1 8 лет назад
volume drops off at 24 seconds
@verolizhimolimi854
@verolizhimolimi854 6 лет назад
OneWayTrip1 And here I was, adjusting my headphones 😂
@NidgeOSullivan
@NidgeOSullivan 3 года назад
Oh thank God, I was just on the phone to the local ear doctor, I thought I'd had a catastrophic aural incident...
@mgigkgeg
@mgigkgeg 13 лет назад
I went to see this episode live!
@2dTones
@2dTones 12 лет назад
The risk is the same for both options, but the payout is higher for 'taking the lot', so that's the better option.
@whade62000
@whade62000 12 лет назад
You're missing a bit of description, now I do not know where to go to find out more about the weird wonderful and throroughly counterintuitive world or game theory.
@spidaminida
@spidaminida 9 лет назад
I want to find out more about the weird wonderful and throroughly counterintuitive world of game theory! Then what??
@mikekenny1698
@mikekenny1698 8 лет назад
+micbenner I remember seeing this, they stopped the show pretty much after this happened because people were abusing it too much, they later had to put a clause in the contract that they weren't allowed to discuss about making deals outside of the show
@BobTheTrueCactus
@BobTheTrueCactus 2 года назад
The best strategy is to not share it, because no matter what the other says, your outcome if you do not share is never worse and at times better than if you said you will share it. If I don't share I get all or nothing. If I don't share, I get half or nothing.
@vidfreak56
@vidfreak56 12 лет назад
If neither know what the other will pick or choose, then assuming the other will choose to "take the lot" is bad because then youll either forfeit the amount or by choosing to share it, or by "taking the lot". You could try and guess if the other guy will choose to "share it" and then "take the lot", but that would require knowing the player and/or working it out prior. Sharing it is the "best" answer assuming both players know nothing of eachother. Tricky if they do know eachother though.
@MiltiadesOfMarathon
@MiltiadesOfMarathon 12 лет назад
Opting to take the money gets you a better outcome whatever the other person does. If they opt to share you get all the money and if they opt to take all of it it's the same material outcome as if you'd opted to share. The satisfaction of not having been screwed over can be considered a slight plus.
@vidfreak56
@vidfreak56 12 лет назад
1 is the answer to |x|=1 but that is just a scalar value. And 1 means that is a scalar quantity that it's a magnitude for both directions of -1 and 1. I can use a scalar quantity to define a magnitude and not a direction because x can be a vector. If x is equal to 200 and i take the absolute value of x, then the result is a scalar quantity. It says nothing about the nature of the original number itself. Its true that i dont know X necessarily, but if I can know it or guess it, its not chance
@CarfDarko
@CarfDarko 11 лет назад
Ofcoures ;p My fave interwebz serie. Best of all... I had the honor to created his new end tune ;)
@dm9910
@dm9910 12 лет назад
in the split or steal situation, no matter what the other player does, the optimal play is to steal: if the opponent splits it's better to steal because you get 50k instead of 25k, and if your opponent steals it doesn't matter what you do because either way you get 0 so you will probably steal just to spite your opponent. so you have the situation where individually the best outcome is to steal. however this way neither player will ever get any money.
@balrog0444
@balrog0444 12 лет назад
The difference is, I intended to share it after the game. If we were aware of having to make the choice beforehand, I would have told them of my plan. If not, I would simply hope that they too have not thought of my plan. "Thinking outside the box is easier if you don't have a box to begin with."
@vidfreak56
@vidfreak56 12 лет назад
This whole debate is about whats the best strategy and game theory. There is no randomness when the weak dominant strat is to always play "take it all". Eventually players who watch previous shows might realize that strat and then play "take it all" everytime. Thats how it works. The only reason why coin tosses are 50/50 each time is because we don't know how to calculate the outcome. If we did, each toss would be 100% one way or the other. No randomness in the face of absolutes.
@uv46jb
@uv46jb 10 лет назад
Actually, I'd say sharing the money is the right choice, because either A: The other person has decided to take all the money in which case you're not getting any no matter what so it doesn't matter. or B: The other person agreed to share the money with you in which case you owe them to do the same.
