Milan_Griffes

• Milan_Griffes 22 Sep 2020 23:50 UTC

  2 points

  in reply to: Benjamin_Todd’s comment on: How Depen­dent is the Effec­tive Altru­ism Move­ment on Dustin Moskovitz and Cari Tuna?

  $80m “other” per year seems very high to me, fwiw.

• Milan_Griffes 22 Sep 2020 21:11 UTC

  2 points

  on: How Depen­dent is the Effec­tive Altru­ism Move­ment on Dustin Moskovitz and Cari Tuna?

  See also: What’s the best struc­ture for op­ti­mal al­lo­ca­tion of EA cap­i­tal?

  So EA is cur­rently in a regime wherein the large ma­jor­ity of cap­i­tal flows from a sin­gle source, and cap­i­tal al­lo­ca­tion is set by a small num­ber of de­ci­sion-mak­ers.

  Rough es­ti­mate: if ~60% of Open Phil grant­mak­ing de­ci­sion­ing is at­tributable to Holden, then 47.2% of all EA cap­i­tal al­lo­ca­tion, or $157.4M, was de­cided by one in­di­vi­d­ual in 2017. 2018 & 2019 will prob­a­bly have similar pro­por­tions.

  It seems like EA en­tered into this regime largely due to his­tor­i­cally con­tin­gent rea­sons (Cari & Dustin de­vel­op­ing a close re­la­tion­ship with Holden, then out­sourc­ing a lot of their philan­thropic de­ci­sion-mak­ing to him & the Open Phil staff).

  It’s not clear that this struc­ture will lead to op­ti­mal cap­i­tal al­lo­ca­tion.

  ...

• Milan_Griffes 21 Sep 2020 20:47 UTC

  2 points

  on: microCOVID.org: A tool to es­ti­mate COVID risk from com­mon activities

  A good thread (a) sum­ma­riz­ing a pa­per on our cur­rent un­der­stand­ing of coro­n­avirus trans­mis­sion dy­nam­ics.

• Milan_Griffes 4 Sep 2020 19:19 UTC

  2 points

  in reply to: Milan_Griffes’s comment on: microCOVID.org: A tool to es­ti­mate COVID risk from com­mon activities

  Though per­haps the effect size they found is im­plau­si­bly large...

  Ex­pressed as rel­a­tive risk, vi­tamin D re­duced the risk of ICU ad­mis­sion 25-fold. Put an­other way, it elimi­nated 96% of the risk of ICU ad­mis­sion. Ex­pressed as an odds ra­tio, which is a less in­tu­itive con­cept but is of­ten used in statis­tics be­cause it gives an es­ti­mate of the effect of the treat­ment that would be con­stant across sce­nar­ios with differ­ent lev­els of risk, vi­tamin D re­duced the odds of ICU ad­mis­sion by 98%. Either way, vi­tamin D prac­ti­cally abol­ished the need for ICU ad­mis­sion.

  
  

  Would be great if this repli­cates in a big­ger study. In the mean­time, sup­ple­ment­ing Vi­tamin D is cheap & safe.

  More Vi­tamin D dis­cus­sion:

  • LessWrong post on Vi­tamin D sup­ple­men­ta­tion (pre-covid)

  • SSC post on the Vi­tamin D liter­a­ture (from 2014)

  • Gw­ern’s com­ment on that SSC post

• Milan_Griffes 4 Sep 2020 18:35 UTC

  2 points

  on: microCOVID.org: A tool to es­ti­mate COVID risk from com­mon activities

  In other covid news, we seem to be learn­ing that Vi­tamin D sup­ple­men­ta­tion is helpful.

  A small RCT was re­cently pub­lished: Castillo et al. 2020

  From Master­john’s com­men­tary (a):

  The trial was con­ducted at the Reina Sofía Univer­sity Hospi­tal in Córdoba, Spain. The trial in­cluded 76 pa­tients with COVID-19 pneu­mo­nia. Although this is no longer the stan­dard of care, all pa­tients were treated with hy­drox­y­chloro­quine and az­ithromycin and, when needed, a broad-spec­trum an­tibiotic. Ad­mis­sion to the ICU was de­ter­mined by a mul­ti­dis­ci­plinary com­mit­tee con­sist­ing of in­ten­sive care spe­cial­ists, pul­mo­nolo­gists, in­ter­nal medicine spe­cial­ists, and mem­bers of the ethics com­mit­tee.

  The pa­tients were ran­domly al­lo­cated to re­ceive or not re­ceive vi­tamin D in a 2:1 ra­tio. This re­sulted in 50 pa­tients in the vi­tamin D group and 26 pa­tients in the con­trol group.

