Evolving strategies through the Noisy Iterated Prisoner’s Dilemma has revealed all sorts of valuable insights into game theory and decision theory. Does anyone know of any similar tournaments where the payouts weren’t constant, so that any particular round might or might not qualify as a classic Prisoner’s Dilemma?
Evolving strategies through the Noisy Iterated Prisoner’s Dilemma has revealed all sorts of valuable insights into game theory and decision theory. Does anyone know of any similar tournaments where the payouts weren’t constant, so that any particular round might or might not qualify as a classic Prisoner’s Dilemma?
There have been many Iterated Prisoner’s Dilemma tournaments; at least a couple were done here on Less Wrong. Most such tourneys haven’t included noise; to find out about the ones that did, try googling for some combination of the phrases “contrite tit for tat”, “generous tit for tat”, “tit for two tats”, “pavlov”, and “grim”.
That question’s potentially ambiguous: does “previous interactions” mean previous moves within a single game, or previous games played? If the former, quite a bit of research on the PD played by finite state machines would fit. If the latter, Toby Ord’s work on the “societal iterated prisoner’s dilemma” would fit.
Not Quite the Prisoner’s Dilemma
Evolving strategies through the Noisy Iterated Prisoner’s Dilemma has revealed all sorts of valuable insights into game theory and decision theory. Does anyone know of any similar tournaments where the payouts weren’t constant, so that any particular round might or might not qualify as a classic Prisoner’s Dilemma?
Do you have a link for the original tournament?
There have been many Iterated Prisoner’s Dilemma tournaments; at least a couple were done here on Less Wrong. Most such tourneys haven’t included noise; to find out about the ones that did, try googling for some combination of the phrases “contrite tit for tat”, “generous tit for tat”, “tit for two tats”, “pavlov”, and “grim”.
Has there been research on Prisoner’s Dilemma where the players have limited amounts of memory for keeping track of previous interactions?
Google gives these:
http://www.pnas.org/content/95/23/13755.full.pdf
http://www.icmp.lviv.ua/journal/zbirnyk.79/33001/art33001.pdf
http://www.complex-systems.com/pdf/19-4-4.pdf
https://editorialexpress.com/cgi-bin/conference/download.cgi?db_name=ASSET2007&paper_id=287
http://ms.mcmaster.ca/~rogern4/pdf/publications_2009/annie_ltm.pdf
http://210.26.51.93/chenlab/images/4/4e/025.pdf
That question’s potentially ambiguous: does “previous interactions” mean previous moves within a single game, or previous games played? If the former, quite a bit of research on the PD played by finite state machines would fit. If the latter, Toby Ord’s work on the “societal iterated prisoner’s dilemma” would fit.