For example, what would be inappropriately off topic to post to LessWrong discussion about?
I couldn't find an answer in the FAQ. (Perhaps it'd be worth adding one.) The closest I could find was this:
What is Less Wrong?
Less Wrong is an online community for discussion of rationality. Topics of interest include decision theory, philosophy, self-improvement, cognitive science, psychology, artificial intelligence, game theory, metamathematics, logic, evolutionary psychology, economics, and the far future.
However "rationality" can be interpreted broadly enough that rational discussion of anything would count, and my experience reading LW is compatible with this interpretation being applied by posters. Indeed my experience seems to suggest that practically everything is on topic; political discussion of certain sorts is frowned upon, but not due to being off topic. People often post about things far removed from the topics of interest. And some of these topics are very broad: it seems that a lot of material about self-improvement is acceptable, for instance.
I propose a few ways of using these words.
"Instrumentally rational" = successful on account of being epistemically rational. Success for other reasons does not count.
"Epistemically rational" = correctly using observation and reasoning to arrive at true beliefs, and making the decisions and actions recommended by those beliefs.
"Irrational" = a privative concept, expressing not merely an absence of rationality, but an absence where presence is seen as having been possible.
How widely or narrowly you draw the line around "irrational" depends on how much of a counterfactual difference you are imagining being possible. Draw it widely enough, and sandstone is irrational for being so easily eroded by the wind. Narrowly enough, and everyone is perfectly rational all the time, in the same way that the Pentium III with the FDIV bug worked flawlessly. Somewhere in between there is a reasonable place to draw that line, but arguing over where to draw it is an argument about what to call things, not an argument about what properties those things have.
I agree.