EA is vetting-constrained

Re: What to do with peo­ple? and After one year of ap­ply­ing for EA jobs: It is re­ally, re­ally hard to get hired by an EA organisation

Epistemic sta­tus: just my per­sonal im­pres­sion. Please prove me wrong.


So we know that:

1) In ag­gre­gate, there are billions of dol­lars in EA

2) There are lots of sur­plus tal­ented peo­ple look­ing for EA work, that can’t get it

and I would like to add:

3) I es­ti­mate that there are at least 10-20 bud­ding or­gani­sa­tions that would love to use this money to get these peo­ple to work, and scale be­yond their cur­rent size, and prop­erly tackle the prob­lem they aim to solve. I know at least 5 founders like that per­son­ally.

So with all the in­gre­di­ents in place for amaz­ing move­ment growth, why isn’t the magic hap­pen­ing?

Know­ing who to del­e­gate to

I agree with the idea that if you want to solve prob­lems, you need to or­ganise your sys­tem in a hi­er­ar­chy, with some kind of di­vide-and-con­quer strat­egy where a prin­ci­pal figures out the sub­prob­lems of a prob­lem and del­e­gates them to some agents, and re­curse.

One prob­lem here is that, even if the agent is al­igned, the prin­ci­pal needs some way to tell that the agent is ca­pa­ble of car­ry­ing out a prob­lem to a cer­tain stan­dard.

Differ­ent sys­tems solve this prob­lem in differ­ent ways. A com­pany might have some stan­dards for hiring and struc­turally re­view the perfor­mance of their em­ploy­ees. Academia re­lies on pres­tige and some met­rics that are prox­ies of qual­ity. The mar­ket gives the most money to those that sell the most pop­u­lar prod­ucts. Com­mu­ni­ties kick out mem­bers that cross bound­aries, and de­prive un­in­ter­est­ing peo­ple of at­ten­tion.

EA, by which I mean the es­tab­lished or­gani­sa­tions of EA, does this kind of thing in two ways: by hiring di­rectly, and by vet­ting pro­jects. For the lat­ter there are grant­mak­ers. As pro­fes­sion­als that have thought a lot about what pro­jects need to hap­pen, they take a long hard look at an ap­pli­ca­tion by a startup founder, and if they ex­pect it to work out well, they fund it. Sim­ple enough.

The state of vet­ting in EA

I want to clar­ify that none of the fol­low­ing is meant to be ac­cusatory. Grant­mak­ing sounds like one of the hard­est jobs in the world, and pro­jects are by no means en­ti­tled to EA money just be­cause they call them­selves EA. I hope that this post keeps a spirit of high trust, which I think is very im­por­tant.

So why aren’t we see­ing more new EA or­gani­sa­tions get­ting fund­ing? Two hy­pothe­ses come to mind:

  • The“high stan­dards” hy­poth­e­sis. Grant­mak­ers think that these new or­gani­sa­tions just aren’t up to stan­dard, and they would there­fore cause dam­age. Per­haps their model is that EA should re­tain a very high stan­dard to make sure that the pres­tige of the move­ment stays in­tact. After all that’s what the move­ment might need to in­fluence big in­sti­tu­tions like academia and gov­ern­ment.

  • The “vet­ting bot­tle­neck” hy­poth­e­sis. Grant­mak­ing or­gani­sa­tions are just way un­der­staffed. It’s not that they’re sure that these or­gani­sa­tions don’t meet the bar, it’s just that they can’t ver­ify that in time, so the safest op­tion is to hold off on fund­ing, or fund a more es­tab­lished or­gani­sa­tion in­stead.

In re­al­ity, it is prob­a­bly a com­bi­na­tion of both of these. Some anec­do­tal ev­i­dence:

  • When one startup got re­jected by a grant­mak­ing or­gani­sa­tion, and they pressed for feed­back, there were told that “We do not pos­sess the do­main ex­per­tise to eval­u­ate scal­able ex­is­ten­tial risk re­duc­tion pro­jects in the way that [other org] would be bet­ter placed to do.” And “as such, we rely more on the strength and qual­ity of refer­ences when mod­el­ing out the po­ten­tial im­pact of pro­jects.” This was af­ter they were in­vited to the in­ter­view stage. It sug­gests that grant­mak­ers fall back on pres­tige be­cause they don’t always have the re­sources to prop­erly eval­u­ate ideas.

  • Another startup con­tacted at least 4 grant­mak­ing or­gani­sa­tions. Three of them deferred to the fourth. This or­gani­sa­tion would in­form them of their de­ci­sion in Novem­ber, then post­poned to De­cem­ber, then post­poned to “af­ter the holi­days”, but they haven’t re­sponded yet. They once men­tioned to the startup that “we be­lieve that we don’t cur­rently have the ca­pac­ity to re­view your ap­pli­ca­tion suffi­ciently”.

  • Quote 80k: “One rea­son why these donors don’t give more is a lack of con­crete “shovel-ready” op­por­tu­ni­ties. This is partly due to a lack of qual­ified lead­ers able to run pro­jects in the top prob­lem ar­eas (es­pe­cially to found non-prof­its work­ing on re­search, policy and com­mu­nity build­ing). But an­other rea­son is a lack of grant­mak­ers able to vet these op­por­tu­ni­ties or start new pro­jects them­selves.”

  • If you look at where, for ex­am­ple, EA Funds spends their money, it seems like most of the funds are just go­ing to safe bets that don’t need much vet­ting.

So what does all of this sug­gest?

What kind of stan­dards grant­mak­ers should have, is up for de­bate. I’m per­son­ally un­der the im­pres­sion that the star­dards are too high. There are a lot of star­tups out there that would in­crease to­tal di­rect value. In you be­lieve the pres­tige of EA (through the av­er­age qual­ity of its pro­jects) is more im­por­tant than its to­tal di­rect value, con­sider that pres­tige is a nega­tive-sum game. but that’s just my 2 low-effort cents and it’s off-topic.

Re­gard­less of the bar that we think an EA startup should meet, I don’t think the cur­rent pat­tern of pay­outs re­flects the set of or­gani­sa­tions that meet that bar. I’d be very sur­prised if ex­actly all of the es­tab­lished orgs are do­ing bet­ter work per marginal dol­lar than ex­actly all of the new ones.

This es­pe­cially be­cause es­tab­lished or­gani­sa­tions are pre­cisely the ones that aren’t fund­ing con­strained. Even the ra­tio­nale of the EA Funds pay­out men­tions “a sense that their work is oth­er­wise much less fund­ing con­strained than it used to be”. The grant­maker sug­gests spend­ing the money on child care and up­grad­ing elec­tron­ics. He doesn’t seem to be aware of any good fund­ing con­strained or­gani­sa­tions (This was in Au­gust, and it looks like they’ve moved to smaller pro­jects now).

So scale up the vet­ting! Then fund more orgs! And all of those amaz­ingly com­pe­tent peo­ple will even­tu­ally find a job in one of them, and who knows, net util­ity of EA might end up or­ders of mag­ni­tude larger.

Again, just my im­pres­sion. Please prove me wrong.