@juuonse
@juuonse 12 лет назад
The occasion where your argument of deceptive strategy being the dominant one is more of an exception than the rule. Factors such as the ones I've already mentioned as well as cultural differences (could be explained through an institutional theory aspect) and many others affect the outcome of the game. Simply put, the co-operative trend is nearly always the dominant strategy. That(!) is the dilemma which game theory attempts to and to some extent does explain.
@nimedbuel
@nimedbuel 2 года назад
Sean Lock will be very missed... 😭
@vidfreak56
@vidfreak56 12 лет назад
Cont. Of course one might assume that the other would pick sharing as the "best" answer and then change it to "take the lot" knowing this. But assuming this, the other might also do the same. The best answer then is to still share it, as it's a guarenteed "win". Basically as many have put it, it comes down to greed, and whether or not a player is "high risk" or "low risk". The lowest risk answer for everyone is still sharing.
@iLuckyyy
@iLuckyyy 12 лет назад
@Mx41204 Since neither player knows what the other will do, the optimal choice is to take all.
@PrrrromotionGiven
@PrrrromotionGiven 11 лет назад
Yes! I never expected anyone to get that here...
@Wheellos
@Wheellos 8 лет назад
So the only 10% chance of getting the result you aim for doesn't count when ur goal is to miss? I'd say that if you attempt to deliberately miss u only have a 10% chance of pulling that off under the rules as they are stated.
@andrew7taylor
@andrew7taylor 8 лет назад
+John Doe - So if he doesn't miss, he hits one of the other guys 90% of the time?
@Wheellos
@Wheellos 8 лет назад
Exactly, you have a 10% chance to hit what you are aiming for (which is anything besides the other two) and a 90% chance to hit one of them.
@wowsa0
@wowsa0 11 лет назад
That's not what stephen fry said though. He said best option is to share. You are correct, and very well put too I enjoyed your follow up comments, and that means Stephen Fry was incorrect.
@Truthiness231
@Truthiness231 12 лет назад
I concur with the whole of that but most importantly that last line: if the two people in a prisoner's dilemma know what they're in for and can confer beforehand, it's best to just work out an agreement (with a contract, even) to split the money. Does kinda kill the point of the dilemma though... but hey as far as I can see this is just figuring out one's absolute best option, and if that is on the table it's the best option by far. ^.^
@TheBlueReptile
@TheBlueReptile 12 лет назад
@z4ck1gr33n The point is that by purposefully missing you are no longer a threat and so neither will waste their bullet on you. It doesn't solve the Liza situation but it certainly does keep you alive.
@influenza99
@influenza99 12 лет назад
Agreed. Stephen got that wrong or misspoke. Without knowing your opponents answer, either decision you make (share or take all) leaves you with a 50% chance of leaving with money...so obviously take the 50% option with the full jackpot.
@ecaepevolhturt
@ecaepevolhturt 11 лет назад
It depends on how you define best. In "Game Theory" the best decision for the player is the best outcome for the player.
@ashy386
@ashy386 12 лет назад
Yeah, you are correct in saying that you should take the lot from a purely mathematical or logical point of view. But the problem with that is that the other guy is thinking the same thing, so you both decide to steal and get nothing. Basically, it is clear that the best outcome for you is to take the lot, regardless of what the other guy does. But then both players do it and get nothing. But in terms of Game Theory, the best answer is for both to split it
@ihave72legs
@ihave72legs 8 лет назад
There are some interesting variations on game theory. In one people were invited to send in codes for the behaviour of their players, who were then pitted against each other repeatedly to see which strategy ended up doing best. I think Dawkins had something to do with it, not 100% though.
@HrHaakon
@HrHaakon 11 лет назад
You're sort of right, but the proper argument is as follows: Assume that the pot is money that we've both earned or something, since game theory generally assumes a zero-summish game. Given that you share, then it's better to steal, since I get all of it. Given that you steal, then well, you certainly aren't going to have it either, so I should also steal. So no matter what YOU choose, stealing is the best option for me. Of course, it's a lot more complicated than that, butyou get the point.
@vidfreak56
@vidfreak56 12 лет назад
I use -1 and 1 here as examples of directions with a magnitude of 1.
@TheBlueReptile
@TheBlueReptile 12 лет назад
@Ensirum Sure, Ok, I believe you...
