References & Resources for LessWrong

A list of refer­ences and re­sources for LW

Up­dated: 2011-05-24

  • F = Free

  • E = Easy (ad­e­quate for a low ed­u­ca­tional back­ground)

  • M = Memetic Hazard (con­tro­ver­sial ideas or works of fic­tion)


Do not flinch, most of LessWrong can be read and un­der­stood by peo­ple with a pre­vi­ous level of ed­u­ca­tion less than sec­ondary school. (And Khan Academy fol­lowed by Bet­terEx­plained plus the help of Google and Wikipe­dia ought to be enough to let any­one read any­thing di­rected at the sci­en­tifi­cally liter­ate.) Most of these refer­ences aren’t pre­req­ui­site, and only a small frac­tion are per­ti­nent to any par­tic­u­lar post on LessWrong. Do not be in­timi­dated, just go ahead and start read­ing the Se­quences if all this sounds too long. It’s much eas­ier to un­der­stand than this list makes it look like.

Nev­er­the­less, as it says in the Twelve Virtues of Ra­tion­al­ity, schol­ar­ship is a virtue, and in par­tic­u­lar:

It is es­pe­cially im­por­tant to eat math and sci­ence which im­p­inges upon ra­tio­nal­ity: Evolu­tion­ary psy­chol­ogy, heuris­tics and bi­ases, so­cial psy­chol­ogy, prob­a­bil­ity the­ory, de­ci­sion the­ory.


This list is hosted on, a com­mu­nity blog de­voted to re­fin­ing the art of hu­man ra­tio­nal­ity—the art of think­ing. If you fol­low the links be­low you’ll learn more about this com­mu­nity. It is one of the most im­por­tant re­sources you’ll ever come across if your aim is to get what you want, if you want to win. It shows you that there is more to most things than meets the eye, but more of­ten than not much less than you think. It shows you that even smart peo­ple can be com­pletely wrong but that most peo­ple are not even wrong. It teaches you to be care­ful in what you emit and to be skep­ti­cal of what you re­ceive. It doesn’t tell you what is right, it teaches you how to think and to be­come less wrong. And to do so is in your own self in­ter­est be­cause it helps you to at­tain your goals, it helps you to achieve what you want.


Why read Less Wrong?

A few ar­ti­cles ex­em­plify­ing in de­tail what you can ex­pect from read­ing Less Wrong, why it is im­por­tant, what you can learn and how it does help you.

Ar­tifi­cial Intelligence

Let an ul­train­tel­li­gent ma­chine be defined as a ma­chine that can far sur­pass all the in­tel­lec­tual ac­tivi­ties of any man how­ever clever. Since the de­sign of ma­chines is one of these in­tel­lec­tual ac­tivi­ties, an ul­tra-in­tel­li­gent ma­chine could de­sign even bet­ter ma­chines; there would then un­ques­tion­ably be an “in­tel­li­gence ex­plo­sion,” and the in­tel­li­gence of man would be left far be­hind. — I. J. Good, “Spec­u­la­tions Con­cern­ing the First Ul­train­tel­li­gent Ma­chine”


Friendly AI

The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for some­thing else. — Eliezer Yud­kowsky, Ar­tifi­cial In­tel­li­gence as a Pos­i­tive and Nega­tive Fac­tor in Global Risk

Ma­chine Learning

Not es­sen­tial but an valuable ad­di­tion for any­one who’s more than su­perfi­cially in­ter­ested in AI and ma­chine learn­ing.

The Tech­nolog­i­cal Singularity

The term “Sin­gu­lar­ity” had a much nar­rower mean­ing back when the Sin­gu­lar­ity In­sti­tute was founded. Since then the term has ac­quired all sorts of un­sa­vory con­no­ta­tions. — Eliezer Yudkowsky

Heuris­tics and Biases

One of the painful things about our time is that those who feel cer­tainty are stupid, and those with any imag­i­na­tion and un­der­stand­ing are filled with doubt and in­de­ci­sion. — Ber­trand Russell

Ig­no­rance more fre­quently begets con­fi­dence than does knowl­edge. — Charles Darwin

The heuris­tics and bi­ases pro­gram in cog­ni­tive psy­chol­ogy tries to work back­ward from bi­ases (ex­per­i­men­tally re­pro­ducible hu­man er­rors) to heuris­tics (the un­der­ly­ing mechanisms at work in the brain).


