Rational Feed: Last Week’s Community Articles and Some Recommended Posts

===Highly Recom­mended Ar­ti­cles:

Why I Am Not A Quaker Even Though It Often Seems As Though I Should Be by Ben Hoff­man—Quak­ers have con­sis­tently got­ten to the right an­swers faster than most peo­ple, or the au­thor. Ar­bi­trage strate­gies to beat the quak­ers. An in­com­plete sur­vey of al­ter­na­tives.

Could A Neu­ro­scien­tist Un­der­stand A Micro­pro­ces­sor by Ra­tion­ally Speak­ing—“Eric Jonas, dis­cussing his provoca­tive pa­per ti­tled ‘Could a Neu­ro­scien­tist Un­der­stand a Micro­pro­ces­sor?’ in which he ap­plied state-of-the-art neu­ro­science tools, like le­sion anal­y­sis, to a com­puter chip. By ap­ply­ing neu­ro­science’s tools to a sys­tem that hu­mans fully un­der­stand he was able to re­veal how sur­pris­ingly un­in­for­ma­tive those tools ac­tu­ally are.”

Rea­son­able Doubt New Look Whether Pri­son Growth Cuts Crime by Open Philos­o­phy—Part1 of a four part, in depth, se­ries on Crim­i­nal Jus­tice re­form. The re­main­ing posts are linked be­low. “I es­ti­mate, that at typ­i­cal policy mar­gins in the United States to­day, de­carcer­a­tion has zero net im­pact on crime. That es­ti­mate is un­cer­tain, but at least as much ev­i­dence sug­gests that de­carcer­a­tion re­duces crime as in­creases it. The crux of the mat­ter is that tougher sen­tences hardly de­ter crime, and that while im­pris­on­ing peo­ple tem­porar­ily stops them from com­mit­ting crime out­side prison walls, it also tends to in­crease their crim­i­nal­ity af­ter re­lease. As a re­sult, “tough-on-crime” ini­ti­a­tives can re­duce crime in the short run but cause offset­ting harm in the long run. Em­piri­cal so­cial sci­ence re­search—or at least non-ex­per­i­men­tal so­cial sci­ence re­search—should not be taken at face value. Among three dozen stud­ies I re­viewed, I ob­tained or re­con­structed the data and code for eight. Repli­ca­tion and re­anal­y­sis re­vealed sig­nifi­cant method­olog­i­cal con­cerns in seven and led to ma­jor rein­ter­pre­ta­tions of four. Th­ese stud­ies en­dured much tougher scrutiny from me than they did from peer re­view­ers in or­der to make it into aca­demic jour­nals. Yet given the stakes in lives and dol­lars, the added scrutiny was worth it. So from the point of view of de­ci­sion mak­ers who rely on aca­demic re­search, to­day’s peer re­view pro­cesses fall well short of the op­ti­mal.”

Deter­rence De Min­i­mus by Open Philos­o­phy—Part 2.

In­ca­pac­i­ta­tion How Much Does Put­ting Peo­ple In­side Pri­son Cut Crime Out­side by Open Philos­o­phy—Part 3.

Aftere­ffects Us Ev­i­dence Says Do­ing More Time Typ­i­cally Leads More Crime After by Open Philos­o­phy—Part 4.

===Scott:

L Dopen Thread by Scott Alexan­der—Bi-weekly pub­lic open thread. Berkeley SSC meetup. New ad for the Green­field Guild, an on­line net­work of soft­ware con­sul­tants. Rea­sons to re­spect the so­ciety of friends.

Med­i­ta­tive States As Men­tal Feed­back Loops by Scott Alexan­der—the main rea­son we don’t see emo­tional pos­i­tive feed­back loops is that peo­ple get dis­tracted. If you do not get dis­tracted you can ex­pe­rience a bliss feed­back look.

Book Re­view Mas­ter­ing The Core Teach­ings Of The Bud­dha by Scott Alexan­der—“Bud­dhism For ER Docs. ER docs are fa­mous for be­ing prac­ti­cal, work­ing fast, and think­ing ev­ery­one else is an idiot. MCTB de­liv­ers on all three counts.” Prac­ti­cal bud­dhism with a fo­cus on get­ting things done. bud­dhism is split into moral­ity con­cen­tra­tion and wis­dom. Dis­cus­sion of “the Dark Night of the Soul” which is a sort of de­pres­sion oc­curs when you have had some but not enough spiritual ex­pe­rience.

