Which Community Building Projects Get Funded?


Dis­clo­sure: This anal­y­sis is mine alone, and does not re­flect the be­lief of any in­di­vi­d­u­als or or­ga­ni­za­tions I’m af­fili­ated with. To pro­tect my anonymity I’m not pro­vid­ing a full his­tory of my ex­pe­riences with the pro­grams I’ve an­a­lyzed, but read­ers can as­sume that I’ve un­suc­cess­fully at­tempted to get fund­ing through one or more of the chan­nels dis­cussed be­low and could po­ten­tially benefit from changes to these grant­mak­ing pro­cesses.

Most Effec­tive Altru­ists (my­self in­cluded) would agree that the stronger the EA com­mu­nity is, the more good it will be able to ac­com­plish. This idea has mo­ti­vated the de­vel­op­ment of mul­ti­ple pro­grams that have granted mil­lions of dol­lars to dozens of or­ga­ni­za­tions, groups, and in­di­vi­d­u­als work­ing on “com­mu­nity build­ing” (CB).

This anal­y­sis ex­am­ines the pro­cesses and grant his­tory of three grant­mak­ing chan­nels: EA Com­mu­nity Build­ing Grants (CBGs), EA Grants, and the EA Meta Fund. Eval­u­at­ing the effi­cacy of spe­cific grantees or grant­mak­ers is not fea­si­ble given the large num­ber of grants, their broad scope, and the lack of pub­li­cly available in­for­ma­tion about many of the pro­jects that have been funded. While I don’t want to be crit­i­cal of any par­tic­u­lar grantee, grant­maker, or plat­form, I do want to call at­ten­tion to con­cern­ing pat­terns re­vealed by a meta-anal­y­sis of these grant­mak­ing chan­nels in ag­gre­gate.

Below, I show that all three of these grant­mak­ing chan­nels use pro­cesses that im­plic­itly or ex­plic­itly re­strict the num­ber and type of CB pro­jects that are con­sid­ered. The out­come of these pro­cesses have been grants that pri­mar­ily fund CB pro­jects in close ge­o­graph­i­cal prox­im­ity to the grant­mak­ers. Just over half of CB fund­ing has gone to peo­ple or pro­jects closely con­nected to “Oxbridge” (Oxford or Cam­bridge) or Lon­don, and 85% of fund­ing has gone to Euro­pean efforts.

As dis­cussed in more de­tail in the sec­tion on the Meta Fund, this con­cen­tra­tion isn’t sim­ply a re­sult of the largest CB or­ga­ni­za­tions be­ing based in Oxford or Lon­don. Nor is it the re­sult of the EA com­mu­nity it­self be­ing ge­o­graph­i­cally con­cen­trated. While the com­mu­nity does have some large hubs, in­di­vi­d­ual EAs (per the EA Sur­vey) and EA groups (per EA Hub) are sig­nifi­cantly more dis­persed than CB fund­ing.

The land­scape for CB fund­ing is shift­ing rapidly. There’s been some turnover in the lead­er­ship of spe­cific fund­ing chan­nels, CEA re­cently an­nounced a new CEO, and the EA Grants pro­gram might be dis­con­tinued al­to­gether. This cre­ates an op­por­tu­nity to re­flect on how CB fund­ing can best be struc­tured go­ing for­ward. I hope my anal­y­sis helps ground this dis­cus­sion in hard data and high­lights some of the is­sues that new pro­cesses should try to re­solve. If my find­ings are cor­rect, there are likely valuable CB pro­jects from cer­tain ge­o­graphic re­gions and/​or in­ter­per­sonal net­works that are be­ing ne­glected.

Note: It’s clear which lo­ca­tions are funded by EA Com­mu­nity Build­ing Grants. How­ever, more in­ter­pre­ta­tion is re­quired for the Meta Fund and EA Grants, as many grantees be­long to mul­ti­ple ge­o­graphic net­works. To high­light the pat­terns I’ve ob­served, the pre­ced­ing table and sub­se­quent analy­ses at­tribute grants to the Oxbridge/​Lon­don net­work where grantees have, or pre­vi­ously had, strong ties to that area. For in­stance, a Meta Fund grant al­low­ing an Oxford grad­u­ate to study at Har­vard is at­tributed to the Oxbridge/​Lon­don area, rather than the Rest of World or a split be­tween Oxbridge/​Lon­don and the Rest of World. Similarly, the Other Europe cat­e­gory is best thought of as “tied to Europe (but not Oxbridge/​Lon­don)”, the Bay Area cat­e­gory as “tied to Bay (but not Europe)”, and the Rest of World as “no ties to Europe or Bay”. All calcu­la­tions and cat­e­go­riza­tions can be found here.

