Reflections on community building in the Netherlands

Authors: Rem­melt Ellen and Sjir Hoeij­mak­er­s

Here, we share our key les­sons since launch­ing the com­mu­nity-build­ing or­gani­sa­tion Effec­tive Altru­ism Nether­lands this year, as well as our plans for 2018. We want your feed­back: both in the com­ments be­low and if you see us at EAG Lon­don.

Below, we have split up our ac­tivi­ties for this year into rough phases:
Jan­uary: start­ing an or­gani­sa­tion
We reg­istered Effec­tive Altru­ism Nether­lands (EAN) as a char­ity in early Jan­uary with the sup­port of Robert and Kel­lie from Effec­tive Giv­ing. We formed an ex­ec­u­tive board and an ad­vi­sory board, and pro­ceeded to set up a bank ac­count, ap­plied for tax-de­ductibil­ity, set up in­ter­nal soft­ware tools, etc.

Fe­bru­ary: man­ag­ing pro­ject­s
In­spired by the de­cen­tral­ised, self-or­ganis­ing struc­tures of Re­think Char­ity and EA NTNU, we de­cided to fo­cus EAN on sup­port­ing pro­jects, based on ei­ther promis­ing ideas that we re­cruited vol­un­teers for or ini­ti­a­tives that emerged from the com­mu­nity. For each pro­ject, we in­tended to write out a frame­work be­fore­hand to set clear ex­pec­ta­tions on the scope of the prob­lem to work on, the means of com­mu­ni­ca­tion and crite­ria for re­view­ing pro­ject re­sults at the end.

EAN was already host­ing monthly strat­egy meet­ings with lo­cal group or­ganisers; to these we added pro­ject-or­ganiser meet­ings (at a co-work­ing hub we started col­lab­o­rat­ing with). We also con­tinued our web­site de­vel­op­ment and one-on-one com­mu­nity out­reach as pro­jects with our vol­un­teer team.

We started two out­reach pro­jects — a crowd-act­ing cam­paign and a page list­ing in an info book­let for fresh­ers’ fairs — and found en­ter­pris­ing peo­ple to lead those. There, we ran into prob­lems: we were act­ing on short-term op­por­tu­ni­ties, so it was difficult to find vol­un­teers deeply in­volved with EA who had the pre­req­ui­site mar­ket­ing skills, and there­fore for us to avoid micro­manag­ing their ex­e­cu­tion.

At this point, EAN was mostly work­ing with groups of stu­dents. Although sev­eral group or­ganisers de­cided to or­ganise monthly pub so­cials, we no­ticed that most of them lacked the time and/​or mo­ti­va­tion to use learned out­comes from meet­ings to de­velop their groups fur­ther.

This high­lighted the need to col­lab­o­rate more with peo­ple who ac­tively shared our goals and had more pro­fes­sional ex­pe­rience. Sjir came into con­tact with two ex­pe­rienced work­shop fa­cil­i­ta­tors and gave them the idea to start giv­ing ca­reer work­shops at uni­ver­si­ties based on the re­search of 80,000 Hours, start­ing with tri­als at stu­dent or­gani­sa­tions in April.

To offi­cially launch EAN, we or­ganised an event on 28 May in­tro­duced by Peter Singer (in the same pe­riod, we ar­ranged for him to at­tend a din­ner dis­cus­sion for large philan­thropists for Effec­tive Giv­ing and over­saw the me­dia pub­lic­ity around a new trans­lated ver­sion of The Most Good You Can Do). The event had 139 par­ti­ci­pants reg­istered – ei­ther through on­line ap­pli­ca­tions or through per­sonal in­vi­ta­tions by us be­cause of their po­si­tion in a cer­tain sec­tor and po­ten­tial for EA-al­igned­ness. We com­bined Peter Singer’s talk with par­allel ses­sions cater­ing to peo­ple work­ing in differ­ent sec­tors and cause ar­eas. This re­sulted in fruit­ful dis­cus­sions and an ex­cited at­mo­sphere (with an av­er­age feed­back score of 6.5/​8 on Has this made you more or less en­thu­si­as­tic about EA?). We over­saw a brain­storm ses­sion on EA pro­jects, but we saw few con­crete re­sults come out of that, prob­a­bly be­cause of the loose ses­sion struc­ture as well as par­ti­ci­pants di­verg­ing in their aims. In an­other ses­sion, we in­vited par­ti­ci­pants in­ter­ested in AI safety. One of them shared a re­cent idea to cre­ate an on­line course about AI safety (now called RAISE: Road to AI-Safety Ex­cel­lence) and was able to in­volve ca­pa­ble vol­un­teers to start work­ing on that.