  
  

  From the ab­stract:

  Of 50 pa­tients treated with cal­cife­diol [a form of Vi­tamin D], one re­quired ad­mis­sion to the ICU (2%), while of 26 un­treated pa­tients, 13 re­quired ad­mis­sion (50%) p-value X^2 Fischer test p < 0.001. Uni­vari­ate Risk Es­ti­mate Odds Ra­tio for ICU in pa­tients with Cal­cife­diol treat­ment ver­sus with­out Cal­cife­diol treat­ment: 0.02 (95%CI 0.002-0.17).

• Milan_Griffes 30 Aug 2020 15:57 UTC

  2 points

  in reply to: Linch’s comment on: microCOVID.org: A tool to es­ti­mate COVID risk from com­mon activities

  Thanks for run­ning the con­ver­sion into micro­morts – that’s helpful.

  fwiw back in 2003 the av­er­age US com­mute was ~30 miles/​day (I couldn’t find more re­cent data). So that’s about one micro­mort/​week from com­mut­ing.

  Here’s some case fatal­ity rate data by age. 0.5% chance of death seems a bit high, though maybe rea­son­able de­pend­ing on how you’re in­cor­po­rat­ing the ex­ter­nal­ity.

• Milan_Griffes 29 Aug 2020 23:08 UTC

  4 points

  on: microCOVID.org: A tool to es­ti­mate COVID risk from com­mon activities

  Thank you for cre­at­ing this!

  I was sur­prised by the risk­i­ness es­ti­mates in the table on this page: https://​​www.micro­covid.org/​​pa­per/​​2-riskiness

  If an event that ac­crues 1000 μCoV is con­sid­ered “bor­der­line reck­less”, wouldn’t that im­ply a very low risk tol­er­ance for ev­ery­day ac­tivi­ties like driv­ing a car? (Be­cause driv­ing is fairly high risk and some of that risk is ex­ter­nal­ized.)

• Milan_Griffes 29 Aug 2020 17:05 UTC

  21 points

  on: Some thoughts on the EA Mu­nich //​ Robin Han­son incident

  Some of Han­son’s writ­ing has prob­a­bly been, on net, detri­men­tal to his own in­fluence...

  https://​​xkcd.com/​​137

• Milan_Griffes 28 Aug 2020 0:03 UTC

  2 points

  in reply to: MathiasKirkBonde’s comment on: It’s Not Hard to Be Mo­rally Ex­cel­lent; You Just Choose Not To Be

  +1 to Parable of the Ta­lents be­ing ex­cel­lent, es­pe­cially given EA’s re­la­tion­ship to scrupu­los­ity.

• Milan_Griffes 19 Aug 2020 21:36 UTC

  8 points

  in reply to: Ozzie Gooen’s comment on: The case of the miss­ing cause pri­ori­ti­sa­tion research

  Thanks!

  100% agree that cul­tural health is very im­por­tant, and that EA is un­der-in­vest­ing in it. (The “we don’t want to just give money to our friends” point res­onates, and other scrupu­los­ity-re­lated stuff is prob­a­bly at play here as well.)

  
  

  In­di­vi­d­u­als at so­cial events will be try­ing to one-up each other with their clev­er­ness. I’m sure I’ve con­tributed to this. I’ve no­ticed my­self be­com­ing jeal­ous when I hear of oth­ers who are similar in some ways do­ing well, which re­ally should make no sense at all.

  Thank you for talk­ing about this!

  I’ve no­ticed similar pat­terns in my own mind, es­pe­cially around how I en­gage with this Fo­rum. (I’ve been step­ping back from it more this year be­cause I’ve no­ticed that a lot of my en­gage­ment wasn’t com­ing from a lov­ing place.)

  Th­ese dy­nam­ics may not make any sense, but there are deep biolog­i­cal & psy­cholog­i­cal forces giv­ing rise to them. [in­sert Robin Han­son’s “ev­ery­thing you do is sig­nal­ing” rant here]

  
  

  … I think in the anony­mous sur­veys 80K did a while back a bunch of peo­ple com­plained that there was a lot of sig­nal­ing go­ing on and that sta­tus was a big deal.

  Right. Last year con­cerns about sta­tus made a lot of heat on the Fo­rum (1, 2, 3), but as far as I know noth­ing has re­ally changed since then, per­haps other than more folks ac­knowl­edg­ing that sta­tus is a thing.

  (Sta­tus seems closely re­lated to scrupu­los­ity & to EA be­ing vet­ting-con­strained; I haven’t un­packed this yet.)

• Milan_Griffes 19 Aug 2020 20:48 UTC

  4 points

  in reply to: Ozzie Gooen’s comment on: The case of the miss­ing cause pri­ori­ti­sa­tion research

  I haven’t seen these spe­cific ex­am­ples, but there definitely seems to be a similar bias in other groups. Many or­ga­ni­za­tions are afraid to make any kinds of es­ti­mates at all...

