Review of CZEA “Intense EA Weekend” retreat

In March the mem­bers and friends of the Czech As­so­ci­a­tion of Effec­tive Altru­ism (CZEA) met for a week­end long in­tense re­treat (you can read more about CZEA in this post ). We would like to share our ex­pe­rience in case any EA group was in­ter­ested in do­ing some­thing similar.

We also be­lieve that the differ­ence be­tween run­ning a good re­treat and an av­er­age re­treat can be as big as be­tween run­ning an av­er­age re­treat and no re­treat at all, so it’s worth to in­vest the effort in de­vel­op­ing the for­mat.

Table of contents

Goals
Format
De­sign principles
Program
Log­ic­tics, costs and tips and tricks
Im­pact evaluation
Fu­ture plans

Goals

Our goals for the re­treat were, in or­der of importance

  1. Com­mu­nity build­ing and net­work­ing. Lo­cal EAs should leave know­ing each other better

  2. In­tro­duc­ing CZEA ac­tivi­ties and en­gag­ing more peo­ple in our projects

  3. An­a­lyz­ing the week­end im­pact and shar­ing our tips and insight

  4. Ed­u­ca­tion in more ad­vanced EA topics

Based on ex­ten­sive ques­tion­naires (EA peo­ple are will­ing to fill them up even if they are long), the event seems to have had im­pact on the goals.

  1. The av­er­age num­ber of EAs par­ti­ci­pants know well grew from 3.7 to 9.6 (op­er­a­tional­ized as “be­ing able to de­scribe in three sen­tences”)

  2. Aver­age self-re­ported knowl­edge of CZEA grew from 4.6 to 6.9 on a scale 1 to 10 (perfect).
    Ex­pected num­ber of hours par­ti­ci­pants plan to spend on EA ac­tivi­ties grew from 11.8 to 13.8 per week. If part of the effect per­sists, it means the event had some lev­er­age.

  3. Aver­age self-re­ported knowl­edge of effec­tive al­tru­ism grew from 5.1 to 6.5 on a scale 1 to 10.

  4. We will con­sider the goal “An­a­lyz­ing the week­end im­pact and shar­ing our tips and in­sight” fulfilled if some­one in EA com­mu­nity run­ning a week­end re­treat reads this and ac­tu­ally uti­lizes some of the info shared, so if you do, please let us know! It is re­ally im­por­tant for our in­ter­nal pri­ori­ti­za­tion.

Format

The event ran from Fri­day evening to Sun­day noon.

There were 25 par­ti­ci­pants in to­tal (and 8 more can­cel­led ap­pli­ca­tions), half of them already ac­tive in CZEA. Others were already fa­mil­iar with EA ba­sics.

De­sign principles

Estab­lish epistemic stan­dards. It helps to es­tab­lish epistemic stan­dard early. By epistemic stan­dard we mean ra­tio­nal rea­son­ing, co­op­er­a­tive solv­ing of dis­agree­ments, ask­ing about un­clear things, and in­de­pen­dent eval­u­a­tion of ev­i­dence and ar­gu­ments.

Mix­ing. The nat­u­ral or­der of things in so­cial groups tends to be as­sor­ta­tive match­ing. In the case of an EA re­treat, it may mean the more se­nior mem­bers talk­ing more among them­selves, and the new mem­bers also. This should be pre­vented. As a lot of things are best learned by os­mo­sis and im­plic­itly, it makes a lot of sense to ar­range the ac­tivi­ties in such a way that there are op­por­tu­ni­ties for this to hap­pen.

Dra­matic arc. Peo­ple en­joy an event more if it fits hu­man-com­pat­i­ble nar­ra­tive struc­ture. For ex­am­ple, if you imag­ine the in­tel­lec­tual in­ten­sity or the emo­tional in­ten­sity as the func­tion of time, it can build up, have some cli­max, fall down, have some much small peak just be­fore the end.

