Transcription
[Music] So, you're playing the game of life. Isn't that nice? Congratulations on making it this far. Maybe you're doing well. Maybe you're doing not so well. Either way, you still have to play all day, every day until the day you die. But death is a different game entirely. And the life game is, how should I say, uh, complicated.
Imagine your life as one long string of situations. By talking to you through this YouTube video, I am touching your string right now. If you leave a comment, you might touch mine. Wouldn't that be naughty? In the game of life, you have many a choice to make. Many rabbit holes to dive down, many games and side quests to play. Some situations are long overarching narratives. Others are quick and unimportant. But almost all situations have an objectively correct strategy to handle.
How often do you kick yourself in the shower, realizing if you had only tried, made, chosen, bet, or said something different, events would have played out differently? You chose wrong on that occasion. And that's okay. Knowing the right path in the moment, every moment is an impossible feat. But this is where game theory steps [Music] in.
Game theory is the idea that the choices you make can be reduced to mathematical formulas and optimal choices can be calculated. Look at the game I'm playing right now. It's called Golden Balls. Well regarded as probably the most brutal game show ever devised for British television. Golden Balls forces its contestants to make a simple choice: split the jackpot with their opponent or steal the lot for themselves. Both players have to make the choice simultaneously, secretly, and with finality. They have no control over what their opponent does besides whatever pleas, bargains, and lies they're able to exchange prior to the choosing. And the game only has three outcomes: one, both players pick split and share the jackpot; two, one player picks steal and takes the entire jackpot; or three, both players pick steal and receive nothing. In other words, Golden Balls is a classic prisoner's dilemma, the most famous problem in game theory.
Assuming your opponent is a pretty cool, sexy, rational person like yourself, what should you choose? Send your votes now to this email [Music] address. Yeah, you steal. Game theory dictates that stealing is always the best option. Why? Because stealing leads to either a win or a draw, while splitting is either a draw or a loss. It has nothing to do with whether it's better to be a greedy bastard or a nice person. The mathematically rational choice is to steal. However, if both players play rationally, as Game Theory assumes, then both players will arrive at this conclusion. They will both pick steal and win nothing. This is known as a Nash equilibrium, where neither player would have anything to gain from making any other choice. It was first proposed by Professor John Nash, the schizophrenic mathematical genius played by Gladiator.
However, the reason the prisoner's dilemma is so well, uh, dilemmatic is because the most favorable outcome is, of course, both players splitting. But the best choice is to steal. And so this leads to a rational paradox where your best choice on paper most often leads to your least desirable outcome. But the life game is not played on paper. In life, people are allowed to talk, to converse, to coerce, to tattle, persuade, manipulate, and guilt trip. The dialogue aspect of Golden Balls makes it truly compelling entertainment, adding a whole new factor that completely skews rational game theory. One analysis of the TV show's 289 glorious episodes found that across all games and players, split was chosen approximately 53% of the time, with only 47% of players choosing steal. Regardless of the outcome of any of these games, it can be concluded that 53% of people behaved completely irrationally from the perspective of a game theory purist. People are imbeciles. Do not be people.
If you ever find yourself in a version of the prisoner's dilemma, be a cold bastard and pick steal. Pick betray, pick defect, rat, lie, and snitch. It's okay. Don't feel guilty. You don't need morals when maths is on your side.
Speaking of morality, who's up for a cheeky game of kidnap? Let's imagine you've kidnapped your favorite celebrity. Should be fairly easy for a twisted little mind like yours. You demand a hefty ransom, and it is promptly paid. Now you are faced with a choice: be true to your word and release the hostage or take care of them. Seems fairly simple. So here is the dilemma: the hostage knows your face and name. Why? Because you got lonely and hung out with them. You sad little [ __ ] Anyway, if released, the hostage would be able to aid the authorities in tracking you down and imprisoning you if they chose to do so, whilst terminating the hostage would leave you rich and free from jail for certain. But you would greatly prefer not to become a murderer, especially as the prisoner is your idol. In an ideal world, you would release them, and they would stay silent. Here are all the options with a rough numerical value assigned to the outcome for each player. What should you do? Cast your votes now by calling this hotline. Yeah, you killed him. It doesn't really matter what you think. Just kill them. It's not a debate. Maths has decided. Game theory dictates that eliminating the hostage, which will again ruin your chances of achieving the most desirable outcome, is still a better option when faced with a 50% chance of imprisonment. But life often has other factors to consider. For example, if your prisoner develops a touch of the old Stockholm syndrome, well, that changes the game. If they love you, it's less likely they will go ahead and rat you out. So, the best option might be to release them. Or maybe you're in a future where mind potatoes could force your hostage to testify against you unwittingly, in which case definitely kill them. But hostage romances and crazy imaginations aside, in base principle, eliminating the captive is always the best option if you're going to play the kidnap game. Ignore the potentially better scenario and take the guaranteed victory. But please, I implore you, do not actually try and play kidnap in real life. Believe me, it's just not very fun.