  Re­minds me of the thing where cor­po­ra­tions don’t want to im­ple­ment in­ter­nal pre­dic­tion mar­kets be­cause im­ple­ment­ing a mar­ket isn’t in the self-in­ter­est of any in­di­vi­d­ual de­ci­sion-maker.

• Milan_Griffes 19 Aug 2020 20:44 UTC

  23 points

  in reply to: Ozzie Gooen’s comment on: The case of the miss­ing cause pri­ori­ti­sa­tion research

  Here’s a list I came up with from think­ing about this for ~30 min­utes:

  Bet­ter ways of mea­sur­ing what matters

  • Bet­ter neu­roimag­ing tech to parse out the neu­rolog­i­cal ba­sis of de­sir­able & un­de­sir­able sub­jec­tive states

  • Bet­ter mea­sures of sub­jec­tive well-being

  
  

  Help EAs see more clearly, un­pack + re­solve per­sonal trau­mas, and boost their effi­cacy + motivation

  • Emo­tional heal­ing as a pre­req­ui­site to rationality

  • CFAR, OAK, Lev­er­age, etc.

  • Plus build­ing meth­ods to au­dit which pro­jects are work­ing, which are failing, which are stagnating

  • Per­haps also a data col­lec­tion pro­ject that vac­u­ums up out­comes from the ob­ject-level pro­jects?

  
  

  Strengthen EA com­mu­nity ties /​ our sense of fellowship

  • More hon­esty about how weird effec­tive re­search meth­ods can be

  • More ac­knowl­edge­ment of the in­ter­de­pen­dent causal com­plex that gives rise to good re­search (e.g. Alex Flint’s in­tro­duc­tion here)

  • More Ben Fran­klin-es­que Juntos

  • Im­port more of Sili­con Valley’s “pay it for­ward” culture

  • Less rep­u­ta­tion man­age­ment /​ more psy­cholog­i­cal safety

  • Less sniping

  • OAK, Bay Area group houses, EA Hotel

  • Again, build­ing out (non-dom­i­nat­ing) ways to au­dit & col­lect data from the ob­ject-level projects

  
  

  Less scrupulosity

  • Ties into the above but de­serves its own bul­let given how our col­lec­tive psy­chol­ogy skews

  • Com­pas­sion­ate fight­ing against the thought-pat­tern Scott Alexan­der de­scribes here

  
  

  Make EA sexier

  • Mar­ket to re­tail donors /​ the broader pub­lic (e.g. Fu­ture Perfect, e.g. 80k, e.g. GiveWell run­ning ads on Vox pod­casts)

  • Mar­ket to im­pact in­vestors (e.g. Lion­heart) and big philanthropy

  • Cul­ti­vat­ing more “I want to be like that” energy

  • Seems easy to walk back if it isn’t work­ing be­cause so many in­ter­est groups are com­pet­ing for mind­share

  
  

  Sup­port EA phys­i­cal health

  • Prop­a­gate effec­tive treat­ments for RSI & back prob­lems, as above

  • Take the mind-body con­nec­tion seriously

  • Prop­a­gate best prac­tices for nu­tri­tion, sleep, ex­er­cise; make the case that at­tend­ing to these is pre­req­ui­site to hav­ing im­pact (rather than trad­ing off against hav­ing im­pact)

  
  

  Ad­vance our fron­tier of knowledge

  • e.g. GPI’s re­search agenda, e.g. the stuff Michael Dick­ens laid out in his comment

  • More work on how to solve co­or­di­na­tion problems

  • More work on gov­er­nance (e.g. Vi­talik’s stuff, e.g. the stuff Pal­la­dium is ex­plor­ing)

  
  

  Fund many moon­shots /​ spec­u­la­tive projects

  • Fund pro­jects that can be walked back if they aren’t work­ing out (which is most pro­jects, though some tech pro­jects may be hard-to-re­verse)

  • Worry less about brand management

• Milan_Griffes 19 Aug 2020 20:13 UTC

  5 points

  in reply to: Ozzie Gooen’s comment on: The case of the miss­ing cause pri­ori­ti­sa­tion research

  Your list re­minds me of this thread: What EA Fo­rum posts do you want some­one to write?

• Milan_Griffes 19 Aug 2020 18:08 UTC

  5 points

  in reply to: trammell’s comment on: The case of the miss­ing cause pri­ori­ti­sa­tion research

  At a glance, Sales­force’s AI Economist seems like an at­tempted im­ple­men­ta­tion of an IAM.

• Milan_Griffes 19 Aug 2020 17:10 UTC

  4 points

  in reply to: MichaelPlant’s comment on: Shifts in sub­jec­tive well-be­ing scales?

