11 core rationalist skills

An ex­cel­lent way to im­prove one’s skill as a ra­tio­nal­ist is to iden­tify one’s strengths and weak­nesses, and then ex­pend effort on the things that one can most effec­tively im­prove (which are of­ten the ar­eas where one is weak­est). This seems es­pe­cially use­ful if one is very spe­cific about the parts of ra­tio­nal­ity, if one de­scribes them in de­tail.

In or­der to fa­cil­i­tate im­prov­ing my own and oth­ers’ ra­tio­nal­ity, I am post­ing this list of 11 core ra­tio­nal­ist skills, thanks al­most en­tirely to Anna Sala­mon.

  • Keep your eyes on the prize. Fo­cus your mod­el­ing efforts on the is­sues most rele­vant to your goals. Be able to quickly re­fo­cus a train of thought or dis­cus­sion on the most im­por­tant is­sues, and be able and will­ing to quickly kill tempt­ing tan­gents. Pe­ri­od­i­cally stop and ask your­self “Is what I am think­ing about at the mo­ment re­ally an effec­tive way to achieve my stated goals?”.

  • En­tan­gle your­self with the ev­i­dence. Real­ize that true opinions don’t come from nowhere and can’t just be painted in by choice or in­tu­ition or con­sen­sus. Real­ize that it is in­for­ma­tion-the­o­ret­i­cally im­pos­si­ble to re­li­ably get true be­liefs un­less you ac­tu­ally get re­li­ably pushed around by the ev­i­dence. Dist­in­guish be­tween knowl­edge and feel­ings.

  • Be Cu­ri­ous: Look for in­ter­est­ing de­tails; re­sist cached thoughts; re­spond to un­ex­pected ob­ser­va­tions and thoughts. Learn to ac­quire in­ter­est­ing an­gles, and to make con­nec­tions to the task at hand.

  • Au­mann-up­date: Up­date to the right ex­tent from oth­ers’ opinions. Bor­row rea­son­able prac­tices for gro­cery shop­ping, so­cial in­ter­ac­tion, etc from those who have already worked out what the best way to do these things is. Take rele­vant ex­perts se­ri­ously. Use out­side views to es­ti­mate the out­come of one’s own pro­jects and the merit of one’s own clever ideas. Be will­ing to de­part from con­sen­sus in cases where there is suffi­cient ev­i­dence that the con­sen­sus is mis­taken or that the com­mon prac­tice doesn’t serve its os­ten­si­ble pur­poses. Have cor­rect mod­els of the causes of oth­ers’ be­liefs and psy­cholog­i­cal states, so that you can tell the differ­ence be­tween cases where the vast ma­jor­ity of peo­ple be­lieve false­hoods for some spe­cific rea­son, and cases where the vast ma­jor­ity ac­tu­ally knows best.

  • Know stan­dard Bi­ases: Have con­scious knowl­edge of com­mon hu­man er­ror pat­terns, in­clud­ing the heuris­tics and bi­ases liter­a­ture; prac­tice us­ing this knowl­edge in real-world situ­a­tions to iden­tify prob­a­ble er­rors; prac­tice mak­ing pre­dic­tions and up­date from the track record of your own ac­cu­rate and in­ac­cu­rate judg­ments.

  • Know Prob­a­bil­ity the­ory: Have con­scious knowl­edge of prob­a­bil­ity the­ory; prac­tice ap­ply­ing prob­a­bil­ity the­ory in real-world in­stances and see­ing e.g. how much to pe­nal­ize con­junc­tions, how to regress to the mean, etc.

  • Know your own mind: Have a mo­ment-to-mo­ment aware­ness of your own emo­tions and of the mo­ti­va­tions guid­ing your thoughts. (Are you search­ing for jus­tifi­ca­tions? Shy­ing away from cer­tain con­sid­er­a­tions out of fear?) Be will­ing to ac­knowl­edge all of your­self, in­clud­ing the petty and un­sa­vory parts. Knowl­edge of your own track record of ac­cu­rate and in­ac­cu­rate pre­dic­tions, in­clud­ing in cases where fear, pride, etc. were strong.

  • Be well cal­ibrated: Avoid over- and un­der-con­fi­dence. Know how much to trust your judg­ments in differ­ent cir­cum­stances. Keep track of many lev­els of con­fi­dence, pre­ci­sion, and sur­pris­ing­ness; dare to pre­dict as much as you can, and up­date as you test the limits of your knowl­edge. Develop as pre­cise a world-model as you can man­age. (Tom McCabe wrote a quiz to test some sim­ple as­pects of your cal­ibra­tion.)

  • Use an­a­lytic philos­o­phy: un­der­stand the habits of thought taught in an­a­lytic philos­o­phy; the habit of fol­low­ing out lines of thought, of tak­ing on one is­sue at a time, of search­ing for counter-ex­am­ples, and of care­fully keep­ing dis­tinct con­cepts dis­tinct (e.g. not con­fus­ing heat and tem­per­a­ture; free will and lack of de­ter­minism; sys­tems for talk­ing about Peano ar­ith­metic and sys­tems for talk­ing about sys­tems for talk­ing about Peano ar­ith­metic).

  • Re­sist Thoughtcrime. Keep truth and virtue ut­terly dis­tinct in your mind. Give no quar­ter to claims of the sort “I must be­lieve X, be­cause oth­er­wise I will be {racist /​ with­out moral­ity /​ at risk of com­ing late to work/​ kicked out of the group /​ similar to stupid peo­ple}”. De­cide that it is bet­ter to merely lie to oth­ers than to lie to oth­ers and to your­self. Real­ize that goals and world maps can be sep­a­rated; one can pur­sue the goal of fight­ing against cli­mate change with­out de­liber­ately fool­ing one­self into hav­ing too high an es­ti­mate (given the ev­i­dence) of the prob­a­bil­ity that the an­thro­pogenic cli­mate change hy­poth­e­sis is cor­rect.