Learn­ing Mathematics

Here’s a phe­nomenon I was sur­prised to find: you’ll go to talks, and hear var­i­ous words, whose defi­ni­tions you’re not so sure about. At some point you’ll be able to make a sen­tence us­ing those words; you won’t know what the words mean, but you’ll know the sen­tence is cor­rect. You’ll also be able to ask a ques­tion us­ing those words. You still won’t know what the words mean, but you’ll know the ques­tion is in­ter­est­ing, and you’ll want to know the an­swer. Then later on, you’ll learn what the words mean more pre­cisely, and your sense of how they fit to­gether will make that learn­ing much eas­ier. The rea­son for this phe­nomenon is that math­e­mat­ics is so rich and in­finite that it is im­pos­si­ble to learn it sys­tem­at­i­cally, and if you wait to mas­ter one topic be­fore mov­ing on to the next, you’ll never get any­where. In­stead, you’ll have ten­drils of knowl­edge ex­tend­ing far from your com­fort zone. Then you can later back­fill from these ten­drils, and ex­tend your com­fort zone; this is much eas­ier to do than learn­ing “for­wards”. (Cau­tion: this back­filling is nec­es­sary. There can be a temp­ta­tion to learn lots of fancy words and to use them in fancy sen­tences with­out be­ing able to say pre­cisely what you mean. You should feel free to do that, but you should always feel a pang of guilt when you do.) — Ravi Vakil




Prob­a­bil­ities ex­press un­cer­tainty, and it is only agents who can be un­cer­tain. A blank map does not cor­re­spond to a blank ter­ri­tory. Ig­no­rance is in the mind. — Eliezer Yudkowsky

Math is fun­da­men­tal, not just for LessWrong. But es­pe­cially Bayes’ The­o­rem is es­sen­tial to un­der­stand the rea­son­ing un­der­ly­ing most of the writ­ings on LW.

Bayes' theorem



All the limi­ta­tive the­o­rems of meta­math­e­mat­ics and the the­ory of com­pu­ta­tion sug­gest that once the abil­ity to rep­re­sent your own struc­ture has reached a cer­tain crit­i­cal point, that is the kiss of death: it guaran­tees that you can never rep­re­sent your­self to­tally. Gödel’s In­com­plete­ness The­o­rem, Church’s Un­de­cid­abil­ity The­o­rem, Tur­ing’s Halt­ing The­o­rem, Tarski’s Truth The­o­rem — all have the flavour of some an­cient fairy tale which warns you that “To seek self-knowl­edge is to em­bark on a jour­ney which … will always be in­com­plete, can­not be charted on any map, will never halt, can­not be de­scribed.” — Dou­glas Hofs­tadter 1979


De­ci­sion theory

It is pre­cisely the no­tion that Na­ture does not care about our al­gorithm, which frees us up to pur­sue the win­ning Way—with­out at­tach­ment to any par­tic­u­lar rit­ual of cog­ni­tion, apart from our be­lief that it wins. Every rule is up for grabs, ex­cept the rule of win­ning. — Eliezer Yudkowsky

Re­mem­ber that any heuris­tic is bound to cer­tain cir­cum­stances. If you want X from agent Y and the rule is that Y only gives you X if you are a de­voted ir­ra­tional­ist then ¬ir­ra­tional. Un­der cer­tain cir­cum­stances what is ir­ra­tional may be ra­tio­nal and what is ra­tio­nal may be ir­ra­tional. Paul K. Fey­er­abend said: “All method­olo­gies have their limi­ta­tions and the only ‘rule’ that sur­vives is ‘any­thing goes’.”

Game Theory

Game the­ory is the study of the ways in which strate­gic in­ter­ac­tions among eco­nomic agents pro­duce out­comes with re­spect to the prefer­ences (or util­ities) of those agents, where the out­comes in ques­tion might have been in­tended by none of the agents. — Stan­ford En­cy­clo­pe­dia of Philosophy


With Re­lease 33-9117, the SEC is con­sid­er­ing sub­sti­tu­tion of Python or an­other pro­gram­ming lan­guage for le­gal English as a ba­sis for some of its reg­u­la­tions. — Will Wall Street re­quire Python?

Pro­gram­ming knowl­edge is not manda­tory for LessWrong but you should how­ever be able to in­ter­pret the most ba­sic pseudo code as you will come across var­i­ous snip­pets of code in dis­cus­sions and top-level posts out­side of the main se­quences.


Python is a gen­eral-pur­pose high-level dy­namic pro­gram­ming lan­guage.


Haskell is a stan­dard­ized, gen­eral-pur­pose purely func­tional pro­gram­ming lan­guage, with non-strict se­man­tics and strong static typ­ing.


