[Question] What should experienced rationalists know?

The ob­vi­ous an­swer is ‘the se­quences’ but imo that is nei­ther nec­es­sary nor suffi­cient. The Se­quences are valuable but they are quite old at this point. They also run to over a mil­lion words (though Ra­tion­al­ity AtZ is only ~600k). Here is a list of core skills and ideas:

1 - Ra­tion­al­ity Techniques

Ideally, an ex­pe­rienced ra­tio­nal­ist would have ex­pe­rience with most of the CFAR man­ual. Any­one try­ing to learn this ma­te­rial needs to ac­tu­ally try the tech­niques; the­o­ret­i­cal knowl­edge is not enough. If I had to make a shorter list of tech­niques I would in­clude:

  • Dou­ble Crux /​ In­ter­nal DC

  • Five-minute Timers

  • Trig­ger Ac­tion Plans

  • Bucket Errors

  • Goal/​Aver­sion Factoring

  • Gears Level Understanding

  • Nega­tive Vi­su­al­i­sa­tion /​ Mur­phy-Jitsu

  • Focusing

2 - AI Risk: Superintelligence

The ra­tio­nal­ity com­mu­nity was founded to help solve AI risk. Su­per­in­tel­li­gence gives an up­dated and com­plete ver­sion of the ‘clas­sic’ ar­gu­ment for AI-risk. Su­per­in­tel­li­gence does not make as many strong claims about take­off as Elizer’s early writ­ings. This seems use­ful given that po­si­tions closer to Paul Chris­ti­ano’s seem to be gain­ing promi­nence. I think the ‘clas­sic’ ar­gu­ments are still very much worth un­der­stand­ing. On the other hand, Su­per­in­tel­li­gence is ~125K words and not easy read­ing.

I think many read­ers can skip the first few chap­ters. The core ar­gu­ment is in chap­ters five through four­teen.

5. De­ci­sive strate­gic advantage

6. Cog­ni­tive superpowers

7. The su­per­in­tel­li­gent will

8. Is the de­fault out­come doom?

9. The con­trol problem

10. Or­a­cles, ge­nies, sovereigns, tools

11. Mul­tipo­lar scenarios

12. Ac­quiring values

13. Choos­ing the crite­ria for choosing

14. The strate­gic picture

3 - Cog­ni­tive Bi­ases: Think­ing Fast and Slow

Prim­ing is the first re­search area dis­cussed in depth in TF&S. Prim­ing seems to be al­most en­tirely BS. I would sug­gest skip­ping the chap­ter on prim­ing and re­mem­ber­ing the dis­cus­sion of the ‘hot hand fal­lacy’ seems in­cor­rect. Another po­ten­tial down­side is the length (~175K words). How­ever, I don’t think there is a bet­ter source over­all. Many of the con­cepts in TF&S re­main fun­da­men­tal. The writ­ing is also quite good and the his­tor­i­cal value is ex­tremely high. Here is a quick re­view from 2016.

4 - Statistics

It is hard to be an in­formed ra­tio­nal­ist with­out a ba­sic un­der­stand­ing of Bayesian think­ing. You need to un­der­stand fre­quen­tist statis­tics to eval­u­ate a lot of rele­vant re­search. Some of the most im­por­tant con­cepts/​goals are listed be­low.

Bayesian Statis­tics:

  • Illus­trate the use of odd’s ra­tio calcu­la­tion in prac­ti­cal situations

  • Derive Laplace’s rule of succession

Fre­quen­tist Stats—Un­der­stand the fol­low­ing con­cepts:

  • Law of large numbers

  • Power, p-val­ues, t-tests, z-tests

  • Lin­ear Regression

  • Limi­ta­tions of the above concepts

5 - Sig­nal­ling /​ The Elephant in the Brain

The Elephant in the Brain is a clear and au­thor­i­ta­tive source. The ideas dis­cussed have cer­tainly been in­fluen­tial in the ra­tio­nal­ist com­mu­nity. But I am not what epistemic sta­tus the com­mu­nity as­signs to the Han­son/​Sim­ler the­o­ries around sig­nal­ing. Any opinions? For refer­ence here are the top­ics.

PART I Why We Hide Our Mo­tives

  • 1 An­i­mal Behavior

  • 2 Competition

  • 3 Norms

  • 4 Cheating

  • 5 Self-Deception

  • 6 Coun­terfeit Reasons

PART II Hid­den Mo­tives in Every­day Life

  • 7 Body Language

  • 8 Laughter

  • 9 Conversation

  • 10 Consumption

  • 11 Art

  • 12 Charity

  • 13 Education

  • 14 Medicine

  • 15 Religion

  • 16 Politics

  • 17 Conclusion

What am I miss­ing? Try to be as spe­cific as pos­si­ble about what ex­actly should be learned. Some pos­si­ble top­ics dis­cussed in the com­mu­nity in­clude:

  • Economics

  • The ba­sics of the other EA cause ar­eas and gen­eral the­ory? (at least the stuff in ‘Do­ing Good Bet­ter’)

  • Eliezer says to study evolu­tion­ary psy­chol­ogy in the eleventh virtue but I have not been im­pressed with evo-psych.

  • Some­thing about men­tal tech? Maybe mind­ful­ness, In­ter­nal Fam­ily Sys­tems, or cir­cling? I am not con­fi­dent any­thing in space fits.