Milan_Griffes

Giv­ing more won’t make you hap­pier

If slow-take­off AGI is some­what likely, don’t give now

What con­se­quences?

• Milan_Griffes 4 Aug 2018 1:10 UTC

  20 points

  in reply to: TaraMacAulay's comment on: Lev­er­age Re­search: re­view­ing the ba­sic facts

  I don’t think that Lev­er­age, Paradigm or re­lated pro­jects are good use of EA time or money

  Found this sur­pris­ing given the pos­i­tive valence of the rest of the com­ment. Could you ex­pand a lit­tle on why you don’t think Lev­er­age et al. are a good use of time/​money?

Do­ing good while clueless

What we talk about when we talk about life satisfaction

• Milan_Griffes 4 Aug 2018 17:01 UTC

  15 points

  in reply to: Evan_Gaensbauer's comment on: Lev­er­age Re­search: re­view­ing the ba­sic facts

  when I’m mak­ing pub­lic com­ments with­out a time crunch

  My hunch is even when there’s a time crunch, fewer words will be big­ger bang for buck :-)

• Milan_Griffes 8 Dec 2018 16:10 UTC

  14 points

  on: Open Thread #43

  I’ve been toy­ing around with the fol­low­ing:

  • There are two mo­ti­va­tions for donat­ing money – ego­tis­tic (e.g. it feels good to do) & al­tru­is­tic (e.g. other peo­ple are bet­ter off)

  • The ego­tis­tic mo­ti­va­tion is highly scope in­sen­si­tive – giv­ing away $500 feels roughly as good as giv­ing away $50,000

    • Prob­a­bly also scope in­sen­si­tive qual­i­ta­tively – giv­ing $5,000 to a low-im­pact char­ity feels about as good as giv­ing $5,000 to an effec­tive char­ity (es­pe­cially if you don’t re­flect very much about im­pact)

    • This scope in­sen­si­tivity is baked in – know­ing about it doesn’t make it go away

  • EA orgs some­times say that giv­ing effec­tively will make you hap­pier (e.g. 80k, e.g. GWWC)

    • Th­ese ar­gu­ments ig­nore the scope in­sen­si­tivity of the ego­tis­tic mo­ti­va­tion – donat­ing some money to char­ity will prob­a­bly make you hap­pier than not donat­ing any at all. It’s less clear that donat­ing more money to char­ity will make you hap­pier than donat­ing some (and es­pe­cially un­clear that the dona­tion <> hap­piness link scales any­thing close to lin­early)

  • Ergo, EA should stop recom­mend­ing effec­tive giv­ing on ego­tis­tic grounds, and prob­a­bly even en­courage peo­ple to not do effec­tive giv­ing if they’re con­sid­er­ing it be­cause they want to be hap­pier (re­lated)

    • If the above is true, effec­tive giv­ing won’t make you much hap­pier than low-im­pact giv­ing, and donat­ing large amounts won’t make you much hap­pier than donat­ing small amounts

    • e.g. $100 to GiveDirectly feels about as good as $1,000 to GiveDirectly; e.g. sav­ing one (statis­ti­cal) life via AMF feels about as good as sav­ing two (statis­ti­cal) lives via AMF

    • Ad­vo­cat­ing for effec­tive giv­ing on ego­tis­tic grounds (e.g. “it will make you hap­pier”) is sorta a false promise

“Just take the ex­pected value” – a pos­si­ble re­ply to con­cerns about cluelessness

• Milan_Griffes 29 Oct 2018 18:40 UTC

  12 points

  in reply to: Halstead's comment on: An­nounc­ing new EA Funds man­age­ment teams

  I’m also cu­ri­ous for more info about how teams were se­lected.

Is in­tel­lec­tual work bet­ter con­strued as ex­plo­ra­tion or perfor­mance?

• Milan_Griffes 10 Jul 2018 15:15 UTC

  11 points

  in reply to: jpaddison's comment on: Open Thread #40

  Why I’m skep­ti­cal of cost-effec­tive­ness analysis

  Re­post­ing as com­ment be­cause mods told me this wasn’t thor­ough enough to be a post.

  Briefly:

  • The en­tire course of the fu­ture mat­ters (more)

  • Pre­sent-day in­ter­ven­tions will bear on the en­tire course of the fu­ture, out to the far future

  • The effects of pre­sent-day in­ter­ven­tions on far-fu­ture out­comes are very hard to predict

  • Any model of an in­ter­ven­tion’s effec­tive­ness that doesn’t in­clude far-fu­ture effects isn’t tak­ing into ac­count the bulk of the effects of the intervention

  • Any model that in­cludes far-fu­ture effects isn’t be­liev­able be­cause these effects are very difficult to pre­dict accurately

How tractable is clue­less­ness?

