Updates from Leverage Research: history, mistakes and new focus

I’m post­ing to share in­for­ma­tion about Lev­er­age Re­search, a non-profit re­search or­gani­sa­tion founded in 2011 that was his­tor­i­cally in­volved in EA. In par­tic­u­lar, I wanted to share a sum­mary of their his­tory and their new fo­cus fol­low­ing a ma­jor re­struc­ture.

My main goals for this post are to:

  1. Estab­lish me as a con­tact per­son for Leverage

  2. Give read­ers a bet­ter un­der­stand­ing of Lev­er­age in the past and an up­date on their new focus

  3. Clar­ify and im­prove the re­la­tion­ship be­tween Lev­er­age and the EA community


I hope to achieve this by writ­ing three main sec­tions:

  1. About me
    Briefly cov­er­ing my role at Lev­er­age, why I’m the per­son post­ing this and my re­la­tion­ship to the EA community

  2. What is Lev­er­age?
    Aiming to give read­ers a ba­sic un­der­stand­ing of what Lev­er­age was do­ing in the past and their plans for the fu­ture, in­clud­ing a lit­tle bit about our re­la­tion­ship with Paradigm Academy.

  3. Lev­er­age and EA: Our mis­takes and what to ex­pect from Lev­er­age mov­ing for­ward
    Clar­ify­ing Lev­er­age’s re­la­tion­ship to the EA com­mu­nity, ad­dress­ing some con­cerns and set­ting out what the EA com­mu­nity can ex­pect from us mov­ing for­ward. I wrote this sec­tion for a nar­rower au­di­ence of peo­ple who’ve in­ter­acted with Lev­er­age in the past and have more con­text on our past re­la­tion­ship with EA.

Many thanks to ev­ery­one who has pro­vided feed­back or oth­er­wise helped me in bring­ing to­gether this post.


1. About me

My name is Larissa Hes­keth-Rowe. I re­cently ac­cepted a job at Lev­er­age Re­search. I’m cur­rently sup­port­ing Lev­er­age with their ex­ter­nal com­mu­ni­ca­tions, al­though longer-term I ex­pect to take on re­spon­si­bil­ity for their Re­search Fel­lows Pro­gram.

I’ve also been an ac­tive mem­ber of the EA com­mu­nity for many years. I started as a vol­un­teer, group leader, and Giv­ing What We Can mem­ber. More re­cently, I worked at the Cen­tre for Effec­tive Altru­ism, first in com­mu­ni­ca­tions and com­mu­nity roles and later as the CEO, run­ning the or­gani­sa­tion in 2018. While I no longer work at an EA or­gani­sa­tion, I still con­sider my­self a mem­ber of the EA com­mu­nity, and I plan to con­tinue to be ac­tively en­gaged in sup­port­ing EA through things like vol­un­teer­ing as a men­tor in the Women and Non-Bi­nary Altru­ism Men­tor­ship (Wan­bam) pro­gramme and fulfilling my Giv­ing What We Can pledge.

My in­volve­ment in EA means I’m in a good po­si­tion to ex­plain to the EA com­mu­nity what Lev­er­age is do­ing and clar­ify Lev­er­age’s re­la­tion­ship to EA. While I don’t ex­pect Lev­er­age to have much di­rect in­volve­ment in EA, both groups are work­ing to im­prove the world, so I am per­son­ally mo­ti­vated to im­prove com­mu­ni­ca­tion be­tween them.

If you ever want to reach out to Lev­er­age for any rea­son or have any ques­tions, you can reach out to me at larissa@lev­er­agere­search.org.


2. What is Lev­er­age?

Lev­er­age has been a com­pli­cated pro­ject with a num­ber of differ­ent com­po­nents. In the past, Lev­er­age was es­sen­tially a small com­mu­nity ex­per­i­ment­ing with how to con­duct use­ful early stage re­search, study­ing cause pri­ori­ti­sa­tion and so­cial sci­ences and run­ning ad-hoc world im­prove­ment pro­jects.

One way we’re hop­ing to make Lev­er­age more in­tel­ligible ex­ter­nally is that Ge­off An­ders (the founder and ED of Lev­er­age) has just started writ­ing an es­say se­ries about Lev­er­age’s his­tory. The first post is now up, which in­cludes a list of the es­says you can ex­pect to see go up over the com­ing months. It’s likely Ge­off will edit the posts as he gets feed­back. The se­ries will ex­plain some of the re­search av­enues Lev­er­age cov­ered and the pro­jects they worked on. Since we wrote most of our con­tent for the in­ter­nal Lev­er­age com­mu­nity, it’s a long pro­ject to work out what to pri­ori­tise shar­ing and how to do that. In this post, I’m merely try­ing to help read­ers un­der­stand what Lev­er­age was do­ing rather than share past con­tent.


