Strong up­voted this one.

A promi­nent Bud­dhist monk in the Thai For­est Tra­di­tion (Ajahn Jayasaro) said the fol­low­ing, which I feel is highly rele­vant here:

Some­one had asked (llama) Ko­hima, “What do you think of ex­pand­ing minds through chem­i­cal means?” He said, if you have an ig­no­rant mind then you just get ex­panded ig­no­rance. I thought he was just on the spot. It is all within the sphere of dark­ness, isn’t it? You are still play­ing around with differ­ent modes of ig­no­rance. You are not ac­tu­ally go­ing be­yond. You are not tran­scend­ing. You are tran­scend­ing one par­tic­u­lar state of ig­no­rance, but you are still in the same build­ing, you haven’t got out of the build­ing, you still haven’t got out of prison. So this so­bri­ety is that whole turn­ing away from all the strange and un­usual ex­pe­riences and vi­sions, phys­iolog­i­cal, men­tal states that are available through chem­i­cal means and tak­ing a delight in the sim­ple down to earth clar­ity of aware­ness.”

effect from boost­ing effi­cacy of cur­rent long-ter­mist la­bor + effect from in­creas­ing the amount of long-ter­mist labor

Let’s go. Up­side 1:

effect from boost­ing effi­cacy of cur­rent long-ter­mist labor

Ad­ding op­ti­mistic num­bers to what I already said:

• Let’s say EAs con­tribute $50m† of re­sources per suc­cess­ful drug be­ing rol­led out across most of the US (mainly con­tribut­ing to re­search and ad­vo­cacy). We ig­nore costs paid by ev­ery­one else.

• This some­how causes rol­lout about 3 years ear­lier than it would oth­er­wise have hap­pened, and doesn’t trade off against the rol­lout of any other im­por­tant drug.

• At any one time, about 100 EAs†† use the now-well-un­der­stood, le­gal drug, and their baseline pro­duc­tivity is av­er­age for long-term-fo­cused EAs.

• This im­proves their pro­duc­tivity by an ex­pected 5%††† vs al­ter­na­tive men­tal health treat­ment.

• Bot­tom line: your $50m buys you about 100 x 5% x 3 = 15 ex­tra EA-years via this mechanism, at a price of $3.3m per per­son-year.

Sup­pose we would trade off $300k for the av­er­age per­son-year††††. This gives a re­turn on in­vest­ment of about $300k/​$3.3m = 0.09x. Even with op­ti­mistic num­bers, up­side 1 jus­tifies a small frac­tion of the cost, and with midline es­ti­mates and model er­rors I’d ex­pect more like a ~0.001x mul­ti­plier. Thus, this part of the ar­gu­ment is in­signifi­cant.

-----

Also, I’ve de­cided to just re­ply to this thread, be­cause it’s the only one that seems de­ci­sion-rele­vant.

† Var­i­ous es­ti­mates of the cost of in­tro­duc­ing a drug here, with a 2014 es­ti­mate be­ing $2.4bn. I guess EAs could only cover the early stages, with much of the rest be­ing picked up by drug com­pa­nies or some­thing.
†† Very, very op­ti­misti­cally, 1,000 long-term-fo­cused EAs in the US, 10% of the pop­u­la­tion suffer from rele­vant men­tal health is­sues, and all of them use the new drug.
††† This looks re­ally high but what do I know.
†††† Pretty made up but don’t think it’s too low. Yes, some­times years are worth more, but we’re look­ing at the whole pop­u­la­tion, not just se­nior staff.

I am con­fused about what ex­actly you are try­ing to com­mu­ni­cate with this post and its part­ner (https://​​fo­rum.effec­tivealtru­ism.org/​​posts/​​FDczXfT4xetcRRWtm/​​an-effec­tive-al­tru­ist-plan-for-so­cial­ism). My sense is that you are say­ing some­thing like:
1) Look, so­cial­ist-lean­ing and cap­i­tal­ist-lean­ing EAs, the poli­cies you prob­a­bly want are es­sen­tially the same, work to­gether and make some­thing hap­pen, and
2) Look, EAs, I can write nearly iden­ti­cal posts with ti­tles that will make you as­sume the posts are at odds—challenge your as­sump­tions. Real­ize the power lan­guage has over you.

Or maybe you pri­mar­ily want en­gage­ment with the poli­cies you are most ex­cited about, but want com­ments from both so­cial­ists and cap­i­tal­ists, and you felt this was the best way to achieve that?