Rhythm. Hard to de­scribe for­mally, but if ac­tivi­ties are struc­tured in a good way, peo­ple get closer to the state of flow, are less tired, learn bet­ter. Think about the state of mind you want the par­ti­ci­pants to be af­ter an ac­tivity, and if the next ac­tivity fits.

Estab­lish com­mon top­ics and knowl­edge. Un­struc­tured talk in coffee breaks is of­ten the best part of sci­en­tific con­fer­ences, but the talks are in­dis­pens­able for es­tab­lish­ing com­mon top­ics and knowl­edge about top­ics.

Do not be afraid of in­ten­sity.

Avoid class­room look&feel Class­rooms tend to in­duce some un­for­tu­nate mind-states. (At least since 14th-cen­tury)

Program

Guided by the goals, we brain­stormed an over­abun­dance of ac­tivi­ties. Guided by the de­sign prin­ci­ples, we cre­ated a pro­gram. We re-checked if the pro­gram seemed to be al­igned with the goals. (About ½ of the ini­tial ideas did not get into the fi­nal pro­gram)

Here we list all the ac­tivi­ties, our rea­son­ing for their in­clu­sion, their rat­ing by the par­ti­ci­pants on 4 scales—to­tal util­ity, fun, learn­ing new ideas, and net­work­ing (U,F,L,N). Some­times we com­ment on ex­pe­rience gained af­ter the event from feed­back form or our im­pres­sions. The graphs ad­ja­cent to the top­ics show rat­ing on a scale from 1 (bad) to 5 (good), to em­pha­size the differ­ence the scale of the plots is from 1.5 to 4.8.

We hope it will be in­spira­tional, but definitely should be changed based on goals of your event, the au­di­ence, and unique op­por­tu­ni­ties aris­ing from the peo­ple pre­sent. We hope the feed­back scores could work as a rough guide how an ac­tivity may help a spe­cific goal.

Friday

Din­ner (18:20-19:20)

We paid for a high qual­ity all ve­gan food. Also snacks and tea were available dur­ing the whole event. In­vest­ing in good qual­ity cater­ing seemed worth it. Good light food and con­stant availa­bil­ity of drinks al­low peo­ple to keep con­cen­trated. Snack & tea bar is a nat­u­ral Schel­ling point for meet­ing peo­ple.

Open­ing Talk (19:30-19:50)

Oper­a­tions info. Also use­ful to in­tro­duce rules like “Pac­man” (a.k.a. “Open your cir­cles”), “To be con­tinued”, “This is not a class­room”

Dou­ble Crux(19:50-20:50)

Dou­ble Crux is a CFAR ra­tio­nal­ity tech­nique for re­solv­ing dis­agree­ments. There were a few CFAR alumni at­tend­ing the re­treat who were able to teach it.

Rea­sons for in­clu­sion: Estab­lish epistemic stan­dards early. Tech­nique use­ful for in­ter­nal CZEA com­mu­ni­ca­tion.

Reflec­tion: We be­lieve that knowl­edge of this tech­nique im­proved the qual­ity of con­ver­sa­tion dur­ing the whole week­end.

Or­ga­ni­za­tional Struc­ture of the Czech As­so­ci­a­tion for EA (20:50-21:20)

Rea­sons for in­clu­sion: Directly helps the goal of peo­ple un­der­stand­ing CZEA struc­ture.

Reflec­tion: Part of the info was un­nec­es­sar­ily du­pli­cated in other talks.

Ex­er­cise in Giv­ing Feed­back (21:30-22:20)

Giv­ing feed­back was mainly in­ter­ac­tive dis­cus­sion mod­er­ated by two CZEA mem­bers who are HR pro­fes­sion­als.

Rea­sons for in­clu­sion: Estab­lish epistemic stan­dards. In­duce peo­ple to give feed­back. Mak­ing the pro­gram more in­ter­ac­tive.

Tea Time (22:20-)

Saturday

Break­fast (9:00-9:30)

Ex­plain­ing Con­cepts (9:30-10:40)

Rea­sons for in­clu­sion: Net­work­ing, mix­ing, rais­ing the level of com­mon knowl­edge.