@acswwwat
@acswwwat 11 лет назад
In a repeated game with a specified cutoff (say 10 iterations) then the rational person will still choose steal the very first game. The reason this is rational is because you will want to steal in the last round, but so will your opponent. So you will want to steal in the next to last round to beat them to the punch. But they will also want to do this. You keep doing this until you realize that you will steal in the first round.
@Terezar
@Terezar 12 лет назад
that is actually the correct answer. You convince the other guy that you are going to hit steal, tell them right out that you are going to hit steal regardless of what they choose, and that if they hit share you will share the money after with them. They will, if you convince them, go for share, and then you do the same and you both win.
@biggusballuz5405
@biggusballuz5405 9 лет назад
A takes Lot, B takes share - A wins 500k. A takes Lot, B takes lot - A wins 0. A takes share, B takes share - A wins 250k. A takes share, B takes lot - A wins 0. Overall you have a 50% chance of getting nothing, 25% chance of winning 500k and 25% chance of winning 250k. In simpler terms, you have a 50/50 chance of winning something or nothing. But taking the lot would allow you to have a higher cash price should your luck push through So your best chance is to say that you will take the lot and pray the fuck that the other guy thinks that saying "share" is a better or safer option.
@TheSummerService
@TheSummerService 9 лет назад
Qin ShiHuang 秦始皇 It's not strictly true to say that you have a "50" or "25"% of anything, because in real life the other person may be biased towards one of the options. Interestingly, this has no effect on the expected values though, because both cases of you earning money depend only on the probability of the other person SHARING, which is the same regardless.
@biggusballuz5405
@biggusballuz5405 9 лет назад
TheSummerService That is why statistics doesn't care about variations, it would make calculations and prediction models impossible to come up with.
@TheSummerService
@TheSummerService 9 лет назад
Qin ShiHuang 秦始皇 Lol. Impossible to come up with? Expected value=probability*payoff. Easy. If we assume the bias towards sharing was 70 to 30: Expected value of taking the lot: 0.7*500+0.3*0=0.7*500 Expected value of sharing: 0.7*250+0.3*0=0.7*250 If we assume the bias towards sharing was 10 to 90: Expected value of taking the lot: 0.1*500+0.9*0=0.1*500 Expected value of sharing: 0.1*250+0.9*0=0.1*250 More generally: Expected value of taking the lot =Pr(sharing)*500 Expected value sharing=Pr(sharing)*250 Pr(sharing)*500>Pr(sharing)*250, ALWAYS, so Expected value of taking the lot>Expected value of taking the lot, ALWAYS. Taking the lot is the rational decision because it gives the highest payoff INDEPENDENT OF THE BIAS. This only works because of the specifics of the game. In other cases/scenarios, the bias would definitely be needed to be taken into account and ignoring it would be unsound.
@biggusballuz5405
@biggusballuz5405 9 лет назад
TheSummerService Which still means I am right. 25 25 50.
@TheSummerService
@TheSummerService 9 лет назад
Qin ShiHuang 秦始皇 No. If the bias is 70:30 towards sharing then: A takes lot, B shares = A wins (500k) with probability of 70% A takes lot, B takes lot = A gets nothing with probability of 30% A shares, B shares = A wins (250k) with probability of 70% A shares, B takes lot = A gets nothing with probability of 30% No matter which action (sharing or taking the lot) you take, you stand a 70% chance of winning some money, and a 30% chance of getting nothing. Nothing to do with 25s or 50s.
@Tomwithnonumbers
@Tomwithnonumbers 11 лет назад
What the original solution (steal) relies on is that of the two people taking part, one is thinking rationally and the other isn't. If they are both rational thinkers, than the correct option is to share, because they realise that since they are both of equal intelligence with equal information, they will both come to the same decision, whatever that decision may be
@diurdi
@diurdi 12 лет назад
You can assign any probabilities to the choice of the other person, and my logic will still hold. It's not based on 50:50. Let's say the other person will share 90% of the time. Your best option if the other person is going to choose to share will still be to "take the lot". You'll recieve all the money instead of half of them. Think about it for a while and you'll probably figure it out.
