First, con­grat­u­la­tions!!

Miranda Dixon-Luinen­burg and I had EA themes through­out our wed­ding cer­e­mony. You’re wel­come to read and bor­row from our cer­e­mony text.

Con­text: we had our wed­ding in a plane­tar­ium, we had our friends write speeches each ac­cord­ing to a theme, those speeches were read while a match­ing starscape was pro­jected above.

Not ex­actly EA, but part of a sci­en­tific wor­ld­view: I had the end of the last para­graph of Ori­gin read at my wed­ding.

Thus, from the war of na­ture, from famine and death, the most ex­alted ob­ject which we are ca­pa­ble of con­ceiv­ing, namely, the pro­duc­tion of the higher an­i­mals, di­rectly fol­lows. There is grandeur in this view of life, with its sev­eral pow­ers, hav­ing been origi­nally breathed into a few forms or into one; and that, whilst this planet has gone cy­cling on ac­cord­ing to the fixed law of grav­ity, from so sim­ple a be­gin­ning end­less forms most beau­tiful and most won­der­ful have been, and are be­ing, evolved.

In fact, it’s pretty un-EA to say that “higher” an­i­mals are “the most ex­alted ob­ject we are ca­pa­ble of con­ceiv­ing,” haha.

...

There’s a lot of Zen stuff about us­ing your in­ti­mate re­la­tion­ships as a sup­port­ive place to learn al­tru­ism which can then be ap­plied to wider and wider cir­cles. That seems pretty ap­pro­pri­ate for a wed­ding. I don’t have any links off the top of my head be­cause I usu­ally hear this kind of thing at dharma talks, but it’s usu­ally along the lines of some­one ask­ing a Zen mas­ter how to be a bet­ter per­son and get­ting the an­swer, “Every day when you wake up, think ‘only for my wife, only for my wife.’ When your wife’s welfare is like your own, think ‘only for my fam­ily’” and so on through the neigh­bor­hood, the com­mu­nity, the city, the coun­try, the world. The lo­cal­ist hi­er­ar­chy isn’t EA, but the idea that you have to level up your com­pas­sion with the sup­port and com­mit­ment of those you are close to brings EA themes to­gether with mar­riage.

This post seems very in­sight­ful to me, and it seems like it worked out very well in terms of up­votes (and it seems like it would in­crease your chances of get­ting fund­ing)? I’d be in­ter­ested to learn who wrote this, but of course no need to say if you pre­fer not to. :)

When long-ter­mist grant ap­pli­ca­tions don’t get funded, the rea­son usu­ally isn’t lack of fund­ing, but one of the fol­low­ing:

• The grant­maker was un­able to vet the pro­ject (due to time con­straints or lack of do­main ex­per­tise) or at least thought it was a bet­ter fit for a differ­ent grant­maker.

• The grant­maker thought the pro­ject came with a high risk of ac­ci­den­tal harm.

This post con­tains great ideas for re­solv­ing the former point, but doesn’t demon­strate high aware­ness of the lat­ter con­cern. Aware­ness of these risks seems im­por­tant to me, es­pe­cially for fun­ders: High-qual­ity grant ap­pli­ca­tions tend to get funded quickly and are thereby elimi­nated from the pool of pro­pos­als available to the EA com­mu­nity, while ap­pli­cants with higher-risk pro­pos­als tend to ap­ply/​pitch to lots of fun­ders. This means that on av­er­age, pro­pos­als sub­mit­ted to fun­ders will be skewed to­wards high-down­side-risk pro­jects, and fun­ders could them­selves eas­ily do harm if they end up sup­port­ing many of them. I’d be in­ter­ested in your thoughts on that.

I re­ally like that you’re pro­vid­ing feed­back to ap­pli­cants! In gen­eral, I wish the EA com­mu­nity was more proac­tive with pro­vid­ing crit­i­cal feed­back.

I haven’t yet finished think­ing about how the EA Fo­rum Team should go about do­ing this, given their par­tic­u­lar re­la­tion­ship to the site’s mem­bers, but here’s a few thoughts.

I think, for a plat­form to be able to in­cen­tivise long-term in­tel­lec­tual progress in a com­mu­nity, it’s im­por­tant that there are in­di­vi­d­u­als trusted on the plat­form to pro­mote the best con­tent to a place on the site that is both last­ing and clearly more im­por­tant than other con­tent, like I and oth­ers have done on the AI Align­ment Fo­rum and LessWrong. Other­wise the site de­volves into a news site, with a cul­ture that de­pends on who turns up that par­tic­u­lar month.