I seek clar­ity.

(I feel stupid, not be­ing able to in­ter­pret you well, albeit only af­ter one quick read-through. But, I think folks should typ­i­cally make com­ments when con­fused, so here I am.)

If we are com­par­ing donat­ing blood to some­thing po­ten­tially more effec­tive we could do with our time, like earn­ing a wage to donate, don’t we need to con­sider whether the op­por­tu­nity to earn that wage is ac­tu­ally available? For ex­am­ple I don’t have the op­por­tu­nity to earn money at any hour of my choos­ing. I get paid to work 40 hours a week, so tak­ing two hours on the week­end to donate blood isn’t sac­ri­fic­ing 2 hours of wage earn­ing be­cause I’ve already earned the max­i­mum I can that week.

I sus­pect that straight­for­wardly tak­ing spe­cific EA ideas and putting them into fic­tion is go­ing to be very hard to do in a non-cringe­wor­thy way (as pointed out by elle in an­other com­ment). I’d be more in­ter­ested in at­tempts to write fic­tion that con­veys an EA mind­set with­out be­ing overly con­cep­tual.

For in­stance, a lot of to­day’s fic­tion seems cyn­i­cal and pes­simistic about hu­man na­ture; the char­ac­ters fre­quently don’t seem to have goals re­lated to any­thing other than their im­me­di­ate so­cial en­vi­ron­ment; and they of­ten don’t pur­sue those goals effec­tively (ap­par­ently for the sake of dra­matic ten­sion). Fic­tion demon­strat­ing peo­ple work­ing effec­tively on am­bi­tious, broadly benefi­cial goals, per­haps with dra­matic ten­sion caused by some­thing other than hu­mans be­ing ter­rible to each other, could help prop­a­gate EA mind­set.

Both the hover-over and side­notes on gw­ern.net are pure JS, and re­quire no mod­ifi­ca­tions to the origi­nal Mark­down or gen­er­ated HTML foot­notes; they just run and mod­ify the ap­pear­ance clientside and de­grade to the origi­nal foot­notes if JS is dis­abled.

Have you heard of Harry Pot­ter and the Meth­ods of Ra­tion­al­ity (http://​​www.hp­mor.com/​​) and/​​or http://​​un­song­book.com ? I think they serve some of this role for the com­mu­nity already.

It’s in­ter­est­ing they are both long-form web fic­tion; we don’t have EA tv shows or rock bands that I know of.

Ar­gu­ment in OP:

In­ter­ven­tions that in­crease the set of well-in­ten­tioned + ca­pa­ble peo­ple also seem quite ro­bust to clue­less­ness, be­cause they al­low for more er­ror cor­rec­tion at each timestep on the way to the far fu­ture.

The psychedelic ex­pe­rience also seems like a plau­si­ble lever on in­creas­ing ca­pa­bil­ity (via re­duc­ing nega­tive self-talk & other men­tal blocks) and im­prov­ing in­ten­tions (via ego dis­solu­tion chang­ing one’s meta­phys­i­cal as­sump­tions).

I view this as a weak ar­gu­ment. I think one could make this sort of ar­gu­ment for a large num­ber of in­ter­ven­tions: read­ing great liter­a­ture, yoga, a huge num­ber of pro­duc­tivity sys­tems, par­ti­ci­pat­ing in healthy com­mu­ni­ties, quan­tified self, vol­un­teer­ing for lo­cal char­i­ties like work­ing at a soup kitchen, etc. Some of these in­ter­ven­tions fo­cus more on the in­creas­ing ca­pa­bil­ity as­pect (pro­duc­tivity sys­tems, pro­duc­tivity sys­tems) and some fo­cus more on im­prov­ing in­ten­tions (par­ti­ci­pat­ing in healthy com­mu­ni­ties, vol­un­teer­ing). Some fo­cus on both to some de­gree.

The rea­son it seems like a weak ar­gu­ment to me is be­cause:

(a) the av­er­age effects of psychedelics on in­creas­ing ca­pa­bil­ity seem un­likely to be strong. They may be high for a small per­centage of peo­ple, but I’m not aware of any par­tic­u­larly strong rea­son to think that the av­er­age effects are large.

They may be large for peo­ple with men­tal health is­sues, but then it’s not re­ally an in­ter­ven­tion for in­creas­ing ca­pa­bil­ity in gen­eral, it’s a men­tal health in­ter­ven­tion. Th­ese are dis­tinct, and as I said above, psychedelics could plau­si­bly be a top in­ter­ven­tion for men­tal health.