===Ra­tion­al­ist:

Im­pres­sion Track Records by Katja Grace—Three rea­sons its bet­ter to keep im­pres­sion track records and be­lief track records sep­a­rate.

Why I Am Not A Quaker Even Though It Often Seems As Though I Should Be by Ben Hoff­man—Quak­ers have con­sis­tently got­ten to the right an­swers faster than most peo­ple, or the au­thor. Ar­bi­trage strate­gies to beat the quak­ers. An in­com­plete sur­vey of al­ter­na­tives.

The Best Self Help Should Be Self Defeat­ing by mindlevelup—“Self-help is sup­posed to get peo­ple to stop need­ing it. But typ­i­cal in­cen­tives in any medium mean that it’s pos­si­ble to get peo­ple hooked on your con­tent in­stead. A mus­ing on how the setup for writ­ing self-help differs from typ­i­cal con­tent.”

No­body Does The Thing That They Are Sup­pos­edly Do­ing by Kaj So­tala—“In gen­eral, nei­ther or­ga­ni­za­tions nor in­di­vi­d­ual peo­ple do the thing that their sup­posed role says they should do.” Evolu­tion­ary in­cen­tives. Psy­chol­ogy of mo­ti­va­tion. Very large num­ber of links.

Out To Get You by Zvi Moshow­itz—“Some things are fun­da­men­tally Out to Get You. They seek re­sources at your ex­pense. Fees are hid­den. Ex­tra op­tions are foisted upon you.” You have four re­sponses: Get Gone, Get Out (give up), Get Com­pact (limit what it wants) or Get Ready for Bat­tle.

In Defense Of Un­re­li­a­bil­ity by Ozy—Zvi claims that when he makes plan with friends in the bay he never as­sumes the plan will ac­tu­ally oc­cur. Ozy de­pends on un­re­li­able trans­port. Get­ting places 10-15 early is also costly. Flak­ing and ago­ra­pho­bia.

Strate­gic Goal Pur­suit And Daily Sched­ules by Rossin (less­wrong) - The au­thor benefit­ted from Anna Sala­mon’s goal-pur­su­ing heuris­tics and daily sched­ules.

Why At­ti­tudes Mat­ter by Ozy—Fo­cus­ing on at­ti­tudes can be bad for some peo­ple. Two ar­gu­ments: “First, for any re­motely com­pli­cated situ­a­tion, it would be im­pos­si­ble to com­pletely list out all the things which are okay or not okay. Se­cond, an at­ti­tude em­pha­sis pre­vents rules-lawyer­ing.”

Hu­mans Cells In Mul­ti­cel­lu­lar Fu­ture Minds by Robin Han­son—In gen­eral hu­mans re­place spe­cific sys­tems with more gen­eral adap­tive sys­tems. See­ing like a State. Most biolog­i­cal and cul­tural sys­tems are not gen­eral. Multi-cel­lu­lar or­ganisms re tremen­dously in­effi­cient. The power of en­trenched sys­tems. Hu­man brains are ex­tremely gen­eral. Hu­man brains may win for a long time vs other forms of in­tel­li­gence.

Rec­og­niz­ing Vs Gen­er­at­ing An Im­por­tant Di­chotomy For Life by Gor­don (Map and Ter­ri­tory) - Bul­let Points → Es­say vs Es­say → Bul­let Points. Gen­er­at­ing ideas vs cri­tique. Most ad­vice is bad since it doesn’t con­vey the rea­sons clearly. Let the other per­son figure out the ac­tual ad­vice for them­selves.

Pre­dic­tion Mar­kets Up­date by Robin Han­son—Pre­dic­tion mar­kets provide pow­er­ful in­for­ma­tion but they challenge pow­er­ful en­trenched in­ter­ests, Han­son com­pares them to “a knowl­edge­able Autist in the C-suite”. Com­pa­nies sel­l­ing straight pre­dic­tion mar­ket tech mostly went un­der. Blockchain plat­forms for pre­dic­tion mar­kets. Some dis­cus­sion of cur­rently promis­ing com­pa­nies.