2.1. “Lev­er­age 1.0” vs “Lev­er­age 2.0”

In the past, the name “Lev­er­age” has been used broadly, not just refer­ring to Lev­er­age the or­gani­sa­tion but also nearby groups with which they co­or­di­nated. To make this eas­ier to fol­low, I’ll make the dis­tinc­tion be­tween “Lev­er­age 1.0” and “Lev­er­age 2.0”. I’m us­ing Lev­er­age 1.0 to re­fer to Lev­er­age from 2011 to 2019, in­clud­ing other or­gani­sa­tions that de­vel­oped out of Lev­er­age such as Paradigm Academy. I’ll use Lev­er­age 2.0 to re­fer to just the or­gani­sa­tion Lev­er­age Re­search and its staff from the sum­mer of this year on­wards. Lev­er­age 2.0 is what we will mean by Lev­er­age mov­ing for­ward.


2.2. Some ways of un­der­stand­ing Lev­er­age 2011 − 2019 (“Lev­er­age 1.0”):

Below I have at­tempted to dis­til nearly nine years of his­tory into its core com­po­nents. My sum­mary will end up be­ing an over­sim­plifi­ca­tion but hope­fully a use­ful one.

Since de­cid­ing to join the Lev­er­age (2.0) team, I have been try­ing to de­velop my own un­der­stand­ing of Lev­er­age 1.0. In do­ing so, I’ve come to the view that the best way to un­der­stand Lev­er­age 1.0 (2011 − 2019) is as a com­bi­na­tion of:

  1. an ex­per­i­ment in build­ing an effec­tive early stage re­search community

  2. a cause pri­ori­ti­sa­tion re­search collaboration

  3. an or­gani­sa­tion fo­cused on un­der­stand­ing ideas, in­di­vi­d­u­als and society

  4. a gen­eral world im­prove­ment project

Much like early mem­bers of the EA com­mu­nity, the origi­nal Lev­er­age team mem­bers were trou­bled by the many prob­lems in the world: global poverty, to­tal­i­tar­i­anism, the threat of nu­clear war. For hu­man­ity to be able to make progress on such is­sues, Lev­er­age be­lieved hu­man­ity first needs to un­der­stand a lot more about the world. In par­tic­u­lar, they thought it was es­sen­tial to un­der­stand bet­ter how to con­duct high-qual­ity re­search, what prob­lems in the world are most im­por­tant to fo­cus on, and how peo­ple, in­sti­tu­tions, and so­cieties shape the world. Lev­er­age Re­search was founded to con­tribute to that un­der­stand­ing.


2.2.1. Lev­er­age as an ex­per­i­ment in build­ing an effec­tive early stage re­search community

The first challenge was to cre­ate a re­search col­lab­o­ra­tion that was ca­pa­ble of mak­ing progress across a wide scope of po­ten­tial re­search av­enues.

To try and cre­ate a pro­duc­tive re­search team, Lev­er­age hired and col­lab­o­rated with peo­ple with a wide range of differ­ent back­grounds, view­points, and cre­den­tials and gave them the free­dom to in­ves­ti­gate what­ever seemed ap­pro­pri­ate and in­ter­est­ing to them. They ex­plored differ­ent ways of con­duct­ing re­search, stud­ied var­i­ous re­search tra­di­tions, and tried out differ­ent de­bate and dis­cus­sion for­mats to help peo­ple learn from each other.

Lev­er­age took an ex­per­i­men­tal ap­proach to set­ting up the team struc­ture such that in many ways, it can make more sense to think of Lev­er­age 1.0 as a re­search com­mu­nity than a re­search or­gani­sa­tion. While hiring some­one and pro­vid­ing them a salary was one way to co­or­di­nate with some­one on re­search, they also had lots of more in­for­mal col­lab­o­ra­tions with vis­it­ing and ex­ter­nal re­searchers.

Lev­er­age was also run in a rea­son­ably de­cen­tral­ised way. In the early days, re­searchers were given a lot of au­ton­omy over their re­search. Later new re­searchers joined ex­ist­ing teams, but these teams still de­vel­oped or­gan­i­cally around the re­search av­enues on which peo­ple wanted to col­lab­o­rate. All of this means that Lev­er­age didn’t have what you might think of as a tra­di­tional hi­er­ar­chy with man­age­ment cen­trally di­rect­ing the re­search. In­stead, the team shared broadly over­lap­ping view­points and plans, and the sup­port and ad­vice of the team lead­ers pro­vided guidance.