The pro­duc­tive in­ter­ac­tions we no­ticed with the form­ing of cause-area/​tal­ent-area sub­groups at the event, as well as the lack of ‘spon­ta­neously-gen­er­ated’ vol­un­teer pro­jects led us to con­clude that we needed to ap­proach build­ing our com­mu­nity more as fa­cil­i­tat­ing spe­cial­ised net­works of peo­ple.
See our re­view of our pro­ject phase.

July: Build­ing net­work­s
We de­cided to test a novel strat­egy that in­volved sup­port­ing or­ganisers in cre­at­ing spe­cial­ised net­works of EAs to work on the most press­ing prob­lems. We would ar­range one-on-one prob­lem-solv­ing ses­sions with the or­ganisers, help con­nect suit­able peo­ple with their net­works and deeply en­gage their mem­bers by fa­cil­i­tat­ing in­ter­ac­tions be­tween them and the wider EA com­mu­nity.

We had the hy­poth­e­sis that we would be able to col­lab­o­rate with five net­works by Oc­to­ber, with each hav­ing clear goals and met­rics, but this didn’t hap­pen for sev­eral rea­sons:

In ad­di­tion to col­lab­o­rat­ing with the ca­reer work­shops fa­cil­i­ta­tors, Effec­tive Giv­ing and RAISE, we ex­plored a num­ber of po­ten­tial col­lab­o­ra­tions with in­di­vi­d­u­als: a pro­fes­sional-ser­vices net­work, a cor­po­rate-en­trepreneurs net­work, a policy net­work, the LessWrong Am­s­ter­dam com­mu­nity, an effec­tive an­i­mal ad­vo­cacy net­work and sev­eral lo­cal groups.
Although we helped a num­ber of these ini­ti­a­tives grow, this did not re­sult in any sig­nifi­cant col­lab­o­ra­tions. We ex­pe­rienced difficulty find­ing ca­pa­ble and al­igned or­ganisers who had the time to form net­works, as well as com­mu­ni­cat­ing a clear value pro­posal to ex­ist­ing or­ganisers.

Our ex­e­cu­tion of this strat­egy was fur­ther ham­pered by Sjir and Rem­melt tak­ing leave dur­ing this time and gen­er­ally spend­ing too much time on reach­ing con­sen­sus in de­ci­sions. Fi­nally, both Sjir and Rem­melt were still work­ing on a vol­un­tary ba­sis, mak­ing it difficult for them to perform deep, undis­tracted work.

EAN ap­plied for fund­ing from EA Grants at the end of June. The staff at the Cen­tre for Effec­tive Altru­ism de­cided to con­nect us to Open Philan­thropy Pro­ject at the end of Au­gust, who then deferred their de­ci­sion back to CEA, who in the end de­cided not to fund us.

CEA staff ex­pressed doubts about two ar­eas: strat­egy syn­chro­ni­sa­tion and our abil­ity to ex­e­cute strat­egy. Here, strat­egy syn­chro­ni­sa­tion was about the risks in ex­pand­ing the num­ber of large or­gani­sa­tions for them to co­or­di­nate with, given how easy it is to make mis­takes in pro­mot­ing effec­tive al­tru­ism (e.g. dilu­tion of ideas and, re­lat­edly, the Unilat­er­al­ist’s Curse). Given that our strat­egy was rather vague, it was difficult for them to gauge these risks. Also, we were ex­plor­ing un­charted ter­ri­tory by try­ing an ap­proach that re­lied on trust­ing other in­di­vi­d­u­als to build net­works with­out dilut­ing the com­plex­ity in­her­ent in striv­ing to do effec­tive al­tru­ism. In terms of ex­e­cu­tion, CEA staff were un­sure whether Sjir and Rem­melt were at least as ca­pa­ble as a re­cruit they would hire to work for them.

In the mean­time, we helped RAISE record their first lec­ture videos as well as nar­row their tar­get au­di­ence to those that seemed most com­mit­ted to and ca­pa­ble of be­com­ing AI-safety re­searchers.
For the ca­reer work­shop fa­cil­i­ta­tors, we con­tacted and found sev­eral uni­ver­sity col­leges to do paid work­shops at.
See our re­view of our net­work-build­ing phase.

Oc­to­ber: a shift to out­reach & events
Our biggest les­son in test­ing our net­work-build­ing strat­egy has been that the Dutch EA com­mu­nity still lacks highly-en­gaged and skil­led in­di­vi­d­u­als – those who could po­ten­tially join the spe­cial­ised net­works we had in mind. At the same time, we see a lot of un­re­al­ised po­ten­tial in those an­a­lyt­i­cal, al­tru­is­tic Dutch peo­ple who have yet to en­counter or en­gaged deeply with EA.