I do think the pre­vi­ous in­car­na­tion of the EA Fo­rum was much more of a news site, where the most ac­tivity oc­curred when peo­ple turned up to de­bate the lat­est con­tro­versy posted there, and that the ma­jor­ity of posts and dis­cus­sion on the new Fo­rum are much more in­ter­ested in dis­cus­sion of the prin­ci­ples and prac­tice of EA, rather than con­flict in the com­mu­nity.

(Note that, while it is not the only or biggest differ­ence, LessWrong and Hacker News both have the same sort­ing al­gorithm on their posts list, yet LW has the best con­tent shown above the re­cent con­tent, and thus is more clearly a site that re­wards the best con­tent over the most re­cent con­tent.)

It’s okay to later build slower and more de­liber­a­tive pro­cesses for figur­ing out what gets pro­moted (al­though you must move much more quickly than the pre­sent day aca­demic jour­nal sys­tem, and with more feed­back be­tween re­searchers and eval­u­a­tors). I think the Fo­rum’s monthly prize sys­tem is a good way to in­cen­tivise good con­tent, but it cru­cially doesn’t en­sure that the re­warded con­tent will con­tinue to be read by new­com­ers 5 years af­ter it was writ­ten. (Added: And similar­ly­cur­rent new EAs on the Fo­rum are not read­ing the best EA con­tent of the past 10 years, just the most re­cent con­tent.)

I agree it’s good for mem­bers of the com­mu­nity to be able to cu­rate con­tent them­selves. Right now any­one can build a se­quence on LessWrong, then the LW team moves some of them up into a cu­rated sec­tion which later get high­lighted on the front page (see the library page, which will be­come more promi­nent on the site af­ter our new front­page re­work). I can imag­ine this be­ing an au­to­matic pro­cess based on vot­ing, but I have an in­tu­ition that it’s good for hu­mans to be in the loop. One rea­son is that when hu­mans make de­ci­sions, you can ask why, but when 50 peo­ple vote, it’s hard to in­ter­ro­gate that sys­tem as to the rea­son be­hind its de­ci­sion, and im­prove its rea­son­ing the next time.

(Thanks for your com­ment Brian, and please don’t feel any obli­ga­tion to re­spond. I just no­ticed that I didn’t in­tu­itively agree with the thrust of your sug­ges­tion, and wanted to offer some mod­els point­ing in a differ­ent di­rec­tion.)

I hear more peo­ple do cold out­reach about be­ing a re­searcher than RA, and my guess is that 3-10x more peo­ple ap­ply for re­searcher than RA jobs even when they are ad­ver­tised. I think it’s a com­bi­na­tion of those two fac­tors.

My recom­men­da­tion would be that peo­ple ap­ply more to RA jobs that are ad­ver­tised, and also reach out to make op­por­tu­ni­ties for them­selves when they are not.

I think about half of re­searchers can use re­search as­sis­tants, whether or not they are cur­rently hiring for one. A ma­jor rea­son re­searchers don’t make re­search as­sis­tant po­si­tions available is they don’t ex­pect to find one worth hiring, and so don’t want to in­cur the ad­minis­tra­tive bur­den. Or maybe they don’t feel com­fortable ask­ing their bosses for this. But if you are a strong can­di­date, coldly reach­ing out may re­sult in you be­ing hired or may trig­ger a hiring round for that po­si­tion. Although of­ten strong can­di­dates would be peo­ple I have met at an EA con­fer­ence, that got far in an in­tern­ship ap­pli­ca­tion, or that has been referred to me.

I don’t think the salaries would be any lower than com­pet­i­tive rates.

It’s good to see peo­ple try­ing small low cost ex­per­i­ments like this al­though I sus­pect there are diminish­ing re­turns to how many peo­ple will be in­ter­ested if you try a similar post again but that’s also why it’s good that you’ve given a frame­work for oth­ers to copy.

At EA Lon­don I haven’t given as many books out in the last 12 months as the pre­vi­ous 12 as I’ve usu­ally only given them to peo­ple af­ter a 1-1 meet­ing or sig­nifi­cant in­ter­est from some­one at an event where it seems more likely that they’ll ac­tu­ally read it and want to take away some ad­vice rather than offer­ing to give it to peo­ple who are more in­ter­ested in the idea of a “free book”.

My im­pres­sion was based mostly on our con­ver­sa­tions sev­eral months ago—quot­ing the notes from that time

lot of the dis­cus­sion and de­bate de­rives from differ­ing as­sump­tions held by the par­ti­ci­pants re­gard­ing the po­ten­tial for bad/​risky pro­jects: Ben­jamin/​Bren­don gen­er­ally point out the lack of data/​sig­nal in this area and be­lieve launch­ing an open pro­ject plat­form could provide data to re­duce un­cer­tainty, whereas Jan is more con­ser­va­tive and pri­ori­tizes cre­at­ing a rigor­ous cu­ra­tion and eval­u­a­tion sys­tem for new pro­jects.