@rancidigloo
@rancidigloo 14 лет назад
Did Fry get this description entirely correct regarding the sharing of a cash prize? It was my understanding that game theory advocates the selfish choice, therefore the player should elect to take all the money; in this case receiving either 50,000 or 0, whereas as following My Fry's advice nets 25,000 or 0, a less favourable set of outcomes.
@vidfreak56
@vidfreak56 12 лет назад
The risk is collectively lower for choosing sharing, but it's higher for those that choose to take the higher payout assuming the other one plays "share it". The only problem with this is its not the "best" choice. Sure its the highest payoff, but if the choice is between winning and losing, sharing guarentees the win, while the other method does not. Both players if they assume the other player will share, will choose all, and both will win nothing, which is an inevitable loss.
@Raveityourway
@Raveityourway 12 лет назад
If you could see what the other person was saying, and they said 'take all', when you're out of any money. It's a matter of choosing vengance or turning the other cheek.
@Aspartem
@Aspartem 10 лет назад
Incredible how so many people can't grasp a simple concept like the prisoners dilemma. Education, where have you gone?
@Hellwyck
@Hellwyck 8 лет назад
+Aspartem *Education, where have you gone?* It fucked off when XFactor turned up and people realised you can be famous for doing nothing.
@441meatloaf
@441meatloaf 12 лет назад
@Proverbs232 that is the prisoner's dilemma. Or alternatively known as a one-shot game. Prisoner's dilemma doesnt have to refer to the "prisoner" its just a game name to describe the game. The game result fry mention is wrong because both players will not cooperate to share the the cash prize of 250K. The pure strategy Nash equilibrium is (0,0) meaning both players would select the choice to take all resulting in o payoffs for each,
@eslington
@eslington 12 лет назад
In this case, if you choose to Share then you will either get nothing or get 50% of the prize. If you choose to Take it all then you either get nothing or all of it depending on what the other person does. So there's no "Dominant Strategy" unless you know what your co-player is more likely to do. In a "normal" Prisoner's dilemma both people choosing to Take it all would result in a reduced prize, meaning the Dominant Strategy is to Take since in both cases the Take option yields higher prizes.
@VikingDobes
@VikingDobes 11 лет назад
Right conclusion, wrong reasoning. You don't have to assume the other person is randomly picking, but you are correct that since you're equal or better off by stealing no matter what the other person does, you should steal every time
@joakimandersson7769
@joakimandersson7769 7 лет назад
Purely mathematically , taking the lot is the strongest option, but when you factor in risk psychology it gets a lot weirder, and fast :D
@mankytoes
@mankytoes 12 лет назад
Surely the best thing for Alan to do would be to refuse to shoot, and just keep his gun aimed at Sean? The others killing each other at the exact same time doesn't seem that likely, and this way he'd have a chance at a shot at the more likely survivor.
@juuonse
@juuonse 12 лет назад
I've already explained twice what the dilemma is about and why I would change my original reasoning based on my own reasoning. I can't make it any clearer than it already is. Because I reason the same way as you I will end up co-operating. Same applies to you. The co-operative trend has been observed in real life countless of times. In controlled experiments as well as in business world between competitive firms (e.g cartels)
@acswwwat
@acswwwat 11 лет назад
That's true. I suspect what he meant was if each followed the weakly dominant strategy then they get nothing so the socially optimal strategy is to share. This is the whole point of the prisoner's dilemma. It attacks the basis of Ayn Rand's philosophy wrt everyone being selfish will give socially optimal results.
@DocLow
@DocLow 12 лет назад
@200131240 Actually, colaboration is only the optimal solution in the iterated prisioners dillema, which is not the case here - one shot per person. Nontheless, Alans best option is still to deliberatedly miss.
@JordGames
@JordGames 12 лет назад
Ignore the probabilities, it isn't a 50:50 thing. The other player's choice cannot be affected by yourself. You should steal. If they steal, you've stopped them getting one over you and if they share you win the lot. In no situation is sharing a good idea
@z4ck1gr33n
@z4ck1gr33n 12 лет назад
Not true missing would leave two people remaining alive and the problem of who has Liza will remain, probably leading to a rematch in which you would probably lose. The best option would be to wait until one of them has used their bullet to shoot the other person, then shoot the last one standing without any risk. This is most likily to be the outcome as the other two would try to kill each other first as they have the better shot and will be the biggest threat.
