Strategy Guide

Why one round is hopeless

Look at a single round on its own. If your opponent settles, suing gets you 5 instead of 3. If they sue, suing gets you 1 instead of 0. Whatever they do, suing scores more. Game theorists call that a dominant strategy, and if both players follow it, both get 1 point when they could have had 3. Two sensible players end up worse off than two trusting ones. That's the dilemma.

Why many rounds change everything

A case in Lawmaxxer lasts many rounds against the same opponent, and what you do today is remembered tomorrow. The political scientist Robert Axelrod called this “the shadow of the future”: when you'll meet again, a reputation for cooperating is worth more than one quick win. Over 20 rounds, steady mutual settling earns 60 points each, while mutual suing earns just 20.

But cooperation only pays against an opponent who returns it. Settling every round against someone who sues every round earns you nothing at all. The skill is working out which kind of opponent you're facing.

What the tournaments found

Around 1980, Axelrod invited game theorists to submit computer programs to play the repeated game against each other. Many entries were clever and complicated. The winner was one of the shortest: cooperate on the first move, then copy whatever the other side did last time. Axelrod ran a second, larger tournament, and the same simple idea won again.

He found that the successful strategies shared four habits:

• Be nice. Never be the first to defect. Strategies that started fights rarely came out ahead.
• Retaliate. Punish a defection straight away, so exploiting you doesn't pay.
• Forgive. Once the other side goes back to cooperating, do the same. Holding a grudge forever locks both sides into the 1-point trap.
• Be clear. Simple, readable behaviour lets the other side learn that cooperating with you pays.

Reading opposing counsel

In Lawmaxxer you don't know who you're up against, so the first rounds are an investigation. Some questions worth asking:

• Do they answer my moves at all? If their moves never change whatever you do, they aren't reacting to you. Then the best reply depends only on what they keep doing, not on keeping the peace.
• Do they punish a single Sue? Try one Sue early and watch the next round or two. An opponent that hits back right away rewards cooperation. One that shrugs it off might be exploitable, but carefully.
• Do they forgive? After they retaliate, settle again. If they come back to settling, peace is available. If they never do, you've made an enemy for the rest of the case, and probing was expensive.
• Is there a pattern at all? If the moves look like coin flips, no amount of good behaviour will train them. Protect yourself.

Every probe has a price. Testing an opponent who never forgives can cost you the whole case, so weigh what you'd learn against what a mistake would cost.

Mistakes happen: playing with noise

Lawmaxxer sometimes misfiles a move as its opposite. That matters more than it sounds. Two strict copycats who both mean to cooperate can be knocked into a long feud by one clerical error: one side “sues” by accident, the other retaliates, the first retaliates back, and so on. Research on noisy games found that a little generosity, such as letting a single defection slide now and then, recovers from accidents much better than strict tit for tat. When a misfile hits you, a quick return to settling often saves the case.

Endgames and the hidden finish

If you knew exactly which round was last, suing in it would be free: there's no tomorrow for them to punish you in. Knowing that, they'd sue in the round before, and the logic unravels all the way back to round one. That's why Lawmaxxer hides the real length of each case: you only find out it's over when the verdict arrives. Without a known last round, the shadow of the future never goes away.

Using the court's tools well

• Objection pays more the earlier you call it, but a wrong guess costs points. Object once you've seen enough moves to be confident, not before.
• A Subpoena is most valuable in a close case, or when you suspect a change of heart.
• A Plea deal doubles your next round, so save it for a round you expect to win big.
• A Continuance undoes a round that went badly and tells you what they filed, which is also a free look at how they think.

Further reading

• Robert Axelrod, The Evolution of Cooperation (1984): the tournaments and what they mean.
• William Poundstone, Prisoner's Dilemma (1992): the history of the idea and the Cold War.
• Nicky Case, The Evolution of Trust: a free interactive explainer on the web.