2.2.2. Lev­er­age as a cause pri­ori­ti­sa­tion re­search collaboration

As well as un­der­stand­ing how best to con­duct re­search, Lev­er­age also had to de­ter­mine what was most im­por­tant to study if you wanted to im­prove the world. One can, there­fore, also un­der­stand Lev­er­age as a cause pri­ori­ti­sa­tion re­search pro­ject.

To tackle the broad scope of po­ten­tial re­search av­enues, Lev­er­age con­ducted ex­tremely open-ended re­search, fol­low­ing nu­mer­ous re­search paths to their nat­u­ral end­point. A re­search av­enue might reach its nat­u­ral end­point if you de­cided it was too hard to be worth con­tin­u­ing, too dan­ger­ous to con­tinue, or you gen­er­ated proof of pos­si­bil­ity.

If a re­search path was too hard, it could be sub­sti­tuted for a more man­age­able prob­lem or de­pri­ori­tised al­to­gether. If you gen­er­ated proof of pos­si­bil­ity, you could ei­ther con­tinue de­vot­ing more time to that av­enue or not, de­pend­ing on what open re­search av­enues seemed to be the most promis­ing.

From the out­side, Lev­er­age’s re­search was un­der­stand­ably con­fus­ing be­cause they were pri­ori­tis­ing mov­ing through a wide range of re­search ar­eas as effi­ciently as pos­si­ble rather than com­mu­ni­cat­ing the re­sults to oth­ers. This ap­proach was de­signed to al­low them to cover more ground with their re­search and nar­row in quickly on ar­eas that seemed the most promis­ing.

If you want to read more about the var­i­ous re­search av­enues Lev­er­age 1.0 ex­plored, keep an eye on Ge­off’s web­site for es­says cov­er­ing these.


2.2.3. Lev­er­age as an or­gani­sa­tion fo­cused on un­der­stand­ing ideas, in­di­vi­d­u­als and society

Ge­off had a par­tic­u­lar in­ter­est in psy­chol­ogy, so­ciol­ogy and philos­o­phy as a means to un­der­stand­ing peo­ple, so­cieties, and how our un­der­stand­ing of the world has pro­gressed. As Lev­er­age made progress down var­i­ous re­search av­enues, their work in these ar­eas looked par­tic­u­larly promis­ing.

In psy­chol­ogy, they con­tinued to de­velop a model of the hu­man mind, and the struc­ture of in­di­vi­d­ual be­liefs called Con­nec­tion The­ory (CT) and de­vel­oped var­i­ous in­tro­spec­tion tech­niques and pro­cesses for over­com­ing differ­ent types of men­tal blocks. In so­ciol­ogy, they stud­ied group dy­nam­ics, past civil­i­sa­tions and in­sti­tu­tions. Lev­er­age also learned a lot about ab­stract method­olog­i­cal re­search, stud­ied the his­tory of sci­ence and built up knowl­edge about how to con­duct early stage re­search. This lat­ter area, in par­tic­u­lar, con­tributed a lot to the work we are now do­ing on early stage sci­ence (the study of how sci­en­tific progress hap­pens in fields with­out well-de­vel­oped sci­en­tific re­search pro­grams). More on this in sec­tion 2.4 on Lev­er­age Re­search to­day.


2.2.4. Lev­er­age as a gen­eral world im­prove­ment project

Fi­nally, much like the EA com­mu­nity, Lev­er­age did not just want to con­duct re­search from their arm­chairs, they wanted to put it into prac­tice. Lev­er­age’s sister com­pany Paradigm Academy, for ex­am­ple, de­vel­oped out of a de­sire to put some of their find­ings in psy­chol­ogy into prac­tice by train­ing in­di­vi­d­u­als. Paradigm pro­vides train­ing to in­di­vi­d­u­als and in­cu­bates star­tups.

They also wanted to meet like-minded peo­ple and were ex­cited about grow­ing the num­ber of peo­ple con­tribut­ing to press­ing prob­lems in the world to­day. As the Lev­er­age com­mu­nity ended up de­vel­op­ing around the same time as EA, Lev­er­age did some work to sup­port the EA com­mu­nity in the early days be­fore there were more cen­tral­ised move­ment-build­ing efforts. For ex­am­ple, Lev­er­age set up THINK, an early EA lo­cal group build­ing pro­ject and they ran some of the first EA con­fer­ences (e.g. the 2013 and 2014 EA Sum­mits) be­fore hand­ing these over to the Cen­tre for Effec­tive Altru­ism (CEA).