We have there­fore shifted to work­ing on a more tried-and-tested ap­proach with clear role di­vi­sions:
Sjir will fo­cus on do­ing tar­geted out­reach to ca­pa­ble peo­ple that share our goals, in­clud­ing by
giv­ing lec­tures on EA (as well as sup­port­ing ca­reer work­shops)
writ­ing mag­a­z­ine ar­ti­cles
rep­re­sent­ing EA(N) to the Dutch me­di­a
Sjir has some ex­pe­rience in these ar­eas (via his work on ba­sic in­come ex­per­i­ments).

See Sjir’s very rough im­pact es­ti­mates.

Rem­melt will fo­cus on co­or­di­nat­ing events to build up the ca­pac­i­ties of these peo­ple, in­clud­ing
• monthly EAN Com­mu­nity Events, start­ing on 12 Novem­ber
an ac­tive Sun­day af­ter­noon where par­ti­ci­pants can se­lect and solve prac­ti­cal EA prob­lems for 4 hours
• bi-an­nual week­end re­treats­
fo­cused on al­low­ing peo­ple we already col­lab­o­rate with to im­merse them­selves deeply in EA and how to strate­gi­cally lead ini­ti­a­tives, pos­si­bly in­clud­ing a Euro­pean EA Or­ganisers Re­treat in Fe­bru­ary.
• an an­nual EAGx Nether­lands, around May/​June­
fo­cused on en­abling new peo­ple to reach the fron­tiers of EA
Rem­melt has ex­pe­rience in man­ag­ing events (incl. mee­tups, 2 speaker tours, and a Hu­man­ist con­fer­ence & week­end). He aims to make each of EAN’s events in 2018 fund it­self through ticket fees.

Our cur­rent strat­egy ap­pears to put us closer to that of EA Lon­don as well as the be­gin­ning years of Effec­tive Altru­ism Foun­da­tion in Switzer­land (the size of the re­gion we’re effec­tively op­er­at­ing in also seems similar).
See our one-pager on our new ap­proach.

2018: what we’re plan­ning for
On sup­port­ing ini­ti­a­tives:
We still see ‘build­ing spe­cial­ised EA net­works’ as a pow­er­ful con­cept for putting peo­ple’s skills to use on the prob­lems where, af­ter due re­flec­tion, they cur­rently ex­pect to make the biggest col­lec­tive im­pact. How­ever, we now think that we should make the pipeline to these more grad­ual – by tar­get­ing and get­ting to know po­ten­tial col­lab­o­ra­tors, and de­sign­ing a so­cial en­vi­ron­ment that re­peat­edly ex­poses them to the nu­anced ideas of EA. In other words, we’ll be more of an ‘in­cu­ba­tor’ than a ‘con­nec­tor’ of ini­ti­a­tives.

Other meta-or­gani­sa­tions have shown suc­cess in in­cu­bat­ing pro­jects in the past, no­table ex­am­ples be­ing the Cen­tre for Effec­tive Altru­ism (with 80,000 Hours & Giv­ing What We Can), and Effec­tive Altru­ism Foun­da­tion (with Sen­tience Poli­tics & Rais­ing for Effec­tive Giv­ing). We will there­fore con­tinue to have one-on-one con­ver­sa­tions with or­ganisers we trust, in­clud­ing from Effec­tive Giv­ing, the ca­reer work­shops fa­cil­i­ta­tors, RAISE and EA Gron­ingen, and sup­port them in solv­ing any bot­tle­necks to build­ing out their net­works.

Sjir has ex­pe­rience in Dutch pub­lic policy and Rem­melt is mak­ing forays into ma­chine learn­ing right now. In 2018, one or both of them may there­fore de­cide to lead a fo­cus pro­ject of his own.

On fund­ing:
We see EAN as a long-term en­deav­our and – as­sum­ing that we will gain suffi­cient trac­tion – plan to ded­i­cate the next years on work­ing on this. We want to avoid our per­sonal in­comes be­ing de­pen­dent on ex­ter­nal fund­ing. In our ex­pe­rience, it has been highly dis­tract­ing to en­gage po­ten­tial donors on a com­pli­cated meta-char­ity to ex­tend our shrink­ing fi­nan­cial run­ways. We want to pre­vent be­ing forced to pre­ma­turely re­duce the time we can spend on com­mu­nity build­ing based on this alone (though we will ac­tively seek out feed­back from in­formed donors on our po­ten­tial rel­a­tive to other com­mu­nity-builders).