===AI:

Fo­cus Areas Of Worst Case Ai Safety by The Foun­da­tional Re­search In­sti­tute—Re­dun­dant safety mea­sures. Trip­wires. Ad­ver­sar­ial ar­chi­tec­tures. De­tect­ing and for­mal­iz­ing suffer­ing. Backup util­ity func­tions. Benign test­ing en­vi­ron­ments.

Srisk Faq by To­bias Bau­mann (EA fo­rum) - Quite de­tailed re­sponses to ques­tions about suffer­ing risks and their con­nec­tion to AGI. sec­tions: Gen­eral ques­tions, The fu­ture, S-risks and x-risks, Mis­cel­la­neous.

===EA:

Rea­son­able Doubt New Look Whether Pri­son Growth Cuts Crime by Open Philos­o­phy—Part1 of a four part, in depth, se­ries on Crim­i­nal Jus­tice re­form. The re­main­ing posts are linked be­low. “I es­ti­mate, that at typ­i­cal policy mar­gins in the United States to­day, de­carcer­a­tion has zero net im­pact on crime. That es­ti­mate is un­cer­tain, but at least as much ev­i­dence sug­gests that de­carcer­a­tion re­duces crime as in­creases it. The crux of the mat­ter is that tougher sen­tences hardly de­ter crime, and that while im­pris­on­ing peo­ple tem­porar­ily stops them from com­mit­ting crime out­side prison walls, it also tends to in­crease their crim­i­nal­ity af­ter re­lease. As a re­sult, “tough-on-crime” ini­ti­a­tives can re­duce crime in the short run but cause offset­ting harm in the long run. Em­piri­cal so­cial sci­ence re­search—or at least non-ex­per­i­men­tal so­cial sci­ence re­search—should not be taken at face value. Among three dozen stud­ies I re­viewed, I ob­tained or re­con­structed the data and code for eight. Repli­ca­tion and re­anal­y­sis re­vealed sig­nifi­cant method­olog­i­cal con­cerns in seven and led to ma­jor rein­ter­pre­ta­tions of four. Th­ese stud­ies en­dured much tougher scrutiny from me than they did from peer re­view­ers in or­der to make it into aca­demic jour­nals. Yet given the stakes in lives and dol­lars, the added scrutiny was worth it. So from the point of view of de­ci­sion mak­ers who rely on aca­demic re­search, to­day’s peer re­view pro­cesses fall well short of the op­ti­mal.”

Deter­rence De Min­i­mus by Open Philos­o­phy—Part 2.

In­ca­pac­i­ta­tion How Much Does Put­ting Peo­ple In­side Pri­son Cut Crime Out­side by Open Philos­o­phy—Part 3.

Aftere­ffects Us Ev­i­dence Says Do­ing More Time Typ­i­cally Leads More Crime After by Open Philos­o­phy—Part 4.

Pay­pal Giv­ing Fund by Jeff Kauf­man—The PayPal giv­ing fund lets you batch dona­tions and PayPal cov­ers the fees if you use it. Jeff thought there must be a catch but it seems le­git.

What Do Dalys Cap­ture by Danae Ar­royos (EA fo­rum) - How Dis­abil­ity Ad­justed life years com­puted. DALYs mis­rep­re­sent men­tal health. DALY’s Miss Indi­rect Effects. Other is­sues.

Against Ea Pr by Ozy—The EA com­mu­nity is the only large en­tity try­ing to pro­duce ac­cu­rate and pub­li­cly available as­sess­ments of char­i­ties. Hence the EA com­mu­nity should not trade away any hon­esty. EAs should sim­ply say which causes and or­ga­ni­za­tions are most effec­tive, they should not worry about PR con­cerns.

Ea Sur­vey 2017 Series Qual­i­ta­tive Com­ments Sum­mary by tee (EA fo­rum) - Are you an EA, how wel­com­ing is EA, lo­cal EA meetup at­ten­dance, con­cerns with not be­ing ‘EA enough’, im­prov­ing the sur­vey.

De­mo­graph­ics Ii by tee (EA fo­rum) - Ra­cial break­down. Per­cent white in var­i­ous ge­o­graphic lo­ca­tions. Poli­ti­cal spec­trum. Poli­tics cor­re­lated with cause area, diet and ge­og­ra­phy, em­ploy­ment, fields of study, year join­ing EA.