(b) The im­prov­ing in­ten­tions as­pect looks to be on even shak­ier grounds. What is the ev­i­dence that tak­ing psychedelics is an effec­tive treat­ment for im­prov­ing in­ten­tions in a man­ner rele­vant to work­ing on the long term? I’ve never heard of any psychedelic or spiritual com­mu­nity be­ing fo­cused on long ter­mism in an EA rele­vant man­ner. Some peo­ple re­port ego dis­solu­tion, but I’m not even aware of any anec­do­tal re­ports that ego dis­solu­tion led to non-EAs think­ing and work­ing on long term things. It sounds like you know some cases where it may have been helpful, but I’m skep­ti­cal that a high qual­ity study would re­port some­thing amaz­ing.

No, I ex­pected that no rigor­ous re­search had been done on NLP as of 2014, and I don’t know how rigor­ous the more re­cent re­search has been.

I like your en­courage­ment to cre­ate more art. How­ever, I no­ticed cring­ing at some of your ideas in the ap­pendix. I worry that they would end up be­ing “poorly ex­e­cuted cul­tural arte­facts [that] may put EA into dis­re­pute” as you put it.

I do not feel ca­pa­ble of ex­plain­ing ex­actly where the cringe re­ac­tion is com­ing from, but a few ex­am­ples:

I do not like the idea in Beau­topia of equat­ing phys­i­cal ap­pear­ance with moral good­ness, given that a) it is already an is­sue that peo­ple as­sume pos­i­tive per­son­al­ity traits when they see phys­i­cally at­trac­tive peo­ple and b) it as­sumes there is some ob­jec­tive and real “good” that can be calcu­lated. And the fi­nal plot line im­ply­ing that it is good to kill peo­ple we think are evil seems like a bad meme to spread.

Dead baby cur­rency seems overly sim­plis­tic and in­sen­si­tive, al­though I am hav­ing a hard time putting words to why. It also trig­gers scrupu­los­ity con­cerns (for ex­am­ple, see http://​​www.giv­ing­gladly.com/​​2012/​​03/​​trade­offs.html ).

Fi­nally, I am wary of how you re­fer to “Africa” mono­lith­i­cally. For more, see https://​​www.the­at­lantic.com/​​in­ter­na­tional/​​archive/​​2013/​​08/​​con­fus­ing-coun­try-con­ti­nent-how-we-talk-about-africa/​​311621/​​.

Prob­a­bly the crux here is that I think ra­tio­nal­ity train­ing & the psychedelic ex­pe­rience can achieve similar kinds of be­hav­ior change (e.g. less en­ergy spent on nega­tive self-talk & un­helpful per­sonal nar­ra­tives) such that their effect sizes can be com­pared.

Whereas you think that ra­tio­nal­ity train­ing & the psychedelic ex­pe­rience are differ­ent enough that be­liev­able com­par­i­son isn’t pos­si­ble.

Does that sound right to you?

Ra­tion­al­ity pro­jects: I don’t care to ar­bi­trate what counts as EA.

Isn’t much of the pre­sent dis­cus­sion about “what counts as EA?”

Maybe I’m get­ting hung up on se­man­tics. The ques­tion I most care about here is: “what top­ics should EAs ded­i­cate re­search ca­pac­ity & cap­i­tal to?”

Does that seem like a worth­while ques­tion?

It would be helpful if you could agree with or con­test with that claim be­fore we move on to the other up­side.

Right. I’m say­ing that the math we should care about is:

• effect from boost­ing effi­cacy of cur­rent long-ter­mist la­bor + effect from in­creas­ing the amount of long-ter­mist la­bor + effect from short-ter­mist benefits

I think that math is likely to work out.

Given your pri­ors, we’ve been dis­count­ing “effect from short-ter­mist benefits” to 0.

So the math is then:

• effect from boost­ing effi­cacy of cur­rent long-ter­mist la­bor + effect from in­creas­ing the amount of long-ter­mist labor

And I think that is also likely to work out, though the case is some­what weaker when we dis­count short-ter­mist benefits to 0.

(I also dis­agree with dis­count­ing short-ter­mist benefits to 0, but that’s doesn’t feel like the crux of our pre­sent dis­agree­ment.)