Moving on. I'm going to flip a coin. Is it heads or tails? Vote now by screaming out your window. It was tails. Pretty easy game, huh? But only about 40% of you won it. What? How? It's a 50/50 chance, right? Well, yes, the coin flip itself is a 50/50 chance, but what you would choose was not. Human beings have a slight bias towards picking heads. Why? Because heads comes before tails. So, it's often the first thing that comes to mind when making such a call. So, if you're a tails-of-fails person, congratulations. You are in a statistical minority. And being a minority means you're cool.
But how is this important for game theory? Well, the human preference for heads is seen as a convention, a rule that people tend to follow. Knowledge of conventions can help you win games. For example, if you're playing a coin game where you have to guess what your opponent will call, you can use the coin convention to assume they will pick heads, theoretically giving you a 60% win rate in what should be a 50/50 game. And conventions are extremely important to game theory, as they relate heavily to real-life applications. Driving is a game. Driving on the correct side of the road is a convention. Knowledge of that convention prevents you from crashing. Unless, of course, you're driving in India, in which case, good luck. Restaurants are a game. Tipping your waiter is a convention. You automatically benefit from a higher level of service if the waiter believes they might earn a tip, regardless of whether you actually tip them. Greetings are a game. Shaking hands is a convention. It's an empirically meaningless action that also manages to automatically build trust. Government is a game. Paying taxes is a convention. If everyone refused, then the government would be powerless, and society would collapse. Society is a game. Money itself is a convention. All money has only imaginary value. It's completely useless. Get rid of your money. Burn it all. Wipe the accounts. Head for the hills. Don't you feel the ship sinking? Your life is only getting harder. Can't you hear the piggies roasting? It's only getting worse. Doom. Doom. Climate collapsing. Nuclear fire. Money is all that's keeping society together. The green paper shield. But it's just an idea. It will not last. Already it's failing. Soon. Come off the doom. We're all going to die. Sorry, sorry. I'm really trying to cut down on my doomsday rants. Uh, going back on topic now, I promise.
Conventions in game theory can reduce uncertainty. However, some games are designed to be wholly [Music] uncertain. You have been challenged by your most bitter of rivals, slapped with a glove, and labeled the great big Nancy. Naturally, your only option is to engage your foe in an unfriendly game of duel. The rules are simple: each player shall be given a pistol with a single bullet. The players will begin a half mile apart and will slowly walk towards each other at roughly the same pace. You are both allowed to fire your pistol at any point, but if you miss your only shot, you must continue to walk towards your opponent until they, in turn, shoot at you. Assuming you and your rival are similar in both body size and accuracy, what is the best time to fire your pistol? Please write your votes on the ballot cards for the upcoming Ribble Valley bi-election. Yeah, there is no best time to shoot. It doesn't exist. In game theory, duel is what's known as a mixed strategy equilibrium. Fire too early, and you lose. Wait too long to fire, and you also lose. Which means with all factors being equal, it doesn't matter when you shoot. Just take a random shot and hope you don't miss. Yeah, it's mental. Do not play duel in real life.
Poker is another example of the value of random tactics. If you only raise your bet when you have a good hand, then opponents will always know when to fold. But if you go all in on every hand, your opponents will quickly realize you're full of [ __ ] and clean you out. However, if you randomly decide to bluff a bad hand every now and then, then your opponents will have a much harder time assessing your strategy. Unless, of course, you have a really obvious tell, but no amount of game theory can fix that. Knowing the nuance between a calculated blunder and giving your opponent free money is the mark of a truly great poker player. And in the larger game of life, intuition is often more important than strategy, more important than game theory. Knowing people's tendencies and reading into their body language and emotions is often much more valuable than relying purely on statistical advantages because emotions are factors that can't really be effectively quantified on paper. As such, in reality, game theory often falls [Music] apart.