I think it is fair to say you ex­pected very low risk from cre­at­ing an open plat­form where peo­ple would just post pro­jects and seek vol­un­teers and fund­ing, while I ex­pected with min­i­mum cu­ra­tion this cre­ates sig­nifi­cant risk (even if the risk is com­ing from small frac­tion of pro­jects). Sorry if I rounded off sug­ges­tions like “let’s make an open plat­form with­out care­ful eval­u­a­tion and see” and “based on the pro­ject ideas lists which ex­isted sev­eral years ago the amount of harm­ful pro­jects seems low” to “wor­ry­ing about them is pre­ma­ture”.

Read­ing your re­cent com­ment, it seems more care­ful, and point­ing out large nega­tive out­comes are more of a prob­lem with x-risk/​long-term ori­ented pro­jects.

In our old dis­cus­sions I also ex­pressed some doubt about your or al­tru­ism.vc abil­ity to eval­u­ate x-risk and similar pro­jects, where your re­cent post states that pro­jects that im­pact x-risks by do­ing some­thing like AI safety re­search has not yet ap­plied to the EA An­gel Group.

I guess part of the dis­agree­ment comes from the fact that I have fo­cus on x-risk and the long-term fu­ture, and I’m more in­ter­ested both in im­prov­ing the pro­ject land­scape in these ar­eas, and more wor­ried about nega­tive out­comes.

If open plat­forms or similar eval­u­a­tion pro­cess also ac­cept miti­gat­ing x-risk and similar pro­pos­als, in my opinion, un­for­tu­nately the bar how good/​ex­pert driven eval­u­a­tions you need is higher, and un­for­tu­nately sig­nals like “this is a com­pe­tent team” which VCs would mainly look at are not enough.

Be­cause I would ex­pect the long-term im­pact will come mainly from long-term, meta-, ex­plo­ra­tory or very am­bi­tious pro­jects, I think you can be ba­si­cally right about low ob­vi­ous risk of all the pro­jects his­tor­i­cally posted on hack­pad or pro­posed to al­tru­ism.vc, and still miss the largest term in the EV.

Milan asked this ques­tion and I an­swered it.

Thanks—both of that hap­pened af­ter I posted my com­ment, and also I still do not see the num­bers which would help me es­ti­mate the ra­tio of pro­jects which ap­plied and which got funded. I take as mildly nega­tive sig­nal that some­one had to ask, and this info was not in­cluded in the post, which so­lic­its pro­ject pro­pos­als and vol­un­teer work.

In my model it seems pos­si­ble you have some­thing like chicken-and-egg prob­lem, not get­ting many great pro­pos­als, and the group of un­named an­gels not fund­ing many pro­pos­als com­ing via that pipeline.

If this is the case and the ac­tual num­ber of suc­cess­fully funded pro­jects is low, I think it is nec­es­sary to state this clearly be­fore invit­ing peo­ple to work on pro­pos­als. My vague im­pres­sion was we may dis­agree on this, which seems to in­di­cate some quite deep dis­agree­ment about how fun­ders should treat pro­jects.

I’m not en­tirely sure what your rea­sons are for hav­ing this opinion, or what you even mean

The whole con­text was, Ryan sug­gested I should have sought some feed­back from you. I ac­tu­ally did that, and your co-founder noted that he will try to write the feed­back on this to­day or to­mor­row, on 11th of Mar—which did not hap­pen. I don’t think this is large prob­lem, as we had already dis­cussed the topic ex­ten­sively.

When writ­ing it I was some­what up­set about the mode of con­ver­sa­tion where crit­ics do ask whether I tried to co­or­di­nate with some­one, but just as­sume I did not. I apol­o­gize for the bad way it was writ­ten.

Over­all my sum­mary is we prob­a­bly still dis­agree in many as­sump­tions, we did in­vest some effort try­ing to over­come them, it seems difficult for us to reach some con­sen­sus, but this should not stop us try­ing to move for­ward.

John Maxwell brought up some in­ter­est­ing points. He sug­gests that plat­forms can ex­pe­rience the chicken and egg prob­lem when it comes to get­ting started, and that in­ten­sive net­work­ing is a way to over­come this is­sue. I agree that plat­forms of­ten have this prob­lem, but the EA An­gel Group re­solved this not by net­work­ing in­tensely but in­stead by offer­ing a lot of value to an­gels. This would in­cen­tivize them to join the plat­form even with­out a large num­ber of ex­ist­ing grant ap­pli­cants which would in turn in­cen­tivize grant ap­pli­cants to ap­ply.