===Poli­tics and Eco­nomics:

Raj Chetty Course Us­ing Big Data Solve Eco­nomic So­cial Prob­lems by Marginal Revolu­tion—Link to an eleven lec­ture course. “Equal­ity of op­por­tu­nity, ed­u­ca­tion, health, the en­vi­ron­ment, and crim­i­nal jus­tice. In the con­text of these top­ics, the course pro­vides an in­tro­duc­tion to ba­sic statis­ti­cal meth­ods and data anal­y­sis tech­niques, in­clud­ing re­gres­sion anal­y­sis, causal in­fer­ence, quasi-ex­per­i­men­tal meth­ods, and ma­chine learn­ing.”

Speech On Cam­pus Re­ply To Brad Delong by Noah Smith—The safe­guard put in place to ex­clude the small minor­ity of gen­uinely toxic peo­ple will be overused. Com­par­i­son to the war on ter­ror. Brad’s ex­clu­sions crite­ria are in­cred­ibly vague. The speech re­stric­tion ap­para­tus is patch­work and in­con­sis­tent. Cul­tural Revolu­tion.

Deon­tol­o­gist Envy by Ozy—The be­hav­ior of your group is highly un­likely to effect the be­hav­ior of your poli­ti­cal op­po­nents. Many peo­ple re­spond to pro­posed tac­tics by ask­ing “What if ev­ery­one did that”. Ozy claims these re­sponses show an im­plicit Kan­tian or de­on­tolog­i­cal point of view.

Peak Fos­sil Fuel by Bayesian In­vestor—Elec­tric cars will have a 99% mar­ket share by 2035. “Elec­tric robo­cars run by Uber-like com­pa­nies will be cheap enough that you’ll have trou­ble giv­ing away a car bought to­day. Uber’s prices will be less than your ob­so­lete car’s costs of fuel, main­te­nance, and in­surance.”

What We Didn’t Get by Noah Smith—We are cur­rently liv­ing in a world en­vi­sioned by the cy­ber­punk writ­ers. the early in­dus­trial sci-fi writ­ers also pre­dicted many in­ven­tions. Why didn’t mid 1900s sci-fi come true? We ran out of the­o­ret­i­cal physics and we ran out of en­ergy. En­ergy den­sity of fuel sources. Some ex­ist­ing or plau­si­ble tech­nol­ogy is just too dan­ger­ous. Dis­cus­sion of whether strong AI, per­sonal up­load, nan­otech and/​or the sin­gu­lar­ity will come true.

Un­pop­u­lar Ideas About Chil­dren by Ju­lia Galef—Ju­lia’s thoughts on why she is col­lect­ing these lists. Par­ent­ing styles, pro and anti-na­tal­ism, sex­u­al­ity, pun­ish­ment, etc. Hap­piness stud­ies. Some other stud­ies find­ing ex­treme re­sults.

The Mar­gin Of Stupid by Noah Smith—Can we trust stud­ies show­ing that mil­len­ni­als are as racist as their par­ents, ex­cept for the ones in col­lege who are ex­treme leftists?

Role of Allies in Queer Spaces by Brute Rea­son—The main pur­pose of hav­ing al­lies in LBGTQA spaces is pro­vid­ing cover for closeted or ques­tion­ing mem­bers. Gen­uinely cis-straight al­lies are ok in some spaces like LBGTQA bands. But straight al­lies cause prob­lems when they are pre­sent in queer sup­port spaces.

The Won­der Of In­ter­na­tional Adop­tion by Bryan Ca­plan—Benefits of in­ter­na­tional adop­tion of third world chil­dren. Adoptees are ex­tremely stunted phys­i­cally on ar­rival but make up some of the differ­ence post adop­tion. In­ter­na­tional adop­tions raises IQ by at least 4 points on av­er­age and per­haps as much as 8.

===Misc:

Coin Flip­ping Prob­lem by pro­tokol2020 - Flip­ping coins un­til you get a pre-com­mit­ted se­quence. You re-start when­ever your flip doesn’t match the se­quence. Re­la­tion­ship be­tween the ex­pected num­ber of flips and the length of the se­quence.

