A Game of Prisoner’s Dilemma

What is the Prisoner’s Dilemma?

The Prisoner’s Dilemma was developed in 1950 at RAND Corporation by Merrill Flood and Melvin Dresher.

In his 1993 book, Prisoner’s Dilemma, author William Poundstone describes the scenario as follows:

Two members of a criminal gang, A and B, are arrested and imprisoned. Each prisoner is in solitary confinement with no means of communicating with the other. The primary charge would result in a sentence of ten years in prison; however, the police lack the evidence for a conviction. They intend to sentence both to two years in prison on a lesser charge, but they offer each prisoner a Faustian bargain: If one confesses to the principal crime, betraying the other, they will be pardoned and set free, while the other will serve the full ten-year sentence instead of the two years for the lesser charge.

This leads to four possible outcomes:

• A and B both remain silent: each serves 2 years in prison.
• A betrays B, B remains silent: A walks free, B serves 10 years in prison.
• A remains silent, B betrays A: A serves 10 years in prison, B walks free.
• A and B both betray each other: each serves 5 years in prison.

This leaves A and B in a troubling predicament. If one plays nice and remains silent, they will either spend 2 or 10 years in prison, depending on what the other one does. If one betrays the other, they will either walk free or spend 5 years in prison.

How we Played the Game

The game was hosted by software delivery expert Matthias Berth.

We participated in the game as pairs and played over multiple rounds, which added complexity as we needed to think about the consequences of our decisions over time.

Pieces of candy served as game tokens, each symbolizing one year in prison. Whoever lost their candy first lost the game.

Several interesting scenarios emerged during the game. Here’s a look at a few:

Consistent Cooperation

In one instance, player B opted for consistent cooperation. However, this strategy did not work out well:

Breaking Trust

In this scenario, the players initially trusted each other. As trust eroded, both ended up paying a penalty:

Consistent Betrayal

Player B tried a constant betrayal approach, but player A adapted quickly to this strategy:

Impact of Reputation

The game dynamics changed when Matthias put our moves up the whiteboard, and asked “What happens if your behavior is up here for everyone to see?” Trust was now dependent on reputation, and we had to change our strategies.

Matthias pointed out that during a tournament organised by Robert Axelrod in the early 1980s, a “tit for tat” strategy, in which players punished bad and rewarded good behavior, appeared to be the most effective.

For example:

Is There a Winning Strategy?

We then discussed ways to manipulate the “tit for tat” strategy with betrayal in the last round, when retaliation is impossible:

What we Learned

After the game we had an interesting conversation, comparing game scenarios with our own experiences.

One of us talked about a star developer whose toxic behavior was never challenged, due to his critical role in the project.

Another participant shared a story about a developer who was shouldering all the work while her colleagues were just coasting along.

We then talked about the importance of forgiveness. Following a round of mutual betrayal, players need to forgive each other, or risk becoming trapped in an endless cycle of punishment.

What became clear to us was that for any collaboration to succeed, addressing poor behavior and offering forgiveness are both essential.

More Games at Play14

During the same Play14 event, Matthias also hosted a game of Okaloa Flowlab, which was an equally interesting experience.