@watergun7
@watergun7 13 лет назад
e.g. is when alan and phil both have same success rate, say p. then when p is at or very close to sean's success rate then none of the 3 should shoot, and as p tends to 0 (keeping sean's success rate constant), there comes a point when they should both shoot sean. I'm just doing these examples intuitively without any calculation (i.e. i'm lazy). Of course you could solve the problem in any case. Game theory is just a maxmin problem (maximizing the minimum outcome of the game for yourself)
@Ensirum
@Ensirum 12 лет назад
@TheBlueReptile I never said that, I said if you lower the chance. See you have this all wrong. These statistics have nothing to do with aiming. It's not about aiming. If I add a bump on the 5 of a die, it would lower the chance of it landing on a 2 (the opposite side of the die). Due to the fact that it would more than likely roll over. So because I've made this chance, the chance of it landing on the other 5 are above 16.66%. Also, dice is plural, you're is YOU ARE. You mean YOUR.
@2dTones
@2dTones 12 лет назад
Well if you're looking to maximize payout across all participants, then yes, you're correct. However, game theory usually applies at an individual level i.e. tends to assume participants aren't altruistic.
@CanoasTC
@CanoasTC 12 лет назад
If you say you'll share it you have a 50% chance of winning half and 50% of not getting anything. If you say you won't share it you have a 50% chance of winning everything and 50% of not getting anything. The best answer is asking for everything.
@ObjectsInMotion
@ObjectsInMotion 12 лет назад
Wait if you say share, then there's two possibilities 1. The other says take and you get nothing 2. The other says share and you get half But if you say take, then the two possibilities are 1. The other says take and you get nothing 2. The other says share and you get everything If you take the average of say 100 people who say share and 100 who say take, those who say share get 25% (the average of half and nothing) but those who say take get 50% (all or nothing) So you should take.
@TimoRutanen
@TimoRutanen 6 лет назад
It seems like an important bit of the puzzle is missing. That they all only have one bullet.
@MiltiadesOfMarathon
@MiltiadesOfMarathon 12 лет назад
@bewilderedkettle Ok, I agree with you there, the fact that I would rather neither player get the money than have the money stolen from me by the other player is a value judgement. So removing that we have a neutral proposition, if he picks steal I can get nothing or I can get nothing. If he picks share I can get all the money or half the money. Stealing gets the player a superior or an equal payout, it is never the wrong option.
@mushroomshrub
@mushroomshrub 12 лет назад
Anyone know what happened with that situation? I've never stooped low enough to watch big brother.
@441meatloaf
@441meatloaf 12 лет назад
@Proverbs232 However, if you turn to another Game theory theoretic where players play repeated games indefinitely. Then the outcome will be cooperation given the grim trigger strategy and if the patience of each player is high enough, cooperation will be played. You cannot think of game theory as cooperation being the best answer. You have think in terms of strategy profiles and how the game is played.
@DutchDread
@DutchDread 12 лет назад
I agree, But that's assuming that the other player is thinking along different lines than I am. If I am playing against myself, both of me get 50%. And if he played against himself, no one would get anything. If people think like him, no one gets anything, if people think like me, everyone gets 50%. Personally I say that the best thing to do is to admit that you are going to take it all, and that you'll give 50% away afterwards if the other chooses share. (all of this assumes only 1 game btw)
@jonboyjon1976
@jonboyjon1976 12 лет назад
@fdasherv That's sometimes the problem with game theory. A lot of it assumes your opponents are as rational as you, and it can make general assumptions. Problem is, a lot of people don't act rationally or like they should
@vidfreak56
@vidfreak56 12 лет назад
noooo. thats if u dont know what the outcome will be. This is about strategy and the luck is in the guessing, at first.The dilemma starts off as a random guess, but fails because both teams (if they play the weak dominant strat) will fail because they will both choose "take it all". That strat wont always work as later teams will keep trying it and trying it until people realize that its pointless. It will work a few times, but the pattern will be learned again as ppl watch previous outcomes.
@jinnwarior
@jinnwarior 11 лет назад
best thing to do is take the lot. if he shares, give him the half anyway. if he takes the lot, he gets nothing. doesn't work quite as well as the regular prisoners dilemma.
