wedrifid comments on Any existential risk angles to the US presidential election? - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (213)
This meme just will not seem to die.
No, not all the assumptions made in the thought experiments designed to show TDT cooperating and CDT defecting (and TDT benefiting from the difference) are present in the specific case of a human deciding whether to vote in a national election. The other agents are not behaving remotely like TDT or UDT agents and a TDT agent would defect and benefit from doing so. And then the next election would come around and they would do the same thing.
TDT doesn't mean act like a care bear!
(Expanding on this out of interest, and assuming a case where there are enough TDTish agents in your population for it to actually be sane to consider cooperating.)
The assumption required is not quite whether you have strong preferences but what the preferences of all the TDTish agents are (or are estimated to be). If there is a group of agents implementing decision theories like TDT who are all willing to cooperate if that's what it takes to make the other people in that group cooperate and it happens that half of them are Greens and half are Blues then they do cooperate by staying home!