Sjir and Rem­melt can cover their ba­sic liv­ing costs un­til the end of Jan­uary 2018. We plan to tran­si­tion to other in­come sources be­fore that dead­line:
Sjir will re­ceive a one-time com­pen­sa­tion from EAN to cover his liv­ing costs from Novem­ber to Jan­uary. He plans to earn his liv­ing af­ter that as a con­sul­tant, speaker and writer on do­ing good effec­tively/​hav­ing a pos­i­tive im­pact more gen­er­ally. In this way, his work can nat­u­rally ex­tend into his ac­tivi­ties for EAN, and he can build a rep­u­ta­tion and net­work that adds value to EAN as well. Since some of his work will be per­son­al­ised to client needs, it is difficult to pin down how much of his time will go di­rectly to EAN’s ac­tivi­ties, but he ex­pects to be able to ded­i­cate a min­i­mum of 10 hours/​week to this in 2018.
Rem­melt is com­plet­ing a grad­u­a­tion pro­ject on multi-agent sys­tems for his de­gree right now and, as such, can live off study loans un­til Fe­bru­ary. Rem­melt is used to liv­ing fru­gally. For any work that he will take on in 2018, he has a strong prefer­ence for it to ac­cel­er­ate the rate at which he can build spe­cial­ised skills in the ar­eas of AI safety re­search and EA com­mu­nity-build­ing. He’s look­ing for part-time work to earn at least €800/​month in the mean­time (please email any sug­ges­tions to rem­melt@effec­tiefaltru­ This may in­clude tak­ing on re­mote re­search jobs or re­ceiv­ing sup­ple­men­tary ticket in­come from or­ganis­ing EA events. In the worst-case sce­nario, he may live off so­cial se­cu­rity for sev­eral months. He in­tends to work a min­i­mum of 20 hours/​week on EAN in 2018.
See Rem­melt’s con­sid­er­a­tions.

This is our base fund­ing sce­nario for EAN. For this, we are still look­ing for fund­ing of €5,900 for 2018 (af­ter sub­tract­ing a re­cent dona­tion of €1,000). This is based on an ap­prox­i­mate to­tal in over­head costs of €575/​month:
€300/​month: tax-ex­empt vol­un­teer re­im­burse­ments for Sjir & Rem­melt
€ 95/​month: venue hire for board & col­lab­o­ra­tive meet­ings
€ 80/​month: other over­head costs
€100/​month: to cover un­fore­seen costs

If you are con­sid­er­ing mak­ing a dona­tion to EAN or con­duct­ing re­search into com­mu­nity build­ing, we wel­come you to email any ques­tions to info@effec­tiefaltru­ We would ap­pre­ci­ate if you re­spond to our an­swers with hon­est feed­back, es­pe­cially your per­sonal anal­y­sis of EAN’s po­ten­tial to con­tribute to build­ing the EA com­mu­nity and any key ar­eas of im­prove­ment you see for us.

We have also very ten­ta­tively set out a growth fund­ing sce­nario for a max­i­mum ad­di­tional amount of €43,480 for 2018. This con­sists of the fol­low­ing roughly es­ti­mated costs:
€5,000 for a one-time liquidity buffer
to both safe­guard EAN’s longevity as an or­gani­sa­tion and to be able to re­spond quickly to high-po­ten­tial op­por­tu­ni­ties (op­tion value).

€ 500/​month max. for an in­tern, for which we’ve opened ap­pli­ca­tions
€2,000/​month max. for a (part-time) event man­ager start­ing from March at the ear­liest
This might be Rem­melt, de­pend­ing on his perfor­mance in terms of at­ten­dance and depth of en­gage­ment but we’re open to re­cruit­ing a pro­fes­sional.
€2,000/​month max. for an ex­pe­rienced (part-time) pro­ject man­ager start­ing from July at the earliest

Our fo­cus in both the ad­vi­sors/​coaches we’re now sur­round­ing our­selves with and the vol­un­teers/​hires we’d con­sider is on strength­en­ing our ‘in­ner cir­cle’ with com­mit­ted EAs who are more ac­tion-ori­ented than us, have had more pro­fes­sional work ex­pe­rience, and have differ­ent life per­spec­tives than we have. Whether and when it would be worth pay­ing for re­cruits de­pends on our trac­tion in 2018, which we cover fur­ther be­low.

See our cash flow out­look for 2018 for more de­tails.
(Italic num­bers de­note the fu­ture in­come streams that we feel we can rely on re­al­is­ing and costs that we ex­pect to make given enough fund­ing.)

