Game theory is the study of what people do when what happens to them depends on what somebody else decides.
That is most of life. Splitting a bill. Deciding whether to trust someone with a secret. Choosing whether to speak up when a group is heading somewhere bad. None of these have a right answer you can work out alone, because the right answer depends on what everyone else does.
Mathematicians started studying this properly in the 1940s, and it turned out that a small number of situations come up again and again, in economics, in biology, in politics, and in the school playground. They have names. They have been studied for eighty years. And once you can recognise one, you handle it better.
GAMETHEORY puts a young person inside five of the most famous of them, with three other players around the table, and then tells them what they did.
Four boats fish the same water. Every fish you take is yours. But the shoal only survives if enough are left behind, and it belongs to nobody, so nobody protects it. Everyone acting sensibly for themselves destroys the thing they all depend on.
This is why fisheries collapse, why aquifers run dry, and why the last slice always disappears. Elinor Ostrom won a Nobel Prize in 2009 for showing that real communities solve it more often than the theory predicted, and for working out how.
Nine paintings, one hundred coins, and four people who each secretly want different things. You are not trying to guess what a painting is worth. You are trying to guess what everyone else thinks it is worth, and pay as little above that as you can.
Auction theory is one of the most practical branches of the field. It sets the price of radio spectrum, advertising, and government debt. Paul Milgrom and Robert Wilson won the Nobel Prize for it in 2020.
A harvest has to be divided. Everyone wants more. But part of it has to go into the walls, and the walls protect everybody, including whoever contributed nothing.
This is the free rider problem, and it is the reason taxes exist, the reason group projects go wrong, and the reason someone always has to be the one who tidies up. You can finish this game with the biggest pile and still lose, in a city that fell.
Four hunters. The stag feeds everyone, but it takes all four. The hare feeds one, and you can take it alone. Nobody is trying to cheat anybody. The problem is simply that you cannot be certain the others will hold.
Rousseau described this in 1755. It is the cleanest example of how trust is not a soft skill but a hard coordination problem, and how one person losing nerve costs everybody.
You send some of your coins. They triple on the way. Then the other person decides how much comes back, and nothing forces them to send anything.
Economists have run this experiment thousands of times, in dozens of countries. Almost nobody sends nothing, and almost nobody keeps everything, which is not what pure self interest predicts. What varies is how quickly trust rebuilds after somebody breaks it, and that is the part worth practising.
These are not invented scenarios. Each one is a standard experimental design that researchers have run for decades, with adults and with children. A short reading list, in case you want to check the ground this stands on.
Tested children aged three to eight with sharing games. Almost all three and four year olds behaved selfishly. By seven or eight, most preferred splits that removed inequality, whether it favoured them or not. Fairness is not something children simply have. It arrives, and it arrives at a particular age.
Followed the same question through childhood and adolescence rather than stopping at eight. This is the age range GAMETHEORY is built for, and it is the period where these preferences are still visibly moving.
A review of how fairness develops, including the finding that children will reject a reward rather than accept an unfair split, and that this appears far earlier than the willingness to give up an advantage.
Theory predicted that shared resources would always be destroyed by self interest. Ostrom went and looked at real fisheries, forests and irrigation systems, and found communities that had managed them for centuries. What they had in common was monitoring, communication and proportional consequences, which is exactly what The Shoal gives a table.
The original trust game, which The Envelope is built on. Pure self interest predicts sending nothing. Almost nobody sends nothing. It has been replicated in dozens of countries since.
People will pay a personal cost to punish someone who took advantage of the group, even with nothing to gain themselves. Cooperation collapses when that option is removed and holds when it is available. This is the research behind reporting someone to the harbour.
Ran a tournament of strategies against each other. The winner was among the simplest: cooperate first, then do whatever the other side did last, and forgive quickly once they stop. Being nice, being provokable and being forgiving beat being clever.
One thing worth saying plainly. This research describes how fairness and trust develop. It does not show that playing these particular games improves a child's behaviour, and we are not going to claim it does. What we can say is that these are the situations the research uses, that the decisions inside them are real, and that afterwards your child gets to see what they actually did rather than being told what they should have done.
You cannot teach someone to read a room by describing it. They have to be in one, make a call, and find out what it cost.
What makes these five situations worth using is that none of them have a safe answer. Play purely for yourself in The Shoal and everyone starves, including you. Play purely generously in The Collection and you go home with nothing. Every game punishes both extremes, so the only way through is judgement: read the table, decide who is worth trusting, and know when to hold a line.
Afterwards, an AI reads back everything that happened and shows the player the moments where the decision actually turned. Not a score. The moment.
Every one of these games is older than the internet. The situations are not going anywhere.