Of course, we do need a stream of in­com­ing grant ap­pli­ca­tions to re­main vi­able, and un­for­tu­nately we en­coun­tered some un­ex­pected is­sues when at­tempt­ing to col­lab­o­rate with EA Grants and speak to many com­mu­nity mem­bers as part of sev­eral strate­gies to ac­quire grant ap­pli­ca­tions. As men­tioned in my progress up­date com­ment, I am cur­rently pur­su­ing al­ter­nate strate­gies to achieve this ob­jec­tive which in­volve steps that I have greater con­trol over (and less steps that re­quire the ap­proval of en­tities whose de­ci­sions I can­not in­fluence). That be­ing said, I think net­work­ing and col­lab­o­ra­tion is highly valuable, and am scal­ing that up even as I pur­sue strate­gies that do not re­quire net­work­ing to suc­ceed.

Con­cern­ing trans­la­tion, it can be a mis­take to imag­ine it’s nec­es­sary to trans­late the whole of texts, or large texts.

In­stead, trans­lat­ing a ti­tle and sum­mary, or first para­graphs, or a con­tents page or back cover of a book, can be enough to help peo­ple de­cide in if they want to trans­late the whole thing into their own lan­guage, or read with the help of google/​dic­tio­nary.

If any­one out there is in­ter­ested in sup­port­ing EAs in In­dia, or vis­it­ing EAs in In­dia, please feel free to mes­sage me via www.ALLFED.info ….or join the Effec­tive Altru­ism In­dia Face­book group

I wrote a progress up­date com­ment re­gard­ing the EA An­gel Group which cov­ered our grant op­por­tu­nity dis­cov­ery ac­tivi­ties over the last few months. We spoke with EA Grants sev­eral months ago, and to the best of my knowl­edge they are still de­ter­min­ing whether to send and re­ceive grant ap­pli­ca­tions with other fun­ders. At least one ma­jor fund­ing group has ex­pressed sig­nifi­cant in­ter­est in send­ing and re­ceiv­ing grant ap­pli­ca­tions with the EA An­gel Group, and we are in the pro­cess of talk­ing with var­i­ous fun­ders about this.

I men­tioned the one con­cern I heard and my re­sponse to it in my progress up­date com­ment:

One ob­jec­tion to shar­ing grant ap­pli­ca­tions among fun­ders is that a fun­der would fund all of the grant pro­pos­als they felt were good and clas­sify all other grant pro­pos­als as not suit­able to be funded. From the fun­der’s per­spec­tive, shar­ing the un­funded grant pro­pos­als would be bad since other or­ga­ni­za­tions could sub­se­quently fund them, and the fun­der clas­sified those grant pro­pos­als as not worth fund­ing. I per­son­ally dis­agree with this ob­jec­tion be­cause the ar­gu­ment as­sumes that a fun­der has de­vel­oped a grant eval­u­a­tion pro­cess that can ac­tu­ally iden­tify suc­cess­ful pro­jects with a high de­gree of ac­cu­racy. Since the norm in the for-profit world in­volves large and suc­cess­ful ven­ture cap­i­tal firms with lots of ex­pe­rienced do­main ex­perts reg­u­larly pass­ing on op­por­tu­ni­ties that later be­come multi­billion-dol­lar com­pa­nies, I find it un­likely that any EA fund­ing or­ga­ni­za­tion will de­velop a grant eval­u­a­tion pro­cess that is so good it jus­tifies hid­ing some or all un­funded ap­pli­ca­tions.

Can you elab­o­rate on:

I think for ex­am­ple that a ‘just-an­other-uni­ver­sal-pro­to­col’ worry would be very rea­son­able to have here.

Are you sug­gest­ing that fun­ders may be con­cerned about adopt­ing a pro­to­col which ends up pro­vid­ing limited value? As I’ve stated in sev­eral other com­ments, I think shar­ing grant ap­pli­ca­tions can be of con­sid­er­able value since ar­bi­trar­ily limit­ing the pool of pro­jects seems pretty sub­op­ti­mal.

To avoid that I think we need to do the hard work of reach­ing out to in­volved par­ties and have many con­ver­sa­tions to in­cor­po­rate their most im­por­tant con­sid­er­a­tions and start mu­tu­ally use­ful col­lab­o­ra­tions. I.e. con­sen­sus build­ing.

I agree. I did some ini­tial out­reach at first and will be­gin ad­di­tional out­reach shortly.