@Truthiness231
@Truthiness231 12 лет назад
@MiltiadesOfMarathon
@MiltiadesOfMarathon 12 лет назад
@bewilderedkettle Again, this isnt about how we can all cooperate and share in kumbaya teddies and marshmallows land. This is game theory, the clue is in the title of this video. The fact is that stealing offers an equal payout with share if the other player opts to steal or a superior payout if they opt to share. Stealing is therefore the only logical option.
@usuckbad2
@usuckbad2 12 лет назад
@entertherat surely that would only work having other people helping you not sure I understood exactly what you meant, an algorithm if met by another algorithm which was the same would decide which would be the slave and master and then one would lose all the time? That would involve people helping eachother?
@Only1DennisBergkamp
@Only1DennisBergkamp 12 лет назад
If any one plays poker and understands estimated value, Typhoontoby, is right, you are better off 'stealing' and bluffing that you will 'share'. However that's not really the general philosophy of the game theory, it's not just about a game show, more to do with economics and how a more sharing and less rival relationship between business etc. can help each other be more successful.
@XhoowieX
@XhoowieX 12 лет назад
The best policy is still to steal. Even if they're 99% likely to steal too, in 99% of cases you'll go away with nothing when you steal and in 1% you'll go away with everything. Still assuming the same chances, in 99% you go away with nothing if you share - and in this case you have the knowledge that some bastard stole it all - and in 1% you go away with only half. Like Typhoontoby said, its not necessarily the moral 'right choice' but certainly it's the best in terms of reward.
@vidfreak56
@vidfreak56 12 лет назад
Collectively the odds are better if share is picked. Thats what i meant by "best" answer. The gain may be higher, but if both players use game theory then everyone will always lose. Share is the best solution.
@JhericFury
@JhericFury 11 лет назад
It really is, and yeah, i kinda got that one the first time round as well, but then i do a lot of maths and probability. I just know the idea that changing gave a better probability completely confused my Mum, to the point where she wouldn't believe it.
@dadeskr
@dadeskr 12 лет назад
I'm not mathmatical, so I think you guys may prove me wrong but: If you choose share then you either get half or you get nothing, if you choose take all you either get all or nothing, take all seem like the better option to me.
@frqdea
@frqdea 13 лет назад
that's hugely interesting
@Vascorian
@Vascorian 12 лет назад
Prisoner's dilemma is the big brother situation
@DutchDread
@DutchDread 11 лет назад
I get that in a single game, stealing is always the better option at that time. The problem is that successive games played in this way create a world where the likelyhood of the other person stealing becomes greater and greater, looking at the bigger picture it's more advantageous for every individual if everyone shares. Ofcourse it's best for you if everyone shares except you, but that sort of thinking is what got us into trouble in the first place. It's the problem of the commons basically.
@mercycollege123
@mercycollege123 11 лет назад
what happens to the sound?
@razorRmc321
@razorRmc321 13 лет назад
I think it depends. If you're only after winning the game prize then yeah they should of gone for the selfish choice. However, if they're concerned with the aftermath of the game show, and they have up to now been portrayed as a character as being nice and moral, then taking the selfish route will undermine their appearance and will leave considerable backlash from the public audience. I think choosing to share the money is the safe option.
@therrydicule
@therrydicule 12 лет назад
Look, the "best outcome" for the 2 prisoner in a prisoner dilemma is to cooperate. It's not the dominant outcome, but it's the best one.
Далее
The Very Best Of Sean Lock On QI!
16:00
Просмотров 1,2 млн
His reaction 😳 (via @kaitlyn.b0506/TT) #shorts
00:10
I Melted Wood With Friction
8:44
Просмотров 714 тыс.
Qi - Onomatopoeia  and Welsh
5:07
Просмотров 486 тыс.
How was Germany Formed?
38:19
Просмотров 69 тыс.
John Sessions impersonates Alan Rickman on QI
2:55
Просмотров 2,1 млн
Ramsey theory on QI (Higher Quality)
3:07
Просмотров 1,3 млн
How Do Restaurants Rip You Off? | QI
4:27
Просмотров 70 тыс.
Hilarious QI Moment - Cricket
1:18
Просмотров 286 тыс.
SCIENCE On QI! Funny And Interesting Rounds!
32:22
Просмотров 538 тыс.