I'm not sure what's silly about it. Just because there's only one game of IPD doesn't mean there can't be multiple rounds of communication before, during, and after each iteration.
I'm not talking about communication in general; only about one-time precommitting. If it is asymmetrical I don't consider that real IPD anymore, and if it isn't, I fail to see how it would change anything, unless you would find some way for both players to precommit and take their opponent's precommitment into consideration.
I didn't mean to offend you by saying that it was stupid; I just don't find it an interesting problem from a game theory perspective.
As for the bet, it is easy to set up an arena of strategies where any particular strategy would fare worse than a certain other strategy—in this case, a pool of RandomBot derivatives would be a fine arena for DefectBot to excel in. Also, "a wide variety" is far too arbitrary for any bets.
What I had in mind as the "best" strategy is the strategy that scores highest against the strategy that is designed to score highest against it. TFT scores 198 points (paperclips, million lives saved, whatever) against 99C1D, which is the best way to deal with TFT. However, I should probably just give you the $500, since it turns out 198 is way too low even for a purely deterministic strategy. I'm now ~100% certain (haha!) that the highest possible score is 248; and if you consider non-deterministic strategies, the score increases to 248.5-e. And sadly, there's no way for you to score higher than 101 respectively 100.5+e against this strategy.
...What I had in mind as the "best" strategy is the strategy that scores highest against the strategy that is designed to score highest against it. TFT scores 198 points (paperclips, million lives saved, whatever) against 99C1D, which is the best way to deal with TFT. However, I should probably just give you the $500, since it turns out 198 is way too low even for a purely deterministic strategy. I'm now ~100% certain (haha!) that the highest possible score is 248; and if you consider non-deterministic strategies, the score increases to 248.5-e. An
Today's post, The Truly Iterated Prisoner's Dilemma was originally published on 04 September 2008. A summary (taken from the LW wiki):
Discuss the post here (rather than in the comments to the original post).
This post is part of the Rerunning the Sequences series, where we'll be going through Eliezer Yudkowsky's old posts in order so that people who are interested can (re-)read and discuss them. The previous post was The True Prisoner's Dilemma, and you can use the sequence_reruns tag or rss feed to follow the rest of the series.
Sequence reruns are a community-driven effort. You can participate by re-reading the sequence post, discussing it here, posting the next day's sequence reruns post, or summarizing forthcoming articles on the wiki. Go here for more details, or to have meta discussions about the Rerunning the Sequences series.