EA Community Building Grants Update

Effec­tive Altru­ism Com­mu­nity Build­ing Grants is a pro­ject run by the Cen­tre for Effec­tive Altru­ism (CEA), pro­vid­ing grants of be­tween $5,000 and $100,000 to in­di­vi­d­u­als and groups do­ing lo­cal effec­tive al­tru­ism com­mu­nity build­ing for the aca­demic year of 2018-19. We be­lieve this could be a high im­pact pro­ject for the EA com­mu­nity in the long term, be­cause it is a way in which we can in­crease the time and effort that goes into EA com­mu­nity build­ing.

This post is an up­date on how the pro­ject is go­ing so far, in­clud­ing an overview of the ap­pli­ca­tion pro­cess, the grants we’ve made, our eval­u­a­tion crite­ria, and plans for the fu­ture.

Background

Effec­tive al­tru­ism groups are an in­te­gral part of the EA com­mu­nity and have played a cen­tral role in helping the com­mu­nity achieve many of its aims. They help di­rect funds to effec­tive causes, in­fluence peo­ple’s ca­reer tra­jec­to­ries (in­clud­ing plac­ing peo­ple in high-im­pact or­gani­sa­tions), con­tribute to the in­tel­lec­tual progress of the com­mu­nity, and help shape EA cul­ture.

The goal of Com­mu­nity Build­ing Grants is to scale the im­pact of high-po­ten­tial effec­tive al­tru­ism groups by in­creas­ing their or­gani­sa­tional ca­pac­ity. We ex­pect this to have a num­ber of benefi­cial effects, but we see a key benefit be­ing the em­pow­er­ment of group mem­bers to work on the world’s most im­por­tant prob­lems. We also ex­pect Com­mu­nity Build­ing Grants to en­able group or­ganisers to bet­ter co­or­di­nate and sup­port each other, to pur­sue valuable pro­jects out­side the scope of EA com­mu­nity build­ing, and to more more effec­tively dis­sem­i­nate a high fidelity un­der­stand­ing of effec­tive al­tru­ism.

This is a new pro­ject for CEA and can be seen as a way of test­ing the hy­poth­e­sised route to im­pact of pro­vid­ing sig­nifi­cant fund­ing to effec­tive al­tru­ism groups. If we judge EA Com­mu­nity Grants to have been suc­cess­ful, we in­tend to ex­pand the pro­ject and the amount of fund­ing al­lo­cated to it. Be­cause the pro­ject is new, we’re also in the pro­cess of re­solv­ing any kinks and ex­pect to con­tinue to tweak the pro­cess as we learn more about how to ex­e­cute it effec­tively.

To read more about EA Com­mu­nity Build­ing Grants and the core ra­tio­nale for the pro­ject, see this launch post.

Ap­pli­ca­tion Process

The EA Com­mu­nity Grants were ad­ver­tised via the EA Fo­rum in March, in the grants an­nounce­ment post, with a link to the ap­pli­ca­tion form. Ap­pli­cants could ap­ply as in­di­vi­d­u­als or with a joint ap­pli­ca­tion. Ap­pli­cants who passed the first in­ter­view stage were then in­ter­viewed by Harri Besceli, with Kerry Vaughan mon­i­tor­ing the in­ter­view pro­cess dur­ing April and May. After the in­ter­views, fi­nal de­ci­sions were made by June. Ap­pli­cants also nom­i­nated refer­ees who were asked to provide a refer­ence in bor­der­line cases.

Eval­u­a­tion Cri­te­ria

The three main crite­ria used to as­sess ap­pli­ca­tions were:

  • Abil­ity and skill set of ap­pli­cant

  • Ap­pli­cant un­der­stand­ing of and en­gage­ment with effec­tive al­tru­ism and effec­tive al­tru­ism com­mu­nity build­ing

  • Fu­ture po­ten­tial of the effec­tive al­tru­ism group (where rele­vant)

Ap­pli­cants were scored on three crite­ria, and the scores were com­bined (us­ing equal weight­ing) to cre­ate an over­all rat­ing of the ap­pli­ca­tion. This acted as a key in­put to the fi­nal de­ci­sions but did not solely de­ter­mine the fi­nal de­ci­sions.

Grant Offers

Since launch­ing in March, 22 grant offers were made, for a to­tal of $623,000, with the sig­nifi­cant ma­jor­ity of grants offers be­ing to en­able group or­ganisers to work full- or part-time on or­ganis­ing their effec­tive al­tru­ism group.

Of the 22 offers made, 18 are to fund in­di­vi­d­u­als di­rectly work­ing on their effec­tive al­tru­ism group, and 4 were for gen­eral ex­penses for groups or spe­cific pro­jects. The table be­low lists the grants made, the num­ber of peo­ple who will be work­ing part- or full-time as a re­sult of the grants, and the length of the grant pe­riod. The ma­jor­ity of the grant pe­ri­ods started at the be­gin­ning of this month. A small num­ber of ap­pli­ca­tions to EA Com­mu­nity Grants were referred to EA Grants for eval­u­a­tion, and aren’t in­cluded be­low.