  Thanks for this!

  How do you think the po­ten­tial con­sis­tency over time (A) squares with the in­con­sis­tency be­tween scales & sub-scales that Kaj pointed out?

[Question] Shifts in sub­jec­tive well-be­ing scales?

• Milan_Griffes 23 Jun 2020 17:08 UTC

  4 points

  in reply to: MichaelPlant’s comment on: HLI’s Men­tal Health Pro­gramme Eval­u­a­tion Pro­ject—Up­date on the First Round of Evaluation

  Note they are mostly to do with in­surance is­sues.

  fwiw I don’t think most of this prob­lem is due to in­surance is­sues, though I agree that the US health­care sys­tem is very weird and falls short in a lot of ways.

  
  

  This also isn’t spe­cific to men­tal health: one might re­tort to donors to AMF that they should be fund­ing im­prove­ments in (say) health treat­ment in gen­eral or malaria treat­ment in par­tic­u­lar.

  I don’t think this anal­ogy holds up: we’ve erad­i­cated malaria in many de­vel­oped coun­tries, but we haven’t figured out men­tal health to the same de­gree (e.g. 1 in 5 Amer­i­cans have a men­tal ill­ness).

  
  

  I sus­pect that if there were a re­ally strong ‘pull’ for goods/​ser­vices to be pro­vided, then we would already have ‘solved’ world poverty, which makes me think dis­tri­bu­tion is weakly re­lated to in­no­va­tion.

  World poverty has been de­creas­ing a lot since 1990 – some good charts here & here.

  M-Pesa and the broad pen­e­tra­tion of smart­phones are ex­am­ples of in­no­va­tions that were quickly dis­tributed. The path from in­no­va­tion to dis­tri­bu­tion is prob­a­bly harder for ser­vices.

• Milan_Griffes 15 Jun 2020 19:17 UTC

  3 points

  in reply to: Aaron Gertler’s comment on: [Link] “Will He Go?” book re­view (Scott Aaron­son)

  I usu­ally do link posts to im­prove the com­mu­nity’s situ­a­tional aware­ness.

  This is up­stream of ad­vo­cat­ing for spe­cific ac­tions, though it’s definitely part of that causal chain.

• Milan_Griffes 15 Jun 2020 17:03 UTC

  4 points

  in reply to: MichaelPlant’s comment on: HLI’s Men­tal Health Pro­gramme Eval­u­a­tion Pro­ject—Up­date on the First Round of Evaluation

  I’m not sure what you mean by go­ing from 0 to 1 vs 1 to n. Can you elab­o­rate?

  The link in my top-level com­ment elab­o­rates the con­cept.

  
  

  how much bet­ter MH treat­ment could be than the cur­rent best practice

  Quick re­ply: prob­a­bly a lot bet­ter. See ec­static med­i­ta­tive states, con­firmed by fMRI & EEG.

  See also Slate Star Codex on the weird­ness of Western men­tal health­care: 1, 2, 3, 4, 5

  
  

  how easy it would be to get it there

  Quick re­ply: not sure about how easy it would be to achieve the pla­tonic ideal of men­tal health­care – QRI is prob­a­bly more opinionated about this.

  Given how much of an im­prove­ment SSRIs and CBT were over the pre­ex­ist­ing stan­dard-of-care, and how much of an im­prove­ment psychedelic, ke­tamine, and so­matic ther­a­pies seem to be over the cur­rent stan­dard-of-care, I’d guess that we’re nowhere close to hit­ting diminish­ing marginal re­turns.

  
  

  how fast this would spread

  Quick re­ply: if global­iza­tion con­tinues, the best prac­tices of the global­ized so­ciety will prop­a­gate “nat­u­rally” (i.e. as a re­sult of the in­cen­tives stake­hold­ers face). From this per­spec­tive, we’re more limited by get­ting the global­ized best prac­tices right than we are by dis­tribut­ing our cur­rent best prac­tices.

• Milan_Griffes 15 Jun 2020 16:46 UTC

  2 points

  in reply to: Aaron Gertler’s comment on: [Link] “Will He Go?” book re­view (Scott Aaron­son)

  From the part I ex­cerpted:

  “You should read it right now (or at least read this Vox in­ter­view), if you want to think through the con­tours of a civ­i­liza­tional Sin­gu­lar­ity that seems at least as plau­si­ble to me as the AI Sin­gu­lar­ity, but whose fixed date of Novem­ber 3, 2020 we’re now hurtling to­ward.”

  
  

  The EA im­pli­ca­tions of the 2020 US pres­i­den­tial elec­tion seem ob­vi­ous?

  See also Dustin & Cari’s $20m grant to the 2016 Clin­ton cam­paign.