Survivor and game theory

[Questor]

On Mon, 26 Sep 2016 11:37:21 -0700, The Horny Goat <lcr...@home.ca> wrote:
>Back about 10 years ago there was a mathematics graduate student who
>demonstrated mathematically that there was a "perfect strategy" for
>games of the sort where one player got evicted every round. It didn't
>specify what the strategy WAS, merely that it was mathematically
>proven one existed. (As someone whose undergraduate degree was
>mathematics I knew enough of game theory to understand about 3/4 of
>it)
>
>The assumptions were (and I'm going from memory here
>- no advance collusion between players
>- all players fighting for 1st place equally with no other side goals
>and 2 or 3 more assumptions I don't recall

Wouldn't that first assumption -- no advance collusion -- invalidate Survivor as
a candidate for this so-called "perfect strategy?" (Not that I think that such
a thing exists for Survivor.) The second assumption is also regularly violated
on Survivor. And then there are those other assumptions...

Despite my skepticism, about ten years ago there was another popular game show
where one player gets evicted every round that probably can be more accurately
modelled as you describe -- "The Weakest Link." There you have the dynamic of
voting smarter people off early so you don't have to face them in the final
round versus keeping them so they can answer the questions correctly and
accumulate prize money. There's no collusion among players between rounds,
everyone has essentially the same goal, and, most importantly from the viewpoint
of game theory, all the information about the game is known to all players.


>The article DID do their calculations on the basis that there were
>'immunity competitions' but this was before the day of the HII which
>in my opinion didn't invalidate the assumptions (though players
>knowing each other in advance with collusion possible as in every
>season since S8 "Allstars" certainly did).
>
>In short the grad student's 'game' was much simpler than the one we
>watch and even from the two assumptions I remember it's clear his
>conclusion wouldn't necessarily hold now.

Yep. I also pursued an undergraduate degree in mathematics and I've thought a
bit about the application of game theory to Survivor. I know that Survivor
resembles -- resembles, mind you -- an iterated, multi-player Prisoner's
Dilemma. But as noted above, there's collusion, conflicting goals, and hidden
information in Survivor. Also, while game theory analysis may provide an
optimal strategy, it is rarely a guarantee of success. Even in the absence of
mistakes, games with an element chance can be lost. An optimal strategy is
proved over many trials, but most people play Survivor (or the Weakest Link)
only once.

While there may never be any definitive, over-all strategy for Survivor, many
individual decisions and isolated events are amenable to analysis and
examination in the light of game theory. I'm sure more than a few players have
sketched payoff matrices in the sand in an effort to determine which alliance
member to eliminate and how different potential jury members might vote at
final tribal council.

Below are some interesting web pages that further explore this topic.

--------------------------------------------------
Outwit, Outplay, Outlast: Survivor and its use of Game Theory
https://blogs.cornell.edu/info2040/2014/09/25/outwit-outplay-outlast-survivor-and-its-use-of-game-theory/

An essay on Richard Hatch's application of game theory during the final immunity
challenge in the first season of Survivor.
--------------------------------------------------
Is the Key to Survivor in 'Non-Cooperative Games'?
http://www.gametheory.net/news/Items/080.html

A general discussion of the possible use of game theory in Survivor, spurred by
Jeff Probst's comments at the end of Survivor Thailand.
--------------------------------------------------
Yul Kwon's use of Prisoner's Dilemma Strategies
http://survivorsucks.yuku.com/topic/1579/Yuls-use-of-Prisoners-Dilemma-Strategies

An overview of how Yul Kwon's actions on Survivor Cook Islands reflect
successful strategies used in an interated Prisoner's Dilemma.
--------------------------------------------------
A Survivor Strategy Blog. Analyzing Survivor: Redemption Island with a touch of
Game Theory, Evolutionary Psychology, and Behavioral Psychology
http://survivortheory.blogspot.com/

After a promising start, this blog stopped with the third episode and only five
posts.
--------------------------------------------------
Teaching Prisoners' Dilemma Strategies in Survivor
http://www.academia.edu/1607681/Teaching_Prisoners_Dilemma_Strategies_in_Survivor_Reality_Television_in_the_IR_Classroom

"The reality television program Survivor is used as a teaching tool for
presenting the prisoners' dilemma. Structural similarities between the format
of reality television and game theory, rule-bound competitions with clear
payoff, enable students to critically examine the strategies that contestants
use, providing a clear pedagogical utility."

This paper is written in dense academ-ese, and it is aimed at using Survivor as
an example for teaching the use of Prisoner's Dilemma strategies in the context
of international relations, but it does have some interesting material
describing season one and season eight (All-Stars).
--------------------------------------------------
Short history of iterated prisoner’s dilemma tournaments
https://egtheory.wordpress.com/2015/03/02/ipd/

[The Horny Goat]

On Mon, 03 Oct 2016 17:16:57 GMT, use...@only.tnx (Questor) wrote:

>Yep. I also pursued an undergraduate degree in mathematics and I've thought a
>bit about the application of game theory to Survivor. I know that Survivor
>resembles -- resembles, mind you -- an iterated, multi-player Prisoner's
>Dilemma. But as noted above, there's collusion, conflicting goals, and hidden
>information in Survivor. Also, while game theory analysis may provide an
>optimal strategy, it is rarely a guarantee of success. Even in the absence of
>mistakes, games with an element chance can be lost. An optimal strategy is
>proved over many trials, but most people play Survivor (or the Weakest Link)
>only once.

Well the big assumption is that 'nobody plays for other than first
place" whereas we all know ego and playing for face time (which may or
may not be linked to post-game personal opportunities) very definitely
plays a role particularly in late game.

Early evictees ALWAYS talk about 'adventure of a lifetime'

Thanks for your links!