Seek Not To Be En­ter­tained by Mr. Money Mus­tache—Don’t be nor­mal, nor­mal peo­ple need con­stant en­ter­tain­ment. You can get en­joy­ment and satis­fac­tion from mak­ing things. Ad­vice for peo­ple less ab­nor­mal than MMM. What you en­joy doesn’t mat­ter, what mat­ters is what is good for you.

Propo­si­tions On Im­mor­tal­ity by sam[]zdat—Fic­tion. A man di­gresses about philos­o­phy, the na­ture of time, the soul, con­scious­ness and mor­tal­ity.

Com­ments For Ghost by Tom Bartleby—Ghost is a blog plat­form hat doesn’t na­tively sup­port com­ments. Three im­por­tant use cases and why they all benefit from com­ments: Ex-Word­press blog­ger who wants things to ‘just work’, Power suers care about pri­vacy and don’t want to use third party com­ments, The Static-Site Fence-Sit­ter since the main dy­namic con­tent you want is com­ments.

Prime Cross­word by pro­tokol2020 - Can you cre­ate a grid larger than [3,7],[1,1] where all the rows and columns are primes? (37, 11, 31 and 71 are prime).

===Pod­cast:

Reihan Salam by The Ezra Klein Show—Re­mak­ing the Repub­li­can party, but not the way Don­ald Trump did it. “The fu­ture of the Repub­li­can Party, the health­care de­bate, and how he would re­form our im­mi­gra­tion sys­tem (and up­end the whole way we talk about it). ”

Into The Dark Land by Wak­ing Up with Sam Har­ris—“Sid­dhartha Mukher­jee about his Pulitzer Prize win­ning book, The Em­peror of All Mal­adies: A Biog­ra­phy of Cancer.”

Con­ver­sa­tion with Larry Sum­mers by Marginal Revolu­tion—“Men­tor­ing, in­no­va­tion in higher ed­u­ca­tion, monopoly in the Amer­i­can econ­omy, the op­ti­mal rate of cap­i­tal in­come tax­a­tion, philan­thropy, Her­mann Melville, the benefits of la­bor unions, Mex­ico, Rus­sia, and China, Fed un­der­shoot­ing on the in­fla­tion tar­get, and Larry’s table ten­nis ad­ven­ture in the sum­mer Jewish Olympics.”

Hilary Clin­ton by The Ezra Klein Show—Hilary’s dream of pay­ing for ba­sic in­come with rev­enue from shared na­tional re­sources. Why she scrapped the plan. Hilary thinks she should per­haps have thrown cau­tion to the wind. Hilary isn’t a rad­i­cal, she is proud of the Amer­i­can poli­ti­cal sys­tem and is an­noyed other’s don’t share her en­thu­si­asm for in­cre­men­tal progress.

David Rem­nick by The Ezra Klein Show—New Yorker ed­i­tor. “Rus­sia’s med­dling in the US elec­tion, Rus­sia’s trans­for­ma­tion from com­mu­nist rule to Boris Yeltsin and Vladimir Putin, his mag­a­z­ine’s cov­er­age of Pres­i­dent Don­ald Trump, how he chooses his re­porters and ed­i­tors, and how to build a real busi­ness around great jour­nal­ism.”

Gabriel Zuc­man by EconTalk—“Re­search on in­equal­ity and the dis­tri­bu­tion of in­come in the United States over the last 35 years. Zuc­man finds that there has been no change in in­come for the bot­tom half of the in­come dis­tri­bu­tion over this time pe­riod with large gains go­ing to the top 1%. The con­ver­sa­tion ex­plores the ro­bust­ness of this re­sult to var­i­ous as­sump­tions and pos­si­ble ex­pla­na­tions for the find­ings.”

Could A Neu­ro­scien­tist Un­der­stand A Micro­pro­ces­sor by Ra­tion­ally Speak­ing—“Eric Jonas, dis­cussing his provoca­tive pa­per ti­tled ‘Could a Neu­ro­scien­tist Un­der­stand a Micro­pro­ces­sor?’ in which he ap­plied state-of-the-art neu­ro­science tools, like le­sion anal­y­sis, to a com­puter chip. By ap­ply­ing neu­ro­science’s tools to a sys­tem that hu­mans fully un­der­stand he was able to re­veal how sur­pris­ingly un­in­for­ma­tive those tools ac­tu­ally are.”