2.3. Con­duct­ing re­search not shar­ing research

Notably, Lev­er­age’s fo­cus was never par­tic­u­larly on shar­ing re­search ex­ter­nally. Some­times this was be­cause it was a quick ex­plo­ra­tion of a par­tic­u­lar av­enue or seemed dan­ger­ous to share. Often though it was a time trade-off. It takes time to com­mu­ni­cate your re­search well, and this is es­pe­cially challeng­ing when your re­search uses un­usual method­ol­ogy or start­ing as­sump­tions. Ge­off will talk more about this in his es­say se­ries, and I dis­cuss it a bit fur­ther in sec­tion 3.2.2: Com­mu­ni­ca­tion about our work.


2.4. Lev­er­age or­gani­sa­tional restructure

By the sum­mer of 2019, Lev­er­age’s pri­mary re­search ar­eas were mostly func­tion­ing as dis­tinct teams. As I have men­tioned, Lev­er­age was de­cen­tral­ised in terms of man­age­ment struc­ture, and the var­i­ous teams acted au­tonomously, mostly in­de­pen­dent of any over­ar­ch­ing man­age­ment struc­ture. As Lev­er­age grew, they came up against more and more challenges in co­or­di­nat­ing across those teams.

For a va­ri­ety of rea­sons, it had be­gun to seem as though much greater cen­tral co­or­di­na­tion was nec­es­sary at Lev­er­age, which would mean more cen­tral­ised man­age­ment guid­ing the teams’ ac­tivi­ties. How­ever, the differ­ent groups at Lev­er­age had already de­vel­oped their own team cul­tures, iden­tities, and plans. Many of the ex­ist­ing staff had ini­tially joined to con­duct re­search un­der a very open-ended re­search man­date, so the move to be­com­ing an or­gani­sa­tion with more cen­tral di­rec­tion was not ap­peal­ing to ev­ery­one.

For these rea­sons, af­ter much re­flec­tion, the re­search col­lab­o­ra­tion that had been Lev­er­age Re­search (“Lev­er­age 1.0”) was for­mally dis­banded ear­lier this year and re­formed into a re­search in­sti­tute fo­cused on early stage sci­ence re­search, in­clud­ing early stage psy­chol­ogy.

Many of the differ­ent teams then for­mally split out to be­come or­gani­sa­tions that are in­de­pen­dently funded and run.


2.5. Lev­er­age Re­search to­day (“Lev­er­age 2.0”): an early stage sci­ence re­search institute

Fol­low­ing the re­struc­ture, Lev­er­age Re­search (“Lev­er­age 2.0”) stopped fo­cus­ing on cause pri­ori­ti­sa­tion re­search and, while cre­at­ing a pro­duc­tive re­search en­vi­ron­ment is still im­por­tant to us as a re­search in­sti­tute, this is no longer a fo­cus of study. Lev­er­age’s more gen­eral world im­prove­ment pro­jects are ei­ther han­dled by sep­a­rate or­gani­sa­tions (e.g. train­ing at Paradigm Academy) or no longer rele­vant to their cur­rent work (e.g. the past work on EA move­ment build­ing. More on this in the sec­tion be­low about Lev­er­age’s re­la­tion­ship to the EA com­mu­nity).

Mov­ing for­ward, Lev­er­age Re­search will fo­cus on early stage sci­ence re­search. Our new mis­sion is to sup­port sci­en­tific progress by ed­u­cat­ing peo­ple about early stage sci­ence, fund­ing promis­ing re­search pro­jects, and con­duct­ing own early stage re­search, par­tic­u­larly in the so­cial sci­ences. If you’d like to know more about our work on early stage sci­ence, check out this page of our web­site.

This new di­rec­tion will en­tail our re­searchers en­gag­ing with academia, hav­ing our work re­viewed ex­ter­nally, and con­nect­ing with other in­di­vi­d­u­als and in­sti­tu­tions in­volved in early stage sci­ence. For this rea­son, Lev­er­age is now work­ing on pub­lish­ing more of our work, in­creas­ing our pub­lic en­gage­ment, and we have just re­leased a beta ver­sion of our web­site.

Lev­er­age 2.0 in this form and with this fo­cus is still new, so I ex­pect things to shift a bit as we con­tinue to de­velop our new strat­egy and re­search agenda. We’re still work­ing out what of our past con­tent is rele­vant to our cur­rent work to write up and how to share our other con­tent.

Since I’m con­scious of mak­ing this post too long for read­ers, if you have any ques­tions about any­thing I’ve not touched on here, please com­ment on this post and check out our web­site for fu­ture up­dates.