• Milan_Griffes 6 Feb 2019 18:37 UTC

  10 points

  on: My pos­i­tive ex­pe­rience tak­ing the an­tide­pres­sant Wel­lbutrin /​ Bupro­pion, & why maybe you should try it too

  Was chat­ting with Gw­ern about this. An ex­cerpt of their thoughts (pub­lished with per­mis­sion):

  Wel­lbutrin/​Bupro­prion is a weird one. Comes up of­ten on SSC and el­se­where as sur­pris­ingly effec­tive and side-effect free, but also with a very wide var­i­ance and messy mechanism (even for an­tide­pres­sants) so anec­dotes differ dra­mat­i­cally.

  With an­tide­pres­sants, you’re usu­ally talk­ing multi-week on­set and washout pe­ri­ods, so blinded self-ex­per­i­ments would take some­thing like half a year for de­cent sam­ple sizes. It’s not that easy to get if you aren’t will­ing to go through a doc­tor (ie no easy or­der­ing on­line from a DNM or clear­net site, like modafinil)…

  Fi­nally, as far as I can tell, my per­sonal prob­lems have more to do with anx­iety than de­pres­sion, and anti-anx­iety is not what bupro­prion is gen­er­ally de­scribed as best for, so my own benefit is prob­a­bly less than usual. I thought about it a lit­tle but de­cided it was too weird and hard to get, and self-ex­per­i­ments would take too long.

• Milan_Griffes 13 Feb 2019 17:45 UTC

  9 points

  on: A sys­tem for scor­ing poli­ti­cal can­di­dates. RFC (re­quest for com­ments) on method­ol­ogy and positions

  Com­ments on any is­sue are gen­er­ally wel­come but nat­u­rally you should try to fo­cus on ma­jor is­sues rather than minor ones. If you post a long line of ar­gu­ments about ed­u­ca­tion policy for in­stance, I might not get around to read­ing and fact-check­ing the whole thing, be­cause the model only gives a very small weight to ed­u­ca­tion policy right now (0.01) so it won’t make a big differ­ence ei­ther way. But if you say some­thing about im­mi­gra­tion, no mat­ter how nu­anced, I will pay close at­ten­tion be­cause it has a very high weight right now (2).

  I think this begs the ques­tion.

  If mod­eler at­ten­tion is dis­tributed pro­por­tional to the model’s cur­rent weight­ing (such that dis­cus­sion of high-weighted is­sues re­ceive more at­ten­tion than dis­cus­sion of low-weighted is­sues), it’ll be hard to iden­tify mis­takes in the cur­rent weight­ing.

• Milan_Griffes 4 Feb 2019 22:35 UTC

  9 points

  on: EA Ho­tel Fundraiser 2: cur­rent guests and their projects

  Great to see so many folks work­ing at cool stuff at the EA Ho­tel!

  Thank you for tak­ing the time to write this up, and for ev­ery­thing else you’ve done to make this hap­pen.

• Milan_Griffes 16 Dec 2018 16:37 UTC

  9 points

  on: New edi­tion of “Ra­tion­al­ity: From AI to Zom­bies”

  I think pa­per books are an un­der-ap­pre­ci­ated tech­nol­ogy these days; very ex­cited that pa­per ver­sions of Ra­tion­al­ity are com­ing out!

  Thanks for mak­ing this hap­pen!

• Milan_Griffes 31 Jan 2019 17:17 UTC

  8 points

  on: Lat­est Re­search and Up­dates for Jan­uary 2019

  Emer­gent Ven­tures is look­ing to fund pro­jects on “ad­vanc­ing hu­mane solu­tions to those fac­ing ad­ver­sity – based on tol­er­ance, uni­ver­sal­ity, and co­op­er­a­tive pro­cesses”

  Recom­mend shoot­ing Tyler Cowen an email check­ing about the chances for your idea be­fore sub­mit­ting an app. He’s pretty re­spon­sive to email & there are ar­eas Emer­gent Ven­tures al­most cer­tainly won’t fund, so check­ing first can save a bunch of time.

• Milan_Griffes 24 Nov 2018 0:11 UTC

  8 points

  in reply to: Benjamin_Todd's comment on: Towards Bet­ter EA Ca­reer Advice

  Th­ese two di­rec­tions put us in a difficult po­si­tion. Given our limited re­sources, if we go nar­rower, then we’ll make our site worse for the broader au­di­ence, and vice versa.

  Has 80k con­sid­ered spin­ning off a sister org that fo­cuses on the broader au­di­ence?

  Seems like serv­ing both the nar­row-ca­reer-ad­vice & broad-ca­reer-ad­vice mar­kets are im­por­tant EA pro­jects.

  80k could be com­par­a­tively well-po­si­tioned to ad­dress both, given its track record & fun­der base.

• Milan_Griffes 4 Aug 2018 18:23 UTC

  8 points

  in reply to: RandomEA's comment on: Prob­lems with EA rep­re­sen­ta­tive­ness and how to solve it

  This is great – con­sider mak­ing it a stan­dalone post?