We ul­ti­mately granted sig­nifi­cantly more fund­ing than we ex­pected to. This was largely due to the qual­ity of ap­pli­cants ex­ceed­ing our ex­pec­ta­tions, but also be­cause we con­cluded that the po­ten­tial for harm­ful in­di­rect effects was smaller than we pre­vi­ously thought.

Support

Re­cip­i­ents of EA com­mu­nity grants will largely work in­de­pen­dently from and au­tonomously to CEA. The main sup­port offered is reg­u­lar calls with CEA staff, a pri­vate Slack chan­nel, and a re­treat which took place in Septem­ber (hosted and co-or­ganised by the Czech As­so­ci­a­tion for Effec­tive Altru­ism). The struc­ture and ex­tent of sup­port for grantees is some­thing that we in­tend to ex­per­i­ment with sig­nifi­cantly over the du­ra­tion of the grant pe­riod.

Evaluation

Eval­u­at­ing the im­pact of groups is challeng­ing, be­cause by they provide a com­plex mix of di­rect and in­di­rect benefits to both in­di­vi­d­u­als and the wider EA com­mu­nity. Our goals in de­sign­ing an eval­u­a­tion pro­cess for EA com­mu­nity grants were to:

  • Give groups clear guidance on what would cause us to eval­u­ate their ac­tivi­ties favourably.

  • Make it easy for the EA com­mu­nity and po­ten­tial fun­ders to un­der­stand and eval­u­ate the suc­cess of EA Com­mu­nity Grants.

  • Provide suffi­cient ev­i­dence of the value pro­duced to en­able CEA to make well-in­formed de­ci­sions on whether to re­new fund­ing for given groups, and whether to scale the EA Com­mu­nity Grants pro­cess.

  • Avoid in­cen­tivis­ing groups to op­ti­mise for our met­rics rather than for what is ac­tu­ally high­est-im­pact.

  • Min­imise the time cost to CEA and to groups in eval­u­at­ing their re­sults.

On the ba­sis of these crite­ria, we’ve cho­sen to eval­u­ate the suc­cess of grants pri­mar­ily by as­sess­ing out­comes per­tain­ing to in­fluenc­ing the ca­reer tra­jec­to­ries of their mem­bers, but also by tak­ing into ac­count how groups are good rep­re­sen­ta­tives of the EA com­mu­nity and by eval­u­at­ing other valuable out­comes on a case-by-case ba­sis.

The pri­mary met­ric used to as­sess grants at the end of the first year is the num­ber of group mem­bers who ap­ply for in­tern­ships or grad­u­ate pro­grams in pri­or­ity ar­eas and reach at least the in­ter­view stage. For sub­se­quent years, the pri­mary met­ric used to as­sess the grants is the num­ber of group mem­bers who go on to work in pri­or­ity ar­eas.

We have cho­sen to pri­ori­tise ca­reer-re­lated out­comes be­cause we be­lieve that this will track a key source of value of the com­mu­nity grants and be­cause we feel like we have a rea­son­able un­der­stand­ing of the value of these kinds of out­comes. Though ca­reer-re­lated out­comes will be a use­ful proxy for suc­cess, it’s worth em­pha­sis­ing that we don’t ex­pect them to cap­ture all of the value that the groups pro­duce. How­ever, we feel it is im­por­tant to have some con­crete met­rics for our pro­jects, es­pe­cially ones that re­quire a lot of peo­ple’s time and money.

We used the 80,000 Hours list of pri­or­ity paths as the ba­sis for our list of ac­cred­ited roles, but ex­panded it to be some­what broader. The ar­eas and roles that we in­tend to ac­credit are still be­ing de­cided upon, and we ex­pect the num­ber of ac­cred­ited roles and ar­eas to in­crease in the fu­ture. We’ve cho­sen a rel­a­tively re­stricted set of crite­ria for the time be­ing, as we think the costs to later re­strict­ing the crite­ria will be sig­nifi­cantly higher than the costs of ex­pand­ing them.

A gen­eral ex­pec­ta­tion of all grantees and a nec­es­sary con­di­tion for re­ceiv­ing fur­ther fund­ing is that they abide by CEA’s Guid­ing Prin­ci­ples of Effec­tive Altru­ism, and are good rep­re­sen­ta­tives of the EA com­mu­nity. By this, we mean that we ex­pect the in­di­rect effects of grantees to the com­mu­nity to be pos­i­tive, both by af­fect­ing the pub­lic per­cep­tion of effec­tive al­tru­ism, and the cul­tural norms of the effec­tive al­tru­ism com­mu­nity.