Thanks for point­ing that out! Jan and I have also talked out­side the EA Fo­rum about our opinions on risk in the EA pro­ject space. I’ve been more op­ti­mistic about the prevalence of nega­tive EV pro­jects, so I thought there was a chance that greater op­ti­mism was be­ing mis­in­ter­preted as a lack of con­cern about nega­tive EV pro­jects, which isn’t my po­si­tion.

I ac­tu­ally stated my opinion in writ­ing in a re­sponse to you two days ago which seems to de­vi­ate highly from your in­ter­pre­ta­tion of my opinion.

I think I’ve seen fo­rum dis­cus­sions where lan­guage has been an un­ac­knowl­edged bar­rier to un­der­stand­ing in the past, so it might be worth flag­ging that Jan is from the Czech Repub­lic and likely does not speak English as his mother tongue.

It seems like Jan is get­ting a lot of crit­i­cal feed­back, so I just want to say, big ups to you Jan for spear­head­ing this. Per­haps it’d be use­ful to sched­ule a Skype call with Habryka, RyanCarey, or oth­ers to try & hash out points of dis­agree­ment.

The point of a pi­lot pro­ject is to gather in­for­ma­tion, but if in­for­ma­tion already ex­ists in the heads of com­mu­nity mem­bers, a pi­lot could just be an ex­pen­sive way of re-gath­er­ing that info. The ideal pi­lot might be some­thing that is con­tro­ver­sial among the most knowl­edgable peo­ple in the com­mu­nity, with some op­ti­mistic and some pes­simistic, be­cause that way we’re gath­er­ing in­for­ma­tive ex­per­i­men­tal data.

I con­sider hav­ing peo­ple giv­ing feed­back to have ‘skin in the game’ to be im­por­tant for the ac­cu­racy of the feed­back. Most peo­ple don’t en­joy dis­cour­ag­ing oth­ers they have so­cial ties with. Often re­view­ers with­out suffi­cient skin in the game might be tempted to not be as openly nega­tive about pro­pos­als as they should be.

Maybe anonymity would be helpful here, the same way sci­en­tists do anony­mous peer re­view?

Since I view this as an im­por­tant idea, I think it’s im­por­tant to get the plan to the strongest point that it can be.

It’s also im­por­tant not to let the perfect be the en­emy of the good. Seems to me like peo­ple are always propos­ing vol­un­teer-lead pro­jects like this and most of them never get off the ground. Re­mem­ber this is just a pi­lot.

I think this is the sort of pro­ject where if you do it badly, it might dis­suade oth­ers from try­ing the same.

The em­piri­cal re­al­ity of the EA pro­ject land­scape seems to be that EAs keep stum­bling on the same pro­ject ideas over and over with lit­tle aware­ness of what has been pro­posed or at­tempted in the past. If this post goes like the typ­i­cal pro­ject pro­posal post, noth­ing will come of it, it will soon be for­got­ten, and 6 months later some­one will in­de­pen­dently come up with a similar idea and write a similar post (which will meet a similar fate).

This is a good post. But since we hear so much about the value of ca­reer cap­i­tal, I thought it’d be use­ful to link this old post which en­couraged peo­ple to de­pri­ori­tize it, just for the sake of an al­ter­nate per­spec­tive.

Given that there’s always go­ing to be a so­cial bias to­wards work­ing at EA orgs

I’m not sure this is true. Just a few years ago, it seemed like there was a so­cial bias against work­ing at EA orgs. The “pri­ori­tize tal­ent gaps” meme was meant to ad­dress this. (I feel like there might be other his­tor­i­cal cases of the EA move­ment over­cor­rect­ing in this man­ner, but no spe­cific in­stances are com­ing to mind.)

In my opinion, or­ga­ni­za­tions may do best to avoid offi­cially en­dors­ing any­thing other than the most cen­tral con­tent that they pro­duce in or­der to re­duce these PR headaches, re­gard­ing both

1. what’s said in the en­dorsed ar­ti­cles and

2. which ar­ti­cles were or weren’t cho­sen to be­gin with (the de­bate over the EA Hand­book comes to mind).

As an al­ter­na­tive, maybe in­di­vi­d­ual peo­ple could cre­ate their own non-CEA-en­dorsed lists of recom­mended con­tent, and these could be made available some­where. Hav­ing many such lists would al­low for di­ver­sity based on in­ter­ests and val­ues. (For ex­am­ple, “The best global poverty ar­ti­cles”, “The best ca­reer ar­ti­cles”, “The best ar­ti­cles for suffer­ing-fo­cused al­tru­ists”, etc.)

Nope, it’s full-time. Right now two of us are do­ing a side pro­ject, but that’s not usual