EA Com­mu­nity Build­ing Grants

Back­ground and process

The EA Com­mu­nity Build­ing Grant (CBG) pro­gram funds paid staff for lo­cal EA Groups. It is run by CEA’s Oxford-based groups team, and an Oxford grad is the main point per­son. (I’m not us­ing names through­out this re­port, as I don’t want to sug­gest that any in­di­vi­d­u­als are do­ing any­thing wrong. As men­tioned ear­lier, I’m try­ing to call at­ten­tion to broader pat­terns that are not the re­spon­si­bil­ity of any one per­son.)

The availa­bil­ity of CBGs was an­nounced via one post on the EA Fo­rum and one post in the EA Group Or­ga­niz­ers Face­book group, but was not pub­li­cized on the main EA Face­book page or in the EA Newslet­ter. This limited pub­lic­ity ap­pears to have trans­lated into limited aware­ness of the pro­gram, even in one of the EA com­mu­nity’s most ac­tive and con­nected hubs: af­ter the win­ners of the first round of grants were an­nounced, one well-con­nected EA re­ported that “When I spoke to ~3 peo­ple about it in the Bay, none of them knew the grant ex­isted or that there was an op­tion for them to work on com­mu­nity build­ing in the bay full time.”

The main an­nounce­ment of the sec­ond round of CBGs was made on the EA Fo­rum on Jan­uary 31, 2019, less than three weeks be­fore the Fe­bru­ary 18 ap­pli­ca­tion dead­line. Less than a week be­fore the ap­pli­ca­tions closed, it was also pub­li­cized through a Fe­bru­ary 13th post on the EA Group Or­ga­niz­ers Face­book group. The an­nounce­ments warned po­ten­tial ap­pli­cants that “this round is for highly promis­ing, time-sen­si­tive op­por­tu­ni­ties that we’d be will­ing to com­mit to be­fore the com­ple­tion of the im­pact eval­u­a­tion. Given the above, the bar for suc­cess­ful ap­pli­ca­tions will be higher than the pre­vi­ous round.” Ac­cord­ingly, the an­nounce­ment in­di­cated that the grant­mak­ing round would have “a bud­get cap of $150,000.” Pre­sum­ably more groups would have been in­clined to ap­ply for fund­ing had they known that in ac­tu­al­ity that round would grant $463,000. (After see­ing a pre­limi­nary ver­sion of this anal­y­sis, CEA clar­ified in a pri­vate com­mu­ni­ca­tion that “$150k was the amount to be spent on *new* grants, though this wasn’t ex­plic­itly stated in the post, and so it’s seems plau­si­ble that po­ten­tial ap­pli­cants would have in­ter­preted this as cov­er­ing both new grants and re­newals.”)

Most re­cently, in Oc­to­ber 2019 CEA an­nounced that it would ac­cept ap­pli­ca­tions on a rol­ling ba­sis. This an­nounce­ment also em­pha­sized that limited fund­ing would be available:

“The bar for suc­cess­ful ap­pli­ca­tions to EA Com­mu­nity Build­ing Grants has risen since our first round of grants in 2018. This means that we ex­pect to be mak­ing a smaller num­ber of new grants and pro­vid­ing less to­tal fund­ing for new grants than pre­vi­ously. Our cur­rent best guess is that we’ll offer be­tween $50,000 and $100,000 in new grants be­tween now and the end of 2019, though these aren’t strict up­per and lower bounds and it’s plau­si­ble that the amount we end up grant­ing lies out­side of this range. ‘New grants’ here does not in­clude re­newed fund­ing for grants made dur­ing pre­vi­ous rounds.”

Grant outcomes

CEA has made ~$1.4 mil­lion in CBGs. While it hasn’t dis­closed how much money went to each group, CEA has re­ported which groups got fund­ing, for how long, and for how many em­ploy­ees. This data shows that CBGs have over­whelm­ingly gone to fund staff for Euro­pean groups, which have re­ceived 89.5% of staff, in­clud­ing 24% for groups based in Oxford, Cam­bridge, or Lon­don. By con­trast, North Amer­i­can groups and Aus­tralian groups re­ceived 9.6% and .8% re­spec­tively. Groups based in Geneva (15.3%) and Oxford (12.2%) each re­ceived fund­ing for more staff than all non-Euro­pean groups com­bined (10.5%); groups in Nor­way (10.2%) and the Czech Repub­lic (9.8%) also came close.

Many, if not most, of the grantees from the lightly pub­li­cized first round ap­pear to have been groups that par­ti­ci­pated in a CEA-run “EA Com­mu­nity Build­ing re­treat in the UK for or­ga­niz­ers of Euro­pean EA groups.” In that round, 91% of funded staff po­si­tions went to Euro­pean groups, in­clud­ing all the grants for 12 month (vs. 6 month) po­si­tions and all the grants for full-time staff.

This Europe-cen­tric grant­mak­ing con­trasts with the ge­o­graph­i­cal dis­tri­bu­tion of EA Groups, where Euro­pean groups ac­count for only a mod­est plu­ral­ity.

Note: When grant re­newals were an­nounced in Oc­to­ber 2019 a to­tal dol­lar amount for those re­newals was not in­cluded, so my figures as­sume these grants pro­vided the same dol­lars per full time staff mem­ber as pre­vi­ous rounds. My anal­y­sis does not in­clude smaller (<$5,000) “Gen­eral Sup­port” grants that some groups have re­ceived, but which are not con­sid­ered part of the CBG pro­gram. The first four of these gen­eral grants (to a Cam­bridge sum­mer pro­gram for math stu­dents, Oxford, the Bay Area Biose­cu­rity group, and In­dia) were an­nounced in Novem­ber 2018. In a pri­vate mes­sage, CEA pro­vided ad­di­tional in­for­ma­tion on gen­eral grants made “from Fe­bru­ary 2019 on­wards.” Th­ese gen­eral grants were less con­cen­trated than CBG (Europe re­ceived 21% of gen­eral fund­ing), though they also pro­vided much less fund­ing (~$80,000 in gen­eral fund­ing vs. ~$1.4 mil­lion in CBG).

EA Grants

Back­ground and process

The EA Grants pro­gram has pri­mar­ily been run by Bay Area-based staff at CEA US, though the ini­tial round also in­volved an Oxford-based em­ployee of CEA UK. The first round of this pro­gram dis­tributed ~$480,000 in 2017. This anal­y­sis will fo­cus on grants to pro­jects CEA clas­sified as “EA Com­mu­nity”, which re­ceived 65% of all fund­ing. (Of the re­main­der, 33% went to “Long-term fu­ture”, while “An­i­mal Welfare” and “Global Health and Devel­op­ment” pro­jects each re­ceived 1%.)

The first round of EA Grants had “con­fus­ing” com­mu­ni­ca­tion and ac­cepted ap­pli­ca­tions for only three weeks, but still at­tracted 722 ap­pli­cants. How­ever, the eval­u­a­tion team re­ported that “Many ap­pli­cants had pro­pos­als for stud­ies and char­i­ties we felt un­der-qual­ified to as­sess. Most of those ap­pli­cants we turned down.”

While CEA pub­li­cly an­nounced a $2.6 mil­lion EA Grants pro­gram with rol­ling ap­pli­ca­tions in De­cem­ber 2017 (and sub­se­quently re­it­er­ated those plans in mid-Fe­bru­ary 2018, April 2018, mid-Au­gust 2018, and late Au­gust 2018), those plans never came to fruition. In­stead, CEA ran a “refer­ral round” of EA Grants (start­ing in Jan­uary 2018), and a Septem­ber 2018 round.

The refer­ral round was, by its na­ture, non-pub­lic. While CEA has pub­li­cly de­scribed this as a “smaller” round, it in fact granted ~80% more money than the origi­nal round. When CEA re-opened EA Grants to pub­lic ap­pli­ca­tions in Septem­ber 2018, they ac­cepted ap­pli­ca­tions for less than a month and since they be­gan “eval­u­at­ing ap­pli­ca­tions be­fore the ap­pli­ca­tion pe­riod ends, [it was] ad­van­ta­geous to sub­mit your ap­pli­ca­tion sooner rather than later.” The Septem­ber round was pub­li­cized in nei­ther the EA Face­book group nor the EA newslet­ter. Even the main EA Grants page wasn’t up­dated when ap­pli­ca­tions opened, so any­one vis­it­ing that page would have been un­der the im­pres­sion that ap­pli­ca­tions were still closed.

Grant outcomes

Look­ing at which “EA Com­mu­nity” pro­jects re­ceived fund­ing, a clear pat­tern emerges. The vast ma­jor­ity of money flowed to Bay Area-based ra­tio­nal­ity pro­jects (56%) or Oxford-con­nected philos­o­phy pro­jects (27%). The largest grant (ac­count­ing for nearly a quar­ter of all EA Com­mu­nity fund­ing) went to a pro­ject run by CEA US’s former Direc­tor of Strat­egy.

The refer­ral round made “22 grants for a to­tal of ~$850,000”; cor­re­spond­ing num­bers for the Septem­ber 2018 round haven’t been made available. While CEA hasn’t pub­lished the grantees for ei­ther of these rounds, they have pro­vided a cou­ple of ex­am­ples (a more com­pre­hen­sive eval­u­a­tion is ap­par­ently in progress). This in­for­ma­tion, along with a few grantees named in other con­texts (such as here and here), pro­vides a (very) limited pic­ture of where this money has gone. This data sug­gests that the refer­ral round re­lied heav­ily on Oxford (and to a lesser ex­tent UK and Bay Area) re­lated net­works.

EA Meta Fund

Back­ground and process

The origi­nal man­ager of this fund is based in the Bay Area but has nu­mer­ous and long­stand­ing Oxford ties. Un­der this man­age­ment, the Meta Fund did not ac­cept ap­pli­ca­tions.

After con­cerns raised by the com­mu­nity, the fund adopted a man­age­ment team in Oc­to­ber 2018. Since then, 4 of the 5 fund man­agers (in­clud­ing the Chair) have been based in Lon­don, with the 5th based in Hong Kong. (In June 2019, the fund re­moved one Lon­don based man­ager and re­placed them with an­other.)

Un­der the new man­age­ment sys­tem, the Meta Fund wel­comes ap­pli­ca­tions for fund­ing. How­ever, de­spite mul­ti­ple com­mu­nity re­quests, the Meta Fund’s web­site does not in­di­cate that this is the case or provide a link to ap­ply (un­like the two other EA Funds that ac­cept ap­pli­ca­tions and the global health fund which ex­plic­itly states that it does not ac­cept ap­pli­ca­tions.)

In­stead, the Meta Fund has so­lic­ited ap­pli­ca­tions through posts on the EA Fo­rum (which are tagged as “Com­mu­nity” posts, mean­ing they do not show up on the Fo­rum’s front page). Per the 2018 EA Sur­vey, only ~20% of the EA com­mu­nity uses the Fo­rum. By com­par­i­son, 53% of re­spon­dents are in the EA Face­book group. How­ever, ap­pli­ca­tions for the Meta Fund have only been so­lic­ited once in the Face­book group, less than 24 hours be­fore that round of ap­pli­ca­tions closed. The Meta Fund has never so­lic­ited ap­pli­ca­tions through the EA Newslet­ter.

Grant outcomes

Since its in­cep­tion, the Meta Fund has pre­dom­i­nantly sup­ported or­ga­ni­za­tions/​in­di­vi­d­u­als based in Oxbridge or Lon­don, and to a lesser ex­tent other parts of Europe. EAs based in other ar­eas have re­ceived min­i­mal fund­ing.

The con­cen­tra­tion in grants isn’t just a func­tion of the biggest meta or­ga­ni­za­tions (CEA, 80K, Founders Pledge) be­ing based near Lon­don. As the fol­low­ing table shows, even ex­clud­ing ~$1.1 mil­lion in grants to those or­ga­ni­za­tions the al­lo­ca­tions to Lon­don and Europe are no­tably high. Non-Euro­pean grants are mod­est (24%) and highly con­cen­trated in Van­cou­ver and the Bay. Edit: This per­spec­tive is also helpful for read­ers who are re­luc­tant to clas­sify CEA and 80,000 Hours as purely Oxford based or­ga­ni­za­tions. CEA has its team split be­tween Oxford and the Bay Area, and 80K was in the US (Bay Area) from Novem­ber 2016 - April 2019.

Note: The Meta Fund’s man­agers re­viewed a pre­limi­nary ver­sion of this anal­y­sis, and ques­tioned whether it was ap­pro­pri­ate to clas­sify re­search grants (to GPI, Forethought Foun­da­tion, and two in­di­vi­d­ual re­searchers) as “com­mu­nity build­ing.” I think this is a bor­der­line case, that re­flects some of the am­bi­guity sur­round­ing what counts as “com­mu­nity build­ing”, what counts as “meta”, and to what ex­tent these cat­e­gories over­lap. I took the ap­proach of in­clud­ing all the Meta Fund’s grants in my anal­y­sis. One could ar­gue that this over­states the amount of CB fund­ing al­lo­cated to Oxford, but it does cap­ture the im­por­tant point that these grants to Oxford re­searchers re­duce the amount of fund­ing available for other CB pro­jects.

Conclusion

Roughly 5 out of 6 dol­lars spent on CB has funded Euro­pean efforts. How should we in­ter­pret this de­gree of con­cen­tra­tion?

One hy­poth­e­sis is that this is an effi­cient al­lo­ca­tion of re­sources. If this were true, it would sug­gest that non-Euro­pean CB pro­jects are sys­tem­at­i­cally less valuable than their Euro­pean coun­ter­parts, and that Euro­pean pro­jects with­out close Oxbridge or Lon­don-ties are sys­tem­at­i­cally less valuable than pro­jects based in those lo­ca­tions. Even if you be­lieve these im­pli­ca­tions are cur­rently true, wouldn’t you want to broaden the base of lo­ca­tions ca­pa­ble of pro­duc­ing highly valuable work? And wouldn’t fund­ing a broader set of CB pro­jects be a nat­u­ral way to do so?

An al­ter­na­tive hy­poth­e­sis, which I find im­mensely more likely, is that CB re­sources are be­ing al­lo­cated in­effi­ciently. Un­der this view, CB fund­ing is too re­li­ant on over­lap­ping grant­maker net­works. This sug­gests it would be benefi­cial to in­cor­po­rate grant­mak­ers with differ­ent net­works (such as by adding one or more North Amer­i­can based man­agers to the Meta Fund com­mit­tee), to adopt CB fund­ing mechanisms that are based less on peo­ple/​per­sonal re­la­tion­ships and more fo­cused on pro­cesses or pro­jects, and to ex­tend the ap­pli­ca­tion pe­ri­ods and in­crease pub­lic­ity for fu­ture grant­mak­ing rounds.

If CB grant­mak­ers are in fact ex­hibit­ing bias to­ward their peo­ple in their net­works, this does not in any way mean that they are act­ing in bad faith or im­prop­erly try­ing to ad­van­tage their friends. It just means they are act­ing ex­actly how we should ex­pect any hu­man to be­have. In-group fa­voritism is a well-known dy­namic, which is pre­cisely why it’s im­por­tant to de­sign sys­tems that limit it.

I want to em­pha­size that there’s noth­ing in­her­ently wrong with re­la­tion­ship-based grant­mak­ing. But I do think its prob­le­matic when all the grant­mak­ing chan­nels in a cause area use this strat­egy, par­tic­u­larly when the de­ci­sion mak­ers have over­lap­ping net­works. In the same vein, the peo­ple mak­ing the grants all have good qual­ifi­ca­tions for do­ing so; the prob­lem as I see it is that these grant­mak­ers don’t com­ple­ment each other as well as they could.

My find­ings largely sup­port the idea that EA is “vet­ting con­strained” (at least with re­spect to CB), and that ad­di­tional grant­mak­ing ca­pac­ity would be helpful. I sus­pect paid grant­mak­ers would be par­tic­u­larly helpful: while we should ap­plaud vol­un­teer grant­mak­ers (like the Meta Fund man­age­ment com­mit­tee) for their time and effort, there’s only so much ca­pac­ity they can provide. The po­ten­tial wind­down of EA Grants will only ex­ac­er­bate these vet­ting con­straints. And if this wind­down also re­duces the amount of CB fund­ing available, it will make it even harder for de­serv­ing pro­jects to get funded.

The pur­pose of the EA com­mu­nity is to do as much good as pos­si­ble. Im­prov­ing how that com­mu­nity is built will make it more effec­tive in pur­su­ing that goal.