Com­puter science

The in­tro­duc­tion of suit­able ab­strac­tions is our only men­tal aid to or­ga­nize and mas­ter com­plex­ity. — Eds­ger W. Dijkstra

One of the fun­da­men­tal premises on LessWrong is that a uni­ver­sal com­put­ing de­vice can simu­late ev­ery phys­i­cal pro­cess and that we there­fore should be able to re­verse en­g­ineer the hu­man brain as it is fun­da­men­tally com­putable. That is, in­tel­li­gence and con­scious­ness are sub­strate-neu­tral.

(Al­gorith­mic) In­for­ma­tion Theory


A poet once said, “The whole uni­verse is in a glass of wine.” We will prob­a­bly never know in what sense he meant that, for po­ets do not write to be un­der­stood. But it is true that if we look at a glass of wine closely enough we see the en­tire uni­verse. — Richard Feynman


Gen­eral relativity

You do not re­ally un­der­stand some­thing un­less you can ex­plain it to your grand­mother. ~ Albert Ein­stein

Quan­tum physics

An elec­tron is not a billiard ball, and it’s not a crest and trough mov­ing through a pool of wa­ter. An elec­tron is a math­e­mat­i­cally differ­ent sort of en­tity, all the time and un­der all cir­cum­stances, and it has to be ac­cepted on its own terms. The uni­verse is not wa­ver­ing be­tween us­ing par­ti­cles and waves, un­able to make up its mind. It’s only hu­man in­tu­itions about QM that swap back and forth. — Eliezer Yudkowsky

I am not go­ing to tell you that quan­tum me­chan­ics is weird, bizarre, con­fus­ing, or alien. QM is coun­ter­in­tu­itive, but that is a prob­lem with your in­tu­itions, not a prob­lem with quan­tum me­chan­ics. Quan­tum me­chan­ics has been around for billions of years be­fore the Sun co­a­lesced from in­ter­stel­lar hy­dro­gen. Quan­tum me­chan­ics was here be­fore you were, and if you have a prob­lem with that, you are the one who needs to change. QM sure won’t. There are no sur­pris­ing facts, only mod­els that are sur­prised by facts; and if a model is sur­prised by the facts, it is no credit to that model. — Eliezer Yudkowsky



(Evolu­tion) is a gen­eral pos­tu­late to which all the­o­ries, all hy­pothe­ses, all sys­tems must hence­for­ward bow and which they must satisfy in or­der to be think­able and true. Evolu­tion is a light which illu­mi­nates all facts, a tra­jec­tory which all lines of thought must fol­low — this is what evolu­tion is. — Pierre Teilhard de Chardin


There is no such thing as philos­o­phy-free sci­ence; there is only sci­ence whose philo­soph­i­cal bag­gage is taken on board with­out ex­am­i­na­tion. — Daniel Den­nett, Dar­win’s Danger­ous Idea, 1995.
Philos­o­phy is a bat­tle against the be­witch­ment of our in­tel­li­gence by means of lan­guage. — Wittgenstein


The Mind

Every­thing of beauty in the world has its ul­ti­mate ori­gins in the hu­man mind. Even a rain­bow isn’t beau­tiful in and of it­self. — Eliezer Yudkowsky


Levels of epistemic ac­cu­racy.



Gen­eral Education


Not es­sen­tial but a good pre­limi­nary to read­ing LessWrong and in some cases helpful to be able to make valuable con­tri­bu­tions in the com­ments. Many of the con­cepts in the fol­low­ing works are of­ten men­tioned on LessWrong or the sub­ject of fre­quent dis­cus­sions.


Elab­o­ra­tion of mis­cel­la­neous terms, con­cepts and fields of knowl­edge you might come across in some of the sub­se­quent and more tech­ni­cal ad­vanced posts and com­ments on LessWrong. The fol­low­ing con­cepts are fre­quently dis­cussed but not nec­es­sar­ily sup­ported by the LessWrong com­mu­nity. Those con­cepts that are con­tro­ver­sial are la­beled M.


Rele­vant web­sites. News and oth­er­wise. F

Fun & Fiction

The fol­low­ing are rele­vant works of fic­tion or playful treat­ments of fringe con­cepts. That means, do not take these works at face value.

Ac­com­pa­ny­ing text: The Log­i­cal Fal­lacy of Gen­er­al­iza­tion from Fic­tional Evidence

Memetic Hazard




A pop­u­lar board game played and analysed by many peo­ple in the LessWrong and gen­eral AI crowd.


This list is a work in progress. I will try to con­stantly up­date and re­fine it.

If you’ve any­thing to add or cor­rect (e.g. a bro­ken link), please com­ment be­low and I’ll up­date the list ac­cord­ingly.