Beyond ca­reer-re­lated out­comes, we will en­courage grantees to sub­mit other out­comes that they’ve pro­duced, which we can eval­u­ate on a case-by-case ba­sis. Such out­comes could in­clude ca­reer-re­lated out­comes not pre­vi­ously speci­fied, or non-ca­reer re­lated out­comes, such as sup­port­ing other EA Groups, in­fluenc­ing dona­tions, etc. This is in part in­tended to min­imise the po­ten­tial of cre­at­ing ad­verse in­cen­tives, and to help us bet­ter un­der­stand the value cre­ated by grantees for fu­ture eval­u­a­tion rounds.

We are aware of how fo­cus­ing on ca­reer-re­lated out­comes could have nega­tive effects. One way this could hap­pen is by dis­in­cen­tivis­ing valuable ac­tivi­ties be­cause they don’t count to­wards the suc­cess crite­ria. This could in­clude:

  • Hav­ing peo­ple en­ter the pri­or­ity area where they will have the high­est ex­pected im­pact (rather than the one where they have the high­est chance of ‘suc­cess’);

  • Helping pro­mote good epistemic norms and con­sid­er­ate cul­ture so that peo­ple en­ter­ing these ar­eas will have larger pos­i­tive ex­ter­nal­ities than oth­er­wise;

  • Helping out other EA groups, with in­for­ma­tion and shared best prac­tices.

Another way this could hap­pen is in­cen­tivis­ing sub­op­ti­mal or harm­ful ac­tivi­ties be­cause they do count to­wards the suc­cess crite­ria. For ex­am­ple:

  • Peo­ple push­ing group mem­bers to ap­ply for the pri­or­ity po­si­tions even when it isn’t right for them;

  • The use of po­ten­tially harm­ful strate­gies such as large-scale out­reach in or­der to achieve the suc­cess con­di­tions;

  • Dis­tort­ing think­ing of the grantees by giv­ing them in­cen­tive to be­lieve this is just the cor­rect thing to do and not ques­tion it too closely.

We hope to avoid these nega­tive effects by pro­vid­ing af­ter-the-fact as­sign­ment of credit for out­comes out­side the scope of the pri­mary suc­cess crite­ria, and by em­pha­sis­ing the crite­rion of ‘be­ing a good rep­re­sen­ta­tive of EA’.

We recog­nise that these mea­sures may be im­perfect, and may seek to make im­prove­ments to fu­ture eval­u­a­tion crite­ria based on our mid-term re­view of cur­rent grantees’ re­ported out­comes.

Fu­ture Plans

CEA will run an­other, smaller, EA Com­mu­nity Grants round in Jan­uary 2019, and in the sum­mer of 2019, when the suc­cess of the first com­mu­nity grants will also be eval­u­ated.

Prior to the sum­mer 2019 fund­ing round, we’ll eval­u­ate the grants pro­vided so far. If we judge spe­cific EA Com­mu­nity Grants to be suc­cess­ful, we’ll re­new the fund­ing pro­vided to those groups in the fund­ing round, and for groups we judge to have been par­tic­u­larly suc­cess­ful, we’ll provide grants for time pe­ri­ods longer than a year. Ad­di­tion­ally, if we judge the pro­cess as a whole to have been suc­cess­ful, we’ll ex­pand the num­ber of groups we fund.

Given that EA Com­mu­nity Grants is in its first iter­a­tion, it is difficult to make strong pre­dic­tions, as its suc­cess cur­rently re­mains to be seen. How­ever, one vi­sion of the EA com­mu­nity build­ing land­scape for within a few years’ time could in­volve:

  • The equiv­a­lent of 50 full-time peo­ple do­ing lo­cal EA com­mu­nity building

  • Full-time EA com­mu­nity builders in the ma­jor­ity of (but not re­stricted to) ~15 ma­jor uni­ver­si­ties and cities in Europe and North America

  • 5 na­tional or re­gional EA or­gani­sa­tions, em­ploy­ing a hand­ful of peo­ple work­ing on both EA com­mu­nity build­ing and ‘di­rect’ pro­jects

  • 5 peo­ple work­ing on sup­port­ing and co­or­di­nat­ing EA Com­mu­nity Gran­tees, pro­vid­ing a com­bi­na­tion of re­treats, con­fer­ences, train­ing, men­tor­ship, on­line in­fras­truc­ture etc.

  • 100+ peo­ple go­ing into high pri­or­ity po­si­tions each year from EA groups with a full-time coordinator

If you’d like to sup­port EA Com­mu­nity Build­ing Grants, you can do so here. Ad­di­tional fund­ing would likely in­crease our will­ing­ness to fund more groups in the Jan­uary fund­ing round, al­though this de­pends on the ap­pli­cant pool.

If you have any ques­tions re­lat­ing to EA Com­mu­nity Grants, please con­tact groups@effec­tivealtru­ism.org.