The par­ti­ci­pants were asked to pick from a list of in­ter­est­ing con­cepts (e.g. tragedy of com­mons, tox­o­plasma of rage, full list here) which ones they would like to have ex­plained to them or ex­plain to oth­ers. We matched small groups and had three runs of fif­teen min­utes long peer to peer ex­plain­ing.

Ex­pe­rience: From the feed­back this was the most pop­u­lar ac­tivity, scor­ing very high in all the crite­ria, con­tribut­ing to all goals.

Talk on Devel­op­ment of the EA Move­ment in the Last Five Years (10:40-11:40)

Rea­sons for in­clu­sion: Con­tribut­ing to­ward the goal of rais­ing knowl­edge about more ad­vanced EA top­ics.

Ex­pe­rience: In prac­tice, it was a use­ful op­por­tu­nity to clear some mis­con­cep­tions spread in early years of effec­tive al­tru­ism, like “it’s about earn­ing to give”, “it’s mostly about pledges”, “it’s mostly about fundrais­ing to GiveWell char­i­ties”

How to Use Slack, Trello, Gdrive etc. (11:45-12:15)

Rea­sons for in­clu­sion: Rhythm, some­thing not so de­mand­ing. Also at CZEA we use a whole stack of col­lab­o­ra­tive tools and we wanted ev­ery­one to be able to use them effec­tively.

Ex­pe­rience: For power users it was bor­ing.

Lunch (12:20-13:20)

Trio Walks (13:30-14:50)

Prac­tic­ing Dou­ble Crux and Mu­tual De­bug­ging in Trios. In­spired by CFAR.

Rea­sons for in­clu­sion: Net­work­ing, mix­ing, im­prov­ing ra­tio­nal­ity. Make peo­ple walk. May be so­cially more de­mand­ing.

Ex­pe­rience: The feed­back scores have bi­modal dis­tri­bu­tion—for some trios it was one of the high­lights of the week­end, for some it did not work at all.

Pre­sen­ta­tion of Czech and Slo­vak EA Pro­jects (15:00-16:00)

Rea­sons for in­clu­sion: Net­work­ing, get­ting ev­ery­body up­dated, ex­plain­ing what we do for new mem­bers.

Ex­pe­rience: Would be bet­ter to join this with “Job fair” which we had on Sun­day.

Group Photo (16:00-16:20)

An­nual Gen­eral Assem­bly of the Czech As­so­ci­a­tion for EA (16:20-18:00)

Rea­sons for in­clu­sion: Le­gal. We have to do it once a year.

Ex­pe­rience: Bor­ing as ex­pected.

Din­ner (18:10-19:50)

EA Themed Pub Quiz (19:30-20:30)

Rea­sons for in­clu­sion: Fun. Have some­thing not-so-se­ri­ous. Net­work­ing.

Com­mon Mis­takes of EA Stu­dents (20:30-21:00)

Talk by Jiří and Anna on top­ics such as “How to not be­come a de­pres­sive al­tru­ist”. Topics cov­ered seems im­por­tant for the wellbe­ing of many EAs (like “how not to be­come com­pletely ob­sessed with x-risk”)

Rea­sons for in­clu­sion: Learn­ing from mis­takes. The rea­son why to do it later in the course of the week­end rather than ear­lier: it costs fewer weird­ness points.

Ex­pe­rience: It was quite funny, peo­ple love to learn from mis­takes of oth­ers.

Split evening block of Ca­reer Plan­ning or EA Fu­ture Plans Brain­storm­ing or CFAR Tech­niques Train­ing (21:00-)

In this bloc, we split the group to three parts, based on differ­ent top­ics res­onat­ing in the group - ca­reer plan­ning, brain­storm­ing fu­ture plans for Czech effec­tive al­tru­ism, and train­ing of CFAR style ap­plied ra­tio­nal­ity. De­tailed de­scrip­tions would be too long, but gen­er­ally, all the op­tions were at­trac­tive.

Sunday

Break­fast (9:00-9:30)

On Heuris­tics (9:40-10:00)

Talk by Aleš. Ex­plain­ing some more ad­vanced EA top­ics.