On eval­u­at­ing perfor­mance:
Cer­tainly, one of our biggest ar­eas of im­prove­ment is in mea­sur­ing our im­pact. Com­pared with EA Lon­don, we have been in­con­sis­tent in track­ing met­rics and in­di­vi­d­ual be­havi­oural changes re­sult­ing from events and cam­paigns. The lack of sta­bil­ity caused by us try­ing out and shift­ing be­tween differ­ent spec­u­la­tive strate­gies and the fund­ing un­cer­tain­ties did make it more difficult to set up sys­tems. But now this is a ma­jor fo­cus point.

In striv­ing to ac­cu­rately as­sess our im­pact in en­gag­ing in­di­vi­d­u­als, we em­pha­sise the fol­low­ing:
per­spec­tives seem to point to the con­clu­sion that the ca­pac­i­ties of in­di­vi­d­u­als to do good lie roughly on a power-law dis­tri­bu­tion. Th­ese con­cern the core eth­i­cal and episte­molog­i­cal con­sid­er­a­tions that a per­son makes, the cause ar­eas and in­ter­ven­tions (s)he chooses to work on, the suc­cess of their work as a com­bined re­sult of their skills and in­cli­na­tions, and each of these effects be­ing am­plified by the spread of in­for­ma­tion and (dis)trust through­out the so­cial net­works within which the per­son is con­nected. Hav­ing said this, any at­tempt to es­ti­mate the fu­ture im­pact of a sup­posed ‘top 1%’ on such a dis­tri­bu­tion will likely be over­con­fi­dent be­cause of the di­men­sions missed (and risks re­in­forc­ing ac­cepted wis­dom on what traits make for an im­pact­ful EA).

Although the pro­cess of es­ti­mat­ing an in­di­vi­d­ual’s ca­pac­ity to do good is ex­tremely com­plex and laden with as­sump­tions, 80,000 Hours seems to have ac­quired the most ex­per­tise in do­ing this amongst the or­gani­sa­tions we’re aware of.

To eval­u­ate our own perfor­mance in en­gag­ing peo­ple to do good, we there­fore plan to count those we re­fer to 80,000 Hours whom are sub­se­quently se­lected for one-on-one coach­ing (and, prefer­ably, the re­sult­ing IASPCs) for our key met­ric. By set­ting 80,000 coach­ing refer­rals as our ‘North Star’, we in­cen­tivise our­selves to se­lec­tively tar­get and sup­port the de­vel­op­ment of the most promis­ing in­di­vi­d­u­als in­stead of pur­su­ing raw ‘mem­ber’ growth, while al­low­ing our­selves to build trust with the in­ter­na­tional com­mu­nity in our abil­ity to ex­e­cute.
A straight­for­ward solu­tion to track­ing this is a refer­ral link to their coach­ing web­page, but we’re con­cerned that this method would miss many of the peo­ple we ad­vise to ap­ply to 80,000 Hours. We will con­sult with them and CEA Groups on what pro­cesses we can adopt here.

Next to this, we’re track­ing the num­ber of peo­ple we are in touch with who are ded­i­cat­ing at least 10% of their time/​money to EA (‘Prac­tis­ing EAs’). Our aim is to bet­ter help the most promis­ing amongst these in­di­vi­d­u­als achieve shared al­tru­is­tic goals. For this, we’ve cre­ated an in­ter­nal database of peo­ple in our com­mu­nity (cur­rently: 234 peo­ple) in which we keep track of the skills, time and con­nec­tions that they are will­ing to con­tribute. Fi­nally, to eval­u­ate our in­di­vi­d­ual perfor­mance, Sjir and Rem­melt will each build in met­rics and feed­back mechanisms for the tar­geted out­reach, ca­pac­ity-build­ing events and one-on-one con­ver­sa­tions with or­ganisers that they will be tak­ing on.

We also seek to more closely col­lab­o­rate with CEA’s Com­mu­nity and Re­search Teams as well as the man­age­ment of Effec­tive Altru­ism Foun­da­tion. We take any ad­vice by them on re­duc­ing the long-term risks that come with pro­mot­ing EA se­ri­ously. At the same time, a sys­tem that re­lies on a few small cir­cles of in­di­vi­d­u­als de­ter­min­ing the in­ter­na­tional agenda for EA com­mu­nity build­ing would be frag­ile. In our view, EAN can add to the body of knowl­edge and the level of dis­cus­sion on this sub­ject.
See more of our views on this.

Next: share your feed­back
We wel­come your com­ments and ques­tions be­low, es­pe­cially of the crit­i­cal, in­sight­ful kind.
If you’re go­ing to EAG Lon­don, feel free to ap­proach Sjir or Rem­melt for a chat.