Welcome to Ultimatum. Again, it's a two-player game. There is a pot of money. Let's call it 100 blank coins. The first player chooses how to split the money up. The second player chooses whether to accept the split or not. If accepted, both players get their respective cash. If declined, both players get nothing. How should this situation play out? Vote now by sending me your thoughts and prayers. Yeah, you all got it wrong. Player one should offer a split of 99 to 1 in their own favor, and player two should accept. A game theorist will reason that player one should try and maximize their gains as much as possible. They can get away with this as player two should accept any amount of money higher than zero. Rationally, they stand to gain nothing from declining free money. But your immediate reaction of fury and dismay is exactly why game theory does not always work: human spite. Game theory does not factor in how the human desire for vengeance greatly trumps their supposed rationality. In reality, the best choice for player one is to offer a 50/50 split as any saying player two is guaranteed to accept it. And if player one offers, say a 55/45 split, well, that's a bit cheeky, but player two is still quite likely to accept. However, if player one's guts get greedy and they go for, say, a 90/10 offer, then the likelihood of player two declining has grown exponentially. So, the further the offer strays from the 50/50, the more likely it is that player one is told to get [ __ ] and both players leave with an abundance of nothing. Humans simply refuse to behave rationally when they feel they're getting [ __ ] over.
However, that is not to say game theory is completely useless. In the case of Ultimatum, you only have to add repetition to the game. In a repeated game, if player two had refused the offer previously, player one is much more likely to move their offer closer to the 50/50 margin to avoid winning nothing once more. Player two can still decline if dissatisfied, but with enough repetitions, both players should eventually reach a mutual equilibrium and make some serious blank coin while the games last. The human ability to learn lessons from our mistakes is something that game theory accurately reflects when games are repeated. And Ultimatum is not the only game where repetition greatly impacts strategy.
Yeah, we're back to Golden Balls. A repeated prisoner's dilemma is a completely different situation to the original game, at least from a mathematical perspective. In a one-off game, you should always steal. Sure. But across multiple games, this is not the most effective strategy. So, what is the best strategy? Well, thankfully, Professor Robert Axelrod of the University of Michigan famously held a couple of nerd contests to answer that question. People from all the nerdiest corners of the globe were invited to create and test computer programs against one another in an attempt to identify the best course of action in a repeated prisoner's dilemma. Many programs were submitted: nice ones, mean ones, sneaky ones, and random ones. The programs played prisoner's dilemmas against each other for 200 rounds, and then the points were tallied. And would you believe it, nice programs performed far better on average than the [ __ ] programs. The most aggressive models would often end up spiraling into steal wars of mutual destruction, whilst nicer programs would lose when paired up against these aggressors, but fare much better overall due to their ability to cooperate in games with other nice programs. Do you get it? Do you get the allegory for life?
One program stood out from all the rest in both contests: Tit for Tat. This program would initially pick split every time. However, as soon as the opposing models choose steal, Tit for Tat would follow suit in the next round. It would then copy the opponent's previous moves forever more. So, if the opponent continued to steal, so would Tit for Tat. And if the opposing program started to split again, so would Tit for Tat. This attitude racked up the most points by far. Axelrod attributed Tit for Tat's success to four main qualities: one, niceness. Tit for Tat was never the first program to pick steal. This made it easy to cooperate with. Two, retaliation. Tit for Tat was by no means a [ __ ]; it could be easily dragged into wars of attrition if the opponent wanted the smoke. Three, forgiveness. Tit for Tat only held grudges for one round. It could easily go back to playing nice if the opponent learned their lesson. And four, clarity. There was always a clear cause and effect to the program's method. It could not be misunderstood or manipulated. Axelrod's tournament proved empirically that an eye for an eye is an objectively better strategy than turn the other cheek. Old Testament beats New Testament. Life may not be a constant prisoner's dilemma, but it is full of repeated interactions. And Tit for Tat exemplifies how to approach the world to maximize your own best outcomes and the best outcomes for other players. Sticking to the Tit for Tat principles allows you to be pleasant without being a pushover, to be strong without throwing your weight around. And if you learn to cut your losses, sacrifice for the greater good, and throw in some wholly random actions from time to time, then bingo, you've got yourself a cheat code for the life game. I mean, obviously, there are many other winning factors, such as being born with intelligence, charisma, monetary wealth, and good physical genetics, but that's just the luck of the draw, I'm afraid. You can only play with the hand you're dealt. The life game is by no means fair. Please don't get too hung up on blaming the dealer, if there even is a dealer. Anyway, for all of humanity's downtrodden, there will always be the loving embrace of Blank Rascal. Have no fear, child. Know that here you are welcome. You are loved. A vit. Goodbye. [Music]