LESSWRONG
LW

2386
csj
1030
Message
Dialogue
Subscribe

Posts

Sorted by New

Wikitag Contributions

Comments

Sorted by
Newest
No posts to display.
No wikitag contributions to display.
Proving Too Much
csj5d21

This just seems like proof by contradiction. Argument bad because can reach obviously false conclusions from it. Doesn't mean original statement is wrong, but go get yourself a narrower argument 

Reply
All Debates Are Bravery Debates
csj6d10

Seems useful to distinguish when disagreement comes from taking different logical steps versus different priors on what the world looks like. The above all being examples of different priors or beliefs about who the average guy is and what he needs to hear not necessarily differences in logic

Reply
Spooky Collusion at a Distance with Superrational AI
csj9d10

Ran a pretty similar experiment just on PD variants using some different framings with temperature = 0. Saw high cooperation rates when the other player was described as "an identical reasoning model" and "another instance of yourself (same model)" for several models. 

Mistral models were a bit of an anomaly; they'd cooperate but often because they were assuming goodwill on behalf of the opponent or prioritizing collective outcomes rather than citing a shared reasoning process. 

code here:  https://www.expectedparrot.com/content/clajelli/superrationality-game-theory

No description has been provided for this image
Reply