Rea­sons for in­clu­sion: Directly aimed at one of the goals.

In­ter­ac­tive Theatre: How to Ex­plain and Talk About EA (10:05-10:45)

Rea­sons for in­clu­sion: Net­work­ing, mix­ing, fun. Devel­op­ment of so­cial skills and abil­ity to ex­plain EA con­cepts.

Ac­tivity based on Fo­rum the­atre form. One per­son acts the role of some­one hav­ing some se­ri­ous mis­con­cep­tion about EA, the other as an effec­tive al­tru­ist try­ing to ex­plain EA ideas. Mem­bers of the au­di­ence can sug­gest differ­ent ac­tions for the ac­tors, or come to “stage” and perform their ac­tions.

Job Fair (10:50-11:50)

“Job fair” for vol­un­teer­ing in pro­jects of CZEA

Rea­sons for in­clu­sion: This seemed the right point where peo­ple should make ac­tion­able plans for the fu­ture.

Ex­pe­rience: This would have been bet­ter joined with pro­ject in­tro­duc­tions.

Lunch

It makes sense to make this lunch break longer, as a space for in­for­mal dis­cus­sion.

Light­ning Talks by participants

Rea­sons for in­clu­sion: En­er­giz­ing ac­tivity, small peak be­fore the end. Allows re­mem­ber­ing peo­ple. Allows adding what was un­in­ten­tion­ally omit­ted.

Clos­ing remarks

Ex­pe­rience: Some peo­ple trav­el­led from far lo­ca­tions and were leav­ing dur­ing Sun­day and didn’t stay for the re­treat clos­ing.

Feed­back (30min)

Filling forms on­line, bring your own lap­top. Su­per-im­por­tant for an­a­lyz­ing the week­end im­pact and im­prov­ing the next run.

Rea­sons for in­clu­sion: It is hard to make peo­ple fill a >30min form at home.

Pro­mo­tion: How peo­ple learned about the event

Al­most all par­ti­ci­pants learned about the event through one of these “via Face­book event”, “via CZEA”, “through friends”.

De­ci­sive rea­sons for ac­tu­ally go­ing, as re­ported by par­ti­ci­pants, were:

  • Wanted to get to know the Czech EAs bet­ter (9x)

  • Par­ti­ci­pa­tion on the pro­gram of the event (4x)

  • Good for­mat, pro­gram (4x)

  • Veg food, lo­ca­tion (2x)

  • Rea­son­able price

Lo­gis­tics, costs and var­i­ous tips and ideas

To­tal ex­penses were around €75 per per­son.

Es­ti­mated time spent on plan­ning, ac­tivity de­vel­op­ment and feed­back anal­y­sis was about 200h, or about 1.5 months FTE.

Ve­gan food and good tea were ap­pre­ci­ated (both rated 910).

A gong is a use­ful tool for or­ga­niz­ers.

A ‘net­work­ing spread­sheet’ where peo­ple could write some­thing about them­selves be­fore the re­treat was a good thing to have.

Par­ti­ci­pants liked “the pac­man rule”: When you are stand­ing in a group, try to leave a place for one more per­son so other peo­ple can join in more eas­ily. (rated 4.4/​5) “To be con­tinued”, “This is not a class­room” , “In­ter­est­ing dis­cus­sion > talk” were some­what pos­i­tive (around 3.2/​5)

It helped us to have clearly defined goals.

Our mis­takes and things to do bet­ter

We didn’t pre­pare an ice­break­ing ac­tivity.

We had only one lec­ture room with no place to go for peo­ple who preferred dis­cus­sion in­stead of listen­ing to a lec­ture.

The level of talks and dis­cus­sion was some­times cho­sen to fit the least in­formed per­son. We could in­clude more ad­vanced top­ics. (The plot show an­swers to ques­tion Was the pro­gram easy or difficult? 1.. very easy 10… very difficult)

We didn’t ac­count for the “fun peak” at Satur­day evening, lead­ing to some­what un­ex­pected mood.

