Conversation

Yes; as I said, my impression is that the LW-derived community has gone well beyond “rationalism” as it is classically defined, and incorporates significant elements of meta-rationality, and I am enthusiastic and supportive of that!
1
2
Rationalists are very aware of “rationalism” as a different body of thought from ours that frustratingly has a similar name. I know it’s confusing & I’m sorry! But blame decision science, not us - they chose to use the word rational for utility maximization, & we followed suit
3
5
This Tweet is from a suspended account. Learn more
Idk, if we had invented a new name for a concept that already exists in decision science I feel like ppl would be griping about that too.
1
5
Yes, some people (like me) will always find something to complain about :) Is your conception of “rationality” in fact restricted to that of decision science? My impression is that the community has a broader sense of it (but individuals vary, and yours may be more specific).
1
I never asked Eliezer directly but assume he must have gotten those definitions from the decision science literature (which includes parts of economics, philosophy, psychology and comp sci relevant to modeling normative decision making) bc that's how that literature uses the word
1
2
The most relevant phrases to how we talk about rationality are probably "normative rationality," "bounded rationality" and "rational choice theory"
2
5
Thank you! I will chew on this overnight… My immediate reaction is “yes, this is familiar mainstream stuff, but I don’t think this is quite what the LW-sphere advocates, even taking into account community diversity”—but I need to think that through carefully before saying so!
1
2
Yeah 99% of the work is figuring out how a human, with a messy human brain, can approximate the normative model. What heuristics work well, in what contexts? When is following explicit rules useful vs. just training your intuition through experience? etc. Lots of diff. ideas here
2
3
Show replies
This Tweet is from a suspended account. Learn more
That’s my kneejerk concern here. It’s not that that is a *wrong* idea, it’s that it tends to point in empirically unhelpful directions. Like “limit inference to k deductive steps / n compute cycles.” Not that anyone in the LW sphere advocates that, but it’s what comes to mind 1st
2
1
Show replies
Do you think that in practice LW type rationalists internalize Bounded Rationality and its ilk? It seems incompatible with the commonly held “shut up and multiply” which tends towards Straw Vulcanism.