Scribd
Upload
24 views2 pages

Game Theory - Steam Update Article

Uploaded by

leagalliencurrier
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
24 views2 pages

Game Theory - Steam Update Article

Uploaded by

leagalliencurrier
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

What game theory and the prisoner's dilemma reveal about

cooperation
The prisoner's dilemma

The prisoner's dilemma is possibly one of the most well-known thought experiments relating to
game theory. Its fame is due to its prevalence in real-world situations; the strategic theory behind
this dilemma models many scenarios in the real world concerning the relationships between two
parties. This is seen everywhere from politics and world conflict to the behaviour of animals in
nature.

To summarise, the prisoner's dilemma involves prisoners separately being interrogated for a crime
they are suspected of committing together.
Each prisoner has two options: to cooperate with their
partner by staying silent or to betray their partner by
confessing. Below shows the combination of possible
outcomes, where R is the sentence given in each scenario.

The theoretical strategy behind this is clear -> with the


unknown of how the other prisoner will act, the option to
testify if favourable- you will either get 0 years or 2 years
vs 1 year or 3 years if you were to stay silent.

With the better strategy for both players being to testify, the
result will end up with both of them getting a 2-year prison sentence, the overall worst result for
them both. This is a clear example of a situation where the use of logical reasoning to act in one’s
self-interest ends up in a situation worse off for everyone. It is seen extremely often in the real
world; an example of this being nuclear armament - the perceived strategic advantage for countries
of owning nuclear bombs has led to a mass weaponisation of the world, the result of which
is negative for everyone, as now every country has to live with the risk of potential mutual
destruction.

The iterative prisoner's dilemma

The problem with the prisoner's dilemma is that it involves only a single interaction, so the
prisoners have no idea what kind of behaviour to expect from one another.

In real-world scenarios, it is extremely uncommon that two parties only ever interact once,
hence a better game theory experiment is the iterated prisoner's dilemma. The
iterated prisoner's dilemma is when two parties engage in the prisoner's dilemma repeatedly, with
the memory of their opponent's past actions and the ability to adjust their strategies accordingly.

Finding the best strategy for this (especially when the number of iterations is unknown), is much
more complex and in 1984, the political scientist Robert Axelrod decided to investigate it by setting
up a ‘tournament’. Axelrod extended invitations to academic peers worldwide to develop computer
strategies for competing in an iterated prisoner's dilemma tournament. The programs would each
play each other for a large number of rounds - greater than 200, however, the exact number they
would play was not given to the program developers.

This was done as if there is a finite number of rounds, and both players are aware of the number of
rounds, the best strategy is to always defect (betray). This comes from the idea that if you know
how many rounds you are playing, you may as well defect on the last round as this is the dominant
strategy for single interactions. However if both algorithms know the other algorithm will defect on
the last round, they may as well defect on the round before, and the round before that, and so on.

The submitted programs exhibited differing characteristics, including algorithmic complexity,


initial aggression, ability to forgive, and other factors. In each round, instead of prison sentence
years, the strategies would accumulate points and, at the end of the tournament, the strategies were
ranked by the points accumulated.

The winning strategy that Axelrod found was also the simplest, written in only four lines of code -
tit for tat. This strategy was to cooperate on the first move, and after that, copy whatever the other
player did on its previous move. He found the worst performing strategies to be those which were
the most ‘greedy’ - the ones designed to defect more in an attempt to take advantage of the other
player.

After analysis of the most successful strategies in his tournament, he identified the best
characteristics a program should have to be successful:

• Nice - the program should cooperate on the first round

• Forgiving - although the program should retaliate, it shouldn’t ‘hold a grudge’ for more than one
round if the other program cooperates again. This avoids cycles of revenge, maximising the points
the strategies can gain by mutually cooperating.

• Non-envious - the program shouldn’t aim to get a better score in the round than the other program.

• Retaliating - the program should retaliate and defect in response to the other program defecting.

I find this result incredibly interesting in what it reveals about the qualities one should have to be
more ‘successful’ overall when interacting with others, especially due to how prevalent
situations exactly like or similar to the iterative prisoner's dilemma are. This game theory idea
governs the principle of cooperation and is thought to be a possible explanation for the mechanism
in which trust and altruism evolved to be a part of human behaviour.

References:

“What Game Theory Reveals About Life, The Universe, and Everything” – YouTube, uploaded by Veritasium, 23
December 2023, Available at: https://www.youtube.com/watch?v=mScpHTIi-kM&t=1413s

“The Prisoner’s Dilemma” – Wikipedia, 20 April 2024, Available at: https://en.wikipedia.org/wiki/Prisoner


%27s_dilemma

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy