CEA Mid-year update (2020)

Link post

We’d like to share an up­date on our re­cent progress.

In a pre­vi­ous post, we set out five goals for 2020:

  1. Devel­op­ing our strat­egy: We’re on track, with more work needed on met­rics and data.

  2. Nar­row­ing our scope by con­sid­er­ing spin­ning off EA Funds and Giv­ing What We Can: We’re slightly ahead of ex­pec­ta­tions; we hired lead­ers and set an ini­tial strate­gic di­rec­tion for each pro­ject.

  3. Ex­pand­ing group and com­mu­nity health sup­port: We’re some­what be­hind ex­pec­ta­tions.

  4. Im­prov­ing on­line dis­cus­sion: We’re some­what ahead of ex­pec­ta­tions. We learned a lot about how to run vir­tual events, and EA Fo­rum en­gage­ment has grown sig­nifi­cantly.

  5. Stream­lin­ing in­ter­nal col­lab­o­ra­tion and pro­cesses: We’re on track or ahead of ex­pec­ta­tions on fi­nances, cus­tomer re­la­tion­ship man­age­ment (CRM), and cul­ture. We’re be­hind on hiring and the Oxford office move.

Over­all, I think we’re mak­ing good progress to­ward our goals for the year.

Spe­cific progress:

  • We hired Jonas Vol­lmer to lead EA Funds, and Luke Free­man to lead Giv­ing What We Can. Jonas and Luke are ex­pe­rienced EAs with ex­ec­u­tive ex­pe­rience and ex­cel­lent skil­lsets for the roles. We plan to in­ves­ti­gate spin­ning off both or­gani­sa­tions from CEA in the next year or so, al­though we will con­tinue to sup­port them both op­er­a­tionally.

  • We hired Cather­ine Low as a con­trac­tor to im­prove sup­port for groups, and she’s had 60+ calls with or­ganisers who gave the calls an NPS of 70+, which is ex­cel­lent. We’ve also up­dated 16 re­sources for lo­cal groups, in­clud­ing an im­proved in­tro­duc­tory fel­low­ship cur­ricu­lum tem­plate, dis­cus­sion group guides, and ad­vice on how to host vir­tual events in light of COVID-19 re­stric­tions. Fur­ther progress here is a ma­jor fo­cus.

  • We re­newed sev­eral Com­mu­nity Build­ing Grants and made three new grants to EA Brown, EA NYC, and EA MIT.

  • Events col­lab­o­rated with EAGx or­ganisers from around the world to cre­ate EAGx Vir­tual. With over 1,400 at­ten­dees, it was the largest EA event ever held. Vir­tual events are un­likely to be as valuable as in-per­son events, but now that we’ve de­vel­oped the ca­pac­ity to hold vir­tual events, we will likely con­tinue to do so (in ad­di­tion to hold­ing live events), since they are less ex­pen­sive and more ac­cessible.

  • Our key Fo­rum met­ric (views of good posts) has dou­bled so far this year, mean­ing that more good con­tent is be­ing posted to the Fo­rum and more peo­ple are read­ing that con­tent. We also in­tro­duced highly-de­manded fea­tures like tags, and some core in­tro­duc­tory con­tent.

  • We’ve fo­cused our com­mu­nity health work on men­tor­ship and net­work­ing for un­der­rep­re­sented EA groups, and de­vel­op­ing long-term plans to fur­ther im­prove di­ver­sity, equity, and be­long­ing in the com­mu­nity.

I also feel that we’ve fixed a lot of in­ter­nal is­sues over the last 18 months and that we con­tinue to grad­u­ally im­prove. We have 13 months of run­way. Staff perfor­mance, morale, and re­ten­tion ap­pear to be solid and im­prov­ing. We’ve im­proved our ac­count­ing sys­tems, grant­mak­ing records, and HR com­pli­ance/​on­board­ing, and in­vested our re­serves to pro­tect against in­fla­tion.

Pro­gram updates


We’re in the early stages of im­prov­ing our sup­port for groups. This is a ma­jor fo­cus at the mo­ment.

Groups support

  • We are work­ing to im­prove the in­tro­duc­tory EA fel­low­ship tem­plate with Com­mu­nity Build­ing Grant re­cip­i­ents Huw Thomas, James Aung, and Alex Hol­ness-Tofts. In­tro­duc­tory EA fel­low­ships are eight-week read­ing and dis­cus­sion groups run on uni­ver­sity cam­puses. We es­ti­mate that ap­prox­i­mately 400 stu­dents par­ti­ci­pated in an in­tro fel­low­ship last year.

  • We brought on Cather­ine Low, who had been an ex­traor­di­nar­ily ac­tive vol­un­teer sup­porter of EA groups, as a con­trac­tor to im­prove our sup­port of groups. Cather­ine has had 60+ calls with group or­ganisers, which so far score an av­er­age Net Pro­moter Score of 70+, which is ex­cel­lent.

  • We’ve part­nered with or­ganisers to im­prove 16 re­sources on the EA Hub re­lated to start­ing and run­ning groups, in­clud­ing dis­cus­sion group guides and ad­vice on run­ning vir­tual events for groups tran­si­tion­ing to on­line mee­tups due to COVID-19.

  • We’ve helped to run three monthly vir­tual mee­tups (with at­ten­dance rang­ing from 40-70 or­ganisers) where or­ganisers made con­nec­tions across differ­ent ge­ogra­phies, ex­plored re­sources, and dis­cussed ways to im­prove their groups.

Com­mu­nity Build­ing Grants

We re­granted a to­tal of $325,000, in­clud­ing grants to the fol­low­ing:

  • Renewals

    • EA Lon­don: 1 FTE, 1 year

    • EA France: 1.2 FTE, 1 year

    • EA Toronto: 0.5 FTE, 1 year

    • EA Nor­way: 2 FTE, 1 year

    • EA Ber­lin: 0.5 FTE, 1 year

  • New Grants

    • EA Brown: 0.5 FTE, 6 months

    • EA MIT: 0.5 FTE, 3 months

    • EA NYC: 2 FTE, 1 year

We col­lab­o­rated with the cur­rent NYC or­ganisers to run an ap­pli­ca­tion round for an EA NYC Com­mu­nity Direc­tor. We were ex­cited by the qual­ity of the ap­pli­ca­tions we re­ceived, and de­cided to make offers to Arushi Gupta and Aaron Mayer to be EA NYC Co-Direc­tors (rather than just mak­ing one offer). This is the first time that we’ve col­lab­o­rated with a group to run an ap­pli­ca­tion round for a spe­cific lo­ca­tion. We’re pleased with how the pro­cess went, and it helped us fur­ther de­velop our pro­cesses for eval­u­at­ing grant ap­pli­ca­tions more gen­er­ally. We may want to run similar ap­pli­ca­tion rounds in the fu­ture — though it took a rel­a­tively long time, so we don’t ex­pect to be able to do this for many lo­ca­tions.


EA Global: Virtual

In re­sponse to COVID-19, we re­placed EA Global: San Fran­cisco 2020 with a vir­tual event.

As the event was sched­uled for March, just as the ex­tent of the coro­n­avirus epi­demic was be­com­ing clear, there was not much time to plan this tran­si­tion. I think the team did a good job an­ti­ci­pat­ing the change and adapt­ing.

We met about half of our goals for the event.

Some key stats:

  • Date: March 21-22, 2020

  • Grip (net­work­ing app): 456 ac­tive users

  • YouTube: 325 peak con­cur­rent views

  • Slack (group dis­cus­sion): 270 members

  • Like­li­hood to recom­mend: 7.82 (av­er­age)

Goals we achieved:

  • Find three case stud­ies of strong im­pact from the conference

  • Im­prove wel­com­ing­ness (such that >80% of re­spon­dents agree that EAG is a place where they feel wel­come); 89% of sur­vey re­spon­dents re­ported feel­ing this way

Goals we didn’t achieve:

  • Aver­age at­tendee feels com­fortable reach­ing out to 5+ new people

    • New con­nec­tions per at­tendee (av­er­age): Vir­tual = 4.4 | Lon­don = 9.6

    • At­ten­dees with 5 or more new con­nec­tions: Vir­tual = 37% | Lon­don = 80%

    • See chart for more data

  • Main­tain or im­prove recom­men­da­tion scores com­pared to EA Global: Lon­don 2019 (see chart be­low for scores from pre­vi­ous events)

See more de­tails in the wrap re­port.

Vir­tual EAGx

In June, we col­lab­o­rated with EAGx or­ganisers from around the world to run the largest ever EA con­fer­ence, with over 1,400 at­ten­dees from 57 coun­tries. For many com­mu­nity mem­bers with­out a lo­cal group nearby, this was their first chance to at­tend an EA event and meet other EAs from across the world. The event re­sulted in 10,456 con­nec­tions and 2,495 meet­ings within the Grip net­work­ing app, and over 3,000 hours of con­tent views.

Some key stats:

  • Date: June 13-14, 2020

  • Grip (net­work­ing app): 1,389 ac­tive users

  • YouTube: 434 peak con­cur­rent views

  • Slack (group dis­cus­sion): 1,025 members

  • Like­li­hood to recom­mend: 8.44 (av­er­age)

This event’s “like­li­hood to recom­mend” score was a big im­prove­ment over our first vir­tual event, and was closer to the usual score for an in-per­son event (see chart above).

Although the pro­por­tion of at­ten­dees mak­ing im­por­tant con­nec­tions was lower than at pre­vi­ous events, 50% of sur­vey re­spon­dents re­ported mak­ing a minor plan change, and over a third changed their mind about some­thing.

We achieved al­most all of our goals for the event, in­clud­ing:

  • 1,000 EAs at­tend­ing over the course of the event (ac­tual num­ber: 1,424)

  • Fa­cil­i­tat­ing at least 700 one-on-one meet­ings (ac­tual num­ber: 2,495)

  • Achiev­ing an NPS over 30 (ac­tual num­ber: 43)

Sys­tems improvements

We shifted our ad­mis­sions sys­tem to new soft­ware, be­gan batch­ing re­sponses to ap­pli­ca­tions, and hired ex­tra ca­pac­ity to help with the ad­mis­sions pro­cess. This seems to be sav­ing time on ad­mis­sions.

It seems that vir­tual events are worth run­ning — they are nor­mally not quite as good as in-per­son events, but cheaper to do and eas­ier for peo­ple to ac­cess. We’re ten­ta­tively think­ing of con­tin­u­ing to run vir­tual events in ad­di­tion to in-per­son events in the fu­ture.


Met­ric progress

The Fo­rum’s key met­ric (num­ber of views of good posts) has more than dou­bled since the start of the year.

While we’ve spent a lot of time fine-tun­ing and val­i­dat­ing this met­ric, we want to flag that the defi­ni­tion of a “good post” is in­her­ently sub­jec­tive. Ex­pe­rienced EAs out­side of CEA have gen­er­ally agreed with our as­sess­ments when we’ve asked them to eval­u­ate posts, but we want to avoid draw­ing strong ev­i­dence of the Fo­rum’s suc­cess from this met­ric alone.

(Some ex­am­ples of “good posts” from this year in­clude writ­ings on ran­domista de­vel­op­ment, per­sonal cruxes on AI safety, and con­sid­er­ing a wider range of longter­mist jobs.)
We’ve also seen many other met­rics of Fo­rum ac­tivity in­crease since the start of this year. For ex­am­ple, views of all posts by logged-in users have roughly dou­bled, and the num­ber of posts with 2+ karma has in­creased by ~80%.

Tech­ni­cal progress

We con­tinue to port fea­tures across from LessWrong. We re­cently im­ported tags, ping­backs and link pre­views. (Tags are our most re­quested fea­ture ever.) We also con­tribute to LW’s code­base.

We also launched a se­quence of posts in­tro­duc­ing peo­ple to EA (this is a first draft — we wel­come feed­back via this form). We’ve be­gun to do more out­reach to po­ten­tial top posters, in­clud­ing pro­mo­tion at EAGxVir­tual, which re­sulted in our biggest spike in new users ever. We’ve also been try­ing to in­crease the Fo­rum’s visi­bil­ity on search en­g­ines (though we are strug­gling to make im­prove­ments, de­spite en­gag­ing SEO con­sul­tants).

We con­tinue to offer edit­ing for Fo­rum posts, and to run the Fo­rum prize, EA Newslet­ter, EA/​CEA so­cial me­dia, etc.

Key issues

Low satis­fac­tion scores: We asked users: “On a scale of 0 to 10, how likely would you be to recom­mend the Fo­rum to some­one who re­cently be­came in­ter­ested in effec­tive al­tru­ism?” The av­er­age re­sponse was 6.9. We think that this low score was partly be­cause peo­ple felt that the Fo­rum is rel­a­tively in­sider-y, and so not suit­able for new peo­ple. How­ever, we’d like to get those scores up.

De­mo­graph­ics: Ac­cord­ing to Google An­a­lyt­ics, 41% of Fo­rum view­ers are fe­male (higher than the pro­por­tion of com­mu­nity mem­bers who are fe­male), but we be­lieve* that males make up a higher pro­por­tion of au­thors than their pro­por­tion in the com­mu­nity. We are con­sid­er­ing how we could en­courage more women to post.

*Many au­thors are anony­mous, so we aren’t cer­tain of this.

Com­mu­nity health

The com­mu­nity health team tries to en­able EA’s long-term suc­cess by re­spond­ing to me­dia en­quiries, me­di­at­ing com­mu­nity dis­putes and com­plaints, and pro­vid­ing ad­vice and sup­port to lo­cal or­ganisers who are fac­ing par­tic­u­larly difficult prob­lems.

One of our goals is to be available to the com­mu­nity to help with our ar­eas of spe­cialty. Here’s the break­down of cases we’ve as­sisted with dur­ing 2020 so far:

  • 87 new me­dia re­quests or situations

  • 12 situ­a­tions or ques­tions on di­ver­sity, equity, and belonging

  • 7 situ­a­tions in­volv­ing harm­ful or un­eth­i­cal behavior

  • 6 per­sonal or men­tal health struggles

  • Around 60 more minor cases or requests

Some of our proac­tive work:

  • Sky May­hew joined the board of di­rec­tors of WANBAM. The data-shar­ing sys­tem she set up al­lows us to re­cruit WANBAM mem­bers as speak­ers for EAG and other events, meetup hosts, job ap­pli­cants, etc.

  • We’ve made some im­prove­ments in in­te­grat­ing di­ver­sity, equity, and be­long­ing intp our hiring pro­cess (e.g. adding re­lated skills in job de­scrip­tions for pub­lic-fac­ing roles, im­prov­ing sourc­ing) but still have out­stand­ing work re­lated to im­prov­ing our as­sess­ment skills in these ar­eas and ad­just­ing our score­card weight­ing.

  • We’ve been work­ing with En­com­pass and with CEA lead­er­ship to clar­ify our org-wide long-term goals with re­spect to di­ver­sity, equity, and be­long­ing.

  • We up­dated our page on di­ver­sity, equity, and be­long­ing, pub­lished an ac­com­pa­ny­ing blog post, and col­lected re­sources on racial jus­tice for the Fo­rum and the groups newslet­ter. Ju­lia Wise gave a talk on re­lated top­ics.

  • We funded Ewelina Tur for pro-bono psy­chother­apy for 140 free or dis­counted ses­sions she pro­vided to EAs over the past year.

  • We in­creased ca­pac­ity for the fu­ture by:

    • Hiring ex­tra ca­pac­ity to help with the ad­mis­sions pro­cess (~.3FTE /​ yr)

    • Clos­ing EA Grants (with some grant-mak­ing trans­ferred to EA Funds), to free up Ni­cole Ross’s time/​focus

    • Con­tract­ing and al­ign­ing with a com­mu­ni­ca­tions ad­vi­sor who can coach EAs be­fore me­dia interviews


Hiring and strategy

Jonas Vol­lmer has joined CEA as Head of EA Funds. Jonas was pre­vi­ously Co-Ex­ec­u­tive Direc­tor of the Cen­ter on Long-Term Risk (pre­vi­ously the EA Foun­da­tion). We were im­pressed by Jonas’ anal­y­sis of the strate­gic op­tions for Funds, and by his plans for ac­tive grant­mak­ing (that is, seek­ing out po­ten­tial grantees more proac­tively rather than only grant­ing to peo­ple who ap­plied for fund­ing).