Peo­ple were leav­ing dur­ing Sun­day and didn’t stay for the re­treat clos­ing.

We didn’t pre­pare how to im­me­di­ately in­volve some of the newly mo­ti­vated par­ti­ci­pants af­ter the re­treat ended. Some may have lost in­ter­est (we will in­ves­ti­gate it fur­ther).

Im­pact eval­u­a­tion and feedback

The re­sponses were gen­er­ally pos­i­tive. At the end of the re­treat, par­ti­ci­pants re­ported a bet­ter un­der­stand­ing of the lo­cal or­ga­ni­za­tion and its pro­jects and were de­ter­mined to spend more time on EA-re­lated ac­tivi­ties. Some of them already joined our on­go­ing pro­jects (e.g. plan­ning of a Prague AI Safety con­fer­ence) and we see more en­gage­ment in our com­mu­nity. Also, com­mu­ni­ties in Brno and Bratis­lava have be­come more ac­tive af­ter their mem­bers at­tended the re­treat.

We will try to track medium-term im­pact by a fol­low-up sur­vey af­ter 6 months.

To give voice to par­ti­ci­pants di­rectly, these were some of the an­swers from feed­back forms :)

What is the most im­por­tant thing about which I have changed my mind?

  • the na­ture of effec­tive altruists

  • EA Bratislava

  • I be­lieve in my knowl­edge even less

  • I will ex­plore more

  • I think about peo­ple, and I think a lot

  • Ca­reer planning

  • I ap­pre­ci­ate the use­ful­ness of self-de­vel­op­ment and CFAR techniques

  • I thought I was just an ob­server, and in the end I was de­ter­mined to get in­volved quite a lot.

  • I’m won­der­ing if EA char­ity con­tri­bu­tions are the best I can do for EA.

  • Earn­ing to give

  • Con­tent—I found it im­por­tant to dis­cuss the pri­ori­ties for a long time and ac­tively seek to find facts that would help me change my mind—eg via dou­ble-crux.

  • How to ad­vise oth­ers with your career

  • Per­haps ex­plo­ra­tion /​ ex­ploita­tion.

  • The area of AI safety is ac­tu­ally worth attention

What sur­prised me the most?

  • The num­ber of ac­tively in­volved people

  • How great was the food

  • En­vi­ron­ment, food

  • Un­ex­pect­edly non-autistic

  • That I felt good among un­known peo­ple.

  • EA Theater

  • In­ten­sity of the program

Which parts of the pro­gram were least use­ful...?

Most com­mon an­swers in­cluded “or­ga­ni­za­tional info! and talk about Trello etc.

Credits

The main or­ga­niz­ers of the re­treat were Jiří Nád­vorník and Anna Ga­j­dová, with help from Kristýna Něm­cová and Jan Kul­veit. Talks and ac­tivi­ties also by Aleš Flídr, Naďa Bed­nárová, Pře­mek Paška, Mar­ika Řežábková, Veronika Portešová and Lenka Ray­manová. The event took place at Eko­cen­trum Louti, cater­ing was by Mo­mos and Ami­taya Tea.

We took a lot of in­spira­tion and some con­tent from CFAR.

We also took in­spira­tion from the method­ol­ogy of PSL and Velky Vuz, Czech or­ga­ni­za­tions de­vel­op­ing Ex­pe­ri­en­tial ed­u­ca­tion tech­niques.

Fu­ture plans

We plan to do a week­end re­treat at least once a year for CZEA.

If you would like to or­ga­nize some­thing similar we would be happy to help you, de­scribe any of the ac­tivi­ties in more de­tail, share ma­te­ri­als, etc.

We also have a few ac­tivi­ties we re­ally wanted to try but were not able to fit in the sched­ule

  • Play co­op­er­a­tive board games about sav­ing the world (e.g. Man­sions of Mad­ness)

  • An AI Safety themed LARP

(Note: Feed­back data anal­y­sis was done and this post was writ­ten co­op­er­a­tively by Anna Ga­j­dová, Jiří Nád­vorník and J.K.)