In which I discuss a classic philosophical problem having to do with trust and betrayal and how it can be used to make games more emotionally compelling.

You and someone you don’t especially care about are arrested in a totalitarian state and accused of a serious crime. You are put in isolated cells and a scary looking dude comes in and tells you the following:

We can put you away for 2 years on some minor felonies that we can pin on you (it doesn’t matter if you really did them). But what we really want is to nail someone for the big crime. If you confess to the big crime, and the other guy stays quiet, we can pin it all on him, in which case you’ll go free and he’ll go away for 8 years. If you stay quiet and he confesses, we’ll put you away for 8 years. If you both confess, I don’t really need either of you and it’ll be 5 years for each of you.

What should you do?

Cooperation vs Betrayal

The problem that you face is this: No matter what your fellow prisonner does, you are better off confessing to the crime. If they stay quiet and you stay quiet, you get 2 years. If they stay quiet and you confess, you go free (+2 points!). If they confess and you stay quiet you go away for 8 years. If they confess and you confess, then you only get 5 years (+2 points!). So clearly, you should confess. However, they are going to go through the same logic as you. If they reach the same conclusion (and they should) then you’ll both go away for 5 years. This is plainly worse than going away for 2 years (10 years of total imprisonment vs 4 years). If only you could coordinate your plans, you could reach an agreement to stay quiet and collectively you’d be better off. Of course even if you did make such a deal there is nothing stopping you from a last minute act of betrayal…

The essence of the prisonner’s dilemma is putting two agents in a position where they’d collectively be better off cooperating but individually better off betraying. Without going too far in to the enormous amount of theory that has been written about this, it’s a very difficult problem and seems to illustrate the impossibility of trust in a one-off interaction.

There is a variation of the dilemma where instead of only playing it once, you allow the prisonners to play it over and over again. Here things change a little. If you betray someone on the first interaction, then they’ll know you for a rotten cheat and so they can stop cooperating with you, while at the same time, trust can be built by a pair of agents that reliably cooperate over and over again. And a devious agent might build trust over a period of time before a strategic betrayal nets them the gain they need to pull decisively ahead.

The Emotional Dimension

Either version of the prisonner’s dilemma is a great beginning to creating rich high-tension multiplayer gameplay. The key elements (a private decision revealed simultaneously for all players, a cooperation strategy that is better for the players collectively set against a betrayal strategy that is better for each individual) can be adapted to a wide range of gameplay situations.

Diplomacy is essentially one long set of prisoner’s dilemmas. On each turn, players jokey for position and make deals, however until the orders are revealed all at once, you can never be sure that your ally actually followed through and there is always a tempation to defect and take advantage of their cooperation.

Any game with a trading mechanism can be turned into a prisoner’s dilemma if you allow either party the opportunity to reneg on the deal (by passing over faulty merchandise or empty boxes).

In PvP games where losing a fight means losing your stuff such as Eve Online, prisoner’s dilemmas often crop up when pirates offer ransoms to their (disabled but not dead) victims. Here, the payoff matrix isn’t symmetrical in that the victim is chosing between the lesser of two losses while the pirate is choosing between the lesser of two victories. In these games, the reputation of the pirate for honouring ransoms is often very carefully protected.

Because most multiplayer games are one-winner-only, you can essentially guarantee that the player won’t care about the other player’s wellbeing. By putting players who are competing with one another for victory in a position where cooperation might give them some collective advantage over the other players you create a very powerful emotional situation. It begins with a strategic element as each player attempts to predict what their opponent/partner will do. This is followed by a moment of suspense between their decision being made and the results being revealed. And finally, the emotional fallout of a betrayal or a cooperation leaves a mark on the rest of the game (and future games).

The Dangers

This issue that any game designer faces when incorporating the prisoner’s dilemma is the same one that makes it so compelling. The high-emotional content that comes from a feeling of betrayal can lead to angry players who aren’t having very much fun. In most modern games with a trading mechanism, players who work out how to reneg on deals are exploiters and scammers instead of clever agents and most game companies spend a lot of effort ensuring that the trading system is bullet-proof in response to angry petitions from infuriated customers.

Because one natural equilibrium point for the repeated dilemma is constant betrayal, it’s possible that the full range of the possibilities never get off the ground and the trusting relationships are never able to form. Some of these issues can be minimized by tightening up the payoffs so that even in the worst case scenario, everyone gets something (soothing the angry egos) or so that betrayal is not fully dominant. You can also offer ways for players to gain public reputation (much as eBay does) or ways for victims to punish their betrayers.

If you can work out the details of the mechanics to your satisfaction, then the possibility of a system of trust and betrayal offers some very compelling gameplay possibilities. In a massively multiplayer game, it can create a nucleus of trust around which communities are formed. In shorter games, it can create a lot of really great showdowns and a cycle of tension and release that leads to really compelling gameplay.