If we think that Funds is mak­ing good progress, we’ll con­sider spin­ning off the pro­ject in the next year or so. Jonas would re­port to our board, but con­tinue to be sup­ported op­er­a­tionally by our team (like 80,000 Hours is).


Although we have been mostly fo­cused on strat­egy and hiring so far this year, cu­mu­la­tive dona­tions YTD to the four main funds are up about 36% (~$4.5m vs. ~$3.3m). In­clud­ing pass-through dona­tions (e.g. peo­ple donat­ing to MIRI), we’re at $5.9m com­pared to $4.1m last year and $5.6m the year be­fore. Note that dona­tions are very lumpy, so this could partly be ran­dom de­vi­a­tion.

Giv­ing What We Can (GWWC)


Luke Free­man just joined us to lead GWWC, with a view to spin­ning off the pro­ject from CEA.

Luke is a long-stand­ing GWWC mem­ber with a back­ground in star­tups, digi­tal mar­ket­ing, and lo­cal com­mu­nity build­ing in Aus­tralia. We’ve been im­pressed by his sup­port­ive and wel­com­ing ap­proach to com­mu­nity build­ing, and plans for spread­ing aware­ness of the pledge.

If we think that GWWC is mak­ing good progress, we’ll con­sider spin­ning off the pro­ject in the next six months or so. Luke would re­port to our board, but con­tinue to be sup­ported op­er­a­tionally by our team (like 80,000 Hours is).


New pledge rates have held rel­a­tively steady (around 10 per week).

As de­tailed in the an­nounce­ment post, Luke is look­ing for feed­back, vol­un­teers, mem­ber sto­ries, and in­tro­duc­tions, and has been hold­ing some on­line events for mem­bers.

Org-wide updates


At the mid-year point, we are 36% be­low bud­get, due to can­cel­ling in-per­son events such as EA Global, slower hiring, lower-than-an­ti­ci­pated salary in­creases, and com­mu­nity-build­ing grant­mak­ing fal­ling more in the sec­ond half of the year. Re­fore­cast­ing our spend to the year end sug­gests we will be 15% un­der bud­get for 2020 over­all.

Tak­ing the above into ac­count, we cur­rently have 13 months of run­way, pro­vid­ing fi­nan­cial se­cu­rity for CEA un­til around halfway through 2021.

We’ve been tidy­ing up our fi­nan­cial sys­tems and im­prov­ing our in­ter­nal re­port­ing to al­low more ac­cu­rate and faster fore­cast­ing. We’ve moved some of our funds to in­ter­est-bear­ing ac­counts, which yields around $3k/​week in in­ter­est pay­ments. We also im­ple­mented a grant man­age­ment sys­tem, which should al­low us to au­to­mate more of our grant­mak­ing, in­crease com­pli­ance, and im­prove mon­i­tor­ing and eval­u­a­tion.


We con­tinue to test and im­prove our strat­egy for the or­gani­sa­tion, in­clud­ing de­vel­op­ing met­rics that can track our progress. This is still a work in progress, so we’re not ready to share more pub­li­cly at the mo­ment. I’m over­all slightly dis­ap­pointed with the rate of progress here, and I’m aiming to fo­cus more on this in the fi­nal six months of the year.



I feel that staff morale is now solid and im­prov­ing. The data sup­port this view, but I think it’s ex­tremely hard to mea­sure this re­li­ably. (The lin­ear re­gres­sion over the last year is for self-re­ported morale to in­crease by an av­er­age of 0.71 points per year on a 10-point scale. This could partly be ex­plained by changes in the ques­tion we ask staff, or how peo­ple were ap­proach­ing the ques­tion, but I think this is at least partly a real effect.)


Re­ten­tion is the per­centage of staff at the be­gin­ning of the pe­riod who still work for the org at the end of the pe­riod.

For his­tor­i­cal rea­sons, we re­port this from Fe­bru­ary to Fe­bru­ary. Note that re­ten­tion and turnover figures for 2020 might change over the course of the year.

Con­clu­sion: progress so far

As I dis­cussed above, I think we’re mak­ing good progress to­ward our goals for the year. We plan to re­view our progress again at the end of the year.