[Question] A ‘Practice of Rationality’ Sequence?

(This is not your typ­i­cal fac­tual-an­swer ques­tion, but I think it makes sense to for­mat this as a ques­tion rather than a post.)

TLDR: Recom­mend some posts for a “prac­tice of ra­tio­nal­ity” se­quence I want to cu­rate! Propos­ing posts that should ex­ist but don’t is also cool.

I’ve been think­ing re­cently that it would be nice if ra­tio­nal­ity were more as­so­ci­ated with a prac­tice—a set of skills which you can keep grind­ing and lev­el­ing up. Testable ra­tio­nal­ity skills (like ac­cu­racy or cal­ibra­tion in fore­cast­ing) are ob­vi­ously a plus, but I’m not refer­ring ex­clu­sively to this—some very real things can be hard to eval­u­ate ex­ter­nally, such as emo­tional well­ness.

A model I have in mind is med­i­ta­tion: med­i­ta­tion is easy to “grind” be­cause the med­i­ta­tor gets con­stant im­me­di­ate feed­back about how well they’re fo­cus­ing (or at least, they get that feed­back if they meet a min­i­mum of fo­cus re­quired to keep track of whether they are fo­cus­ing). Yet it’s quite difficult to eval­u­ate progress from the out­side.

(In fact, when I men­tioned this de­sire for a “prac­tice” of ra­tio­nal­ity to one friend, they were like “I agree, and in fact I think the prac­tice should just be in­sight med­i­ta­tion.”)

This is ba­si­cally re­it­er­at­ing Brienne’s call for tor­toise skills (see also), ex­cept what I want to do is col­lect pro­posed things which could be part of a prac­tice.

Ob­vi­ously, some CFAR con­tent could already qual­ify. CFAR doesn’t ex­actly teach it that way—as far as I’ve ob­served, CFAR’s fo­cus is on mind­set in­ter­ven­tions. “Mind­set in­ter­ven­tion” is the fancy psy­chol­ogy term for get­ting some­one to think differ­ently by hav­ing them do some­thing once. For ex­am­ple, the point of “growth mind­set” in­ter­ven­tions is that you ex­plain it once and this has long-last­ing im­pact on some­one’s be­hav­ior. Another mind­set in­ter­ven­tion is: you ask peo­ple to write about what mat­ters to them. Do­ing this once has shown long-term re­sults.

In my first CFAR ex­pe­rience (which was an MSFP, fwiw), the phrase “It’s not about the ex­er­cises!” was kind of a motto. It was ex­plained at the be­gin­ning that CFAR teaches ex­er­cises not be­cause peo­ple learn the ex­er­cises and then go out and use the ex­er­cises, but rather, go­ing through the ex­er­cises a few times changes how you think about things. (The story was that peo­ple of­ten go to a CFAR work­shop and then im­prove a bunch of things in their life, but say “but I haven’t been do­ing the ex­er­cises!”.)

But many of the things CFAR teaches could be used as a prac­tice, and (again refer­ring to my first CFAR ex­pe­rience) CFAR does do some things which en­courage you to look at them that way, like the fol­low-up emails which en­courage you to over­learn one ex­er­cise per week (prac­tic­ing that one thing a bunch so that it be­comes an au­to­matic men­tal mo­tion).

Another ex­am­ple point­ing at what I want here is be­well­tuned.com. The con­tent may or may not be right, but the sort of thing seems ex­actly right to me—ac­tion­able skills you can keep work­ing on reg­u­larly af­ter get­ting sim­ple ex­pla­na­tions of how to do it. And fur­ther­more, the pre­sen­ta­tion seems ex­actly right. LessWrong has a ten­dency to fo­cus on wordy ex­pla­na­tions of in­tel­lec­tual top­ics, which is great, but the be­well­tuned style seems like an ex­cel­lent coun­ter­bal­ance.

I’m us­ing the “ques­tion” for­mat so that an­swers can recom­mend spe­cific things (per­haps rep­re­sented by ex­ist­ing LW posts, per­haps not), whereas com­ments can dis­cuss this more broadly (such as what more gen­eral crite­ria should be ap­plied to filter sug­ges­tions, or whether this is even a good idea). The an­swer list here could serve as a big repos­i­tory. I’ll prob­a­bly cre­ate a se­quence which can be my own highly opinionated cu­ra­tion of the sug­ges­tions here, plus my own writ­ing on the sub­ject.

I origi­nally in­tended Be­com­ing Unusu­ally Truth Ori­ented to be the start of a se­quence on the sub­ject writ­ten en­tirely by me. How­ever, some re­sult­ing dis­cus­sion made me ques­tion my ap­proach (hence the mo­ti­va­tion for this ques­tion).

One friend of mine (go­ing off of some of the dis­cus­sion in com­ments to that post) voiced a con­cern about the ra­tio­nal­ity com­mu­nity fal­ling into the same pit­falls as mar­tial arts. Sev­eral ar­ti­cles about this have been writ­ten on LW. (I’m not find­ing all the ones I re­mem­ber! If you put links to more of them in the com­ments I’ll prob­a­bly edit this to add them.) The con­cern is that a mar­tial art of ra­tio­nal­ity could lead to the same kinds of epistemic vi­cious­ness which are seen in literal mar­tial arts—a prac­tice di­vorced from re­al­ity due to the con­straints and in­cen­tives of train­ing/​teach­ing.

That same friend sug­gested that the solu­tion was to fo­cus on em­piri­cally ver­ifi­able skills, namely fore­cast­ing. But in the in-per­son ra­tio­nal­ist com­mu­nity in the bay area, I’ve en­coun­tered some crit­i­cism of ex­treme fo­cus on fore­cast­ing which sug­gests that it’s mak­ing the very mis­take we’re afraid of here—Good­hart­ing on the prob­lem. One per­son asked me to give any ex­am­ples of Su­perfore­cast­ing-like skills re­sult­ing in ac­tual ac­com­plish­ments, sug­gest­ing that plan­ning is the far more valuable skill and varies sig­nifi­cantly from fore­cast­ing. Another per­son re­counted their ex­pe­rience sit­ting down with sev­eral other ra­tio­nal­ists to learn su­perfore­cast­ing skills. It was a group of rather com­mit­ted and also in­di­vi­d­u­ally com­pe­tent ra­tio­nal­ists, but they quickly came to the con­clu­sion that while they could put in the effort to be­come much bet­ter at fore­cast­ing, the ac­tual skills they’d learn would be highly spe­cific to the task of win­ning points in pre­dic­tion tasks, and they aban­doned the pro­ject, con­clud­ing that it would not mean­ingfully im­prove their gen­eral ca­pa­bil­ity to ac­com­plish things!!

So, this seems like a hard prob­lem.

What could/​should be a part of a ‘prac­tice’ of ra­tio­nal­ity?