Psychedelic in­ter­ven­tions seem promis­ing be­cause they can plau­si­bly in­crease the num­ber of ca­pa­ble peo­ple fo­cused on long-ter­mist work, in ad­di­tion to plau­si­bly boost­ing the effi­cacy of those already in­volved.

Point­ing out that there are two up­sides is helpful, but I had just made this claim:

The math for [the bold part] seems re­ally un­likely to work out.

It would be helpful if you could agree with or con­test with that claim be­fore we move on to the other up­side.

-

Ra­tion­al­ity pro­jects: I don’t care to ar­bi­trate what counts as EA. I’m go­ing to steer clear of pre­sent-day state­ments about spe­cific orgs, but you can see my dona­tion record from when I was a trader on my LinkedIn pro­file.

I’m not ar­gu­ing against try­ing to com­pare things. I was say­ing that the com­par­i­son wasn’t in­for­ma­tive. Com­par­ing dis­similar effects is valuable when done well, but com­par­ing d-val­ues of differ­ent effects from differ­ent in­ter­ven­tions tells you very lit­tle.

If each of them donated an av­er­age of U$ 1 for this cause, they would match all of GDs trans­fers in 2017

fwiw I think it’s very hard to get peo­ple to donate to things.

From sec­tion 4(b) of the OP: “Roughly $40 mil­lion has been com­mit­ted to psychedelic re­search since 2000.”

Got it, thanks.

So, I es­ti­mate GD re­sults in U$200/​QALY…

Enthea’s es­ti­mate of psychedelics liber­al­iza­tion is of $472/​DALY

As far as I know, GiveWell con­sid­ers cost-effec­tive­ness es­ti­mates as in­for­ma­tive for effi­cacy differ­ences that are or­ders of mag­ni­tude apart.

For two in­ter­ven­tions that are on the same or­der of mag­ni­tude, the analy­ses aren’t gran­u­lar enough to be­liev­ably in­form which is more effec­tive.

OK, I re­al­ised the flaw in my ar­gu­men­ta­tion. If I have 1000 GBP to give away, I could ei­ther ‘walk’ 1000 GBP in di­rec­tion of char­ity x or 1000 GBP in di­rec­tion of char­ity y but only sqrt(x^2 + y^2) in a com­bi­na­tion of x and y, e.g. the max­i­mal gra­di­ent. The op­ti­mal al­lo­ca­tion (x, y) of money is what max­imises the scalar product of gra­di­ent (dU/​dx, dU/​dy) * (x, y) un­der the re­stric­tion that x + y = 1000. If dU/​dx = dU/​dy a ⁵⁰⁄₅₀ al­lo­ca­tion as good as an al­lo­ca­tion of all money to the most effec­tive char­ity. Other­wise giv­ing all money to the most effec­tive char­ity max­imises util­ity. Sorry for the con­fu­sion and thanks for the dis­cus­sion.

We’ve jumped from emo­tional blocks & un­helpful per­sonal nar­ra­tives to life satis­fac­tion & treat­ment-re­sis­tant de­pres­sion, which are very differ­ent.

fwiw I think nega­tive self-talk (a kind of emo­tional block) & un­helpful per­sonal nar­ra­tives are big parts of the sub­jec­tive ex­pe­rience of de­pres­sion.

As you note, the two effects you’re now com­par­ing (life satis­fac­tion & treat­ment-re­sis­tant de­pres­sion) aren’t re­ally the same at all.

Com­par­ing dis­similar effects is a core part of EA-style anal­y­sis, right?

The lat­ter seems very likely false. You would need the ad­di­tional cost of re­search­ing, ad­vo­cat­ing for and im­ple­ment­ing a spe­cific new treat­ment

Does this mean you think that pro­jects like CFAR & Paradigm Academy shouldn’t be as­so­ci­ated with the EA plank?

… speci­fi­cally long-term-fo­cused EAs in that ge­og­ra­phy (<0.001% of the pop­u­la­tion). The math for that seems re­ally un­likely to work out.

Psychedelic in­ter­ven­tions seem promis­ing be­cause they can plau­si­bly in­crease the num­ber of ca­pa­ble peo­ple fo­cused on long-ter­mist work, in ad­di­tion to plau­si­bly boost­ing the effi­cacy of those already in­volved. (See sec­tion 3(a) of the OP.)

The marginal value of each ad­di­tional value-al­igned + ca­pa­ble long-ter­mist is prob­a­bly quite high.