2.6. Lev­er­age Re­search (Lev­er­age 2.0) and Paradigm Academy

Of the var­i­ous or­gani­sa­tions that ei­ther had been part of Lev­er­age 1.0, Ge­off An­ders now runs only Paradigm Academy and Lev­er­age Re­search. The oth­ers are now all run in­de­pen­dently so I’m afraid I can’t speak for them. Since I some­times work with Paradigm staff and we share a man­age­ment team, if you have ques­tions about Paradigm I may be able to an­swer them here but the best per­son to speak to about Paradigm is Mindy McTeigue (mindy@paradig­macademy.com).

Over the years, Lev­er­age has, at many points, con­sid­ered hav­ing Paradigm Academy run much more in­de­pen­dently. How­ever, there con­tinues to be over­lap, es­pe­cially since the re­struc­ture, where Paradigm is well-po­si­tioned to provide op­er­a­tions sup­port as part of their in­cu­ba­tion pro­gramme, Paradigm’s train­ing benefits from Lev­er­age’s psy­chol­ogy re­search and Lev­er­age staff benefit from Paradigm train­ing. I, there­fore, ex­pect this over­lap to con­tinue. Cur­rently, Lev­er­age con­tracts Paradigm to run their op­er­a­tions while we look for our own op­er­a­tions man­ager. We also con­tract some of their train­ers to provide train­ing to Lev­er­age staff. Of the team, Ge­off An­ders and Mindy McTeigue work at both or­gani­sa­tions. Mindy joins Ge­off at Lev­er­age (pre­vi­ously she worked at Paradigm) since be­ing pro­moted to Chief Oper­at­ing Officer, sup­port­ing Ge­off in man­age­ment. You can see the cur­rent Lev­er­age team on our team page.

Paradigm cur­rently con­tinues to provide in­di­vi­d­ual train­ing and startup in­cu­ba­tion. Once more of the ground­work has been laid for Lev­er­age to fo­cus on its new mis­sion, Ge­off and Mindy will likely fo­cus more on up­dat­ing Paradigm’s web­site and com­mu­ni­cat­ing about their work.


3. Our mis­takes and what to ex­pect mov­ing forward

Hope­fully, my last sec­tion has helped peo­ple un­der­stand the ba­sics of what Lev­er­age was and will be. In this next sec­tion, I’d like to talk about some ques­tions and con­cerns that peo­ple in the EA com­mu­nity have brought up about Lev­er­age start­ing with some brief con­text.

The mis­takes dis­cussed here are from Lev­er­age 1.0. While I work at Lev­er­age and have writ­ten this on their be­half, I’ve been au­tho­rised to speak on be­half of Paradigm here too, to make the same com­mit­ments mov­ing for­ward for Lev­er­age and Paradigm. I’ll of­ten re­fer to the team as “we” in­stead of “they” as I have so far when de­scribing Lev­er­age in the past when I was not in­volved. I chose to write this way be­cause I’m con­vey­ing apolo­gies from the en­tire team that I will be helping to en­sure we keep these com­mit­ments mov­ing for­ward, so this sec­tion feels to me like a team effort. Apolo­gies if this be­comes con­fus­ing.

I don’t ex­pect this sec­tion to put an end to all dis­agree­ments or set­tle all con­cerns. If you’re work­ing on some­thing that’s both im­por­tant and highly un­cer­tain, there are bound to be dis­agree­ments, and some amount of this seems healthy for broad­en­ing your per­spec­tive and challeng­ing one an­other to do bet­ter. How­ever, we think the EA com­mu­nity is do­ing im­por­tant work, and so we don’t want to jeop­ar­dise that.


3.1 Lev­er­age and the EA community

Lev­er­age 1.0 started up around the same time as the EA com­mu­nity did, and shared similar mo­ti­va­tions; a deep com­mit­ment to im­prov­ing the world and a be­lief that through care­ful rea­son­ing, we can do more good. In the past, we sup­ported the growth of the EA com­mu­nity and were in­volved in EA move­ment-build­ing pro­jects.

And yet, Lev­er­age has, as a friend re­cently de­scribed it, of­ten seemed like “a bit of a square peg in a round hole in the EA com­mu­nity”. Lev­er­age 1.0 started from differ­ent ideas and as­sump­tions. While Lev­er­age is made up of in­di­vi­d­u­als with differ­ent views, in gen­eral Lev­er­age has much more baseline scep­ti­cism of main­stream in­sti­tu­tions than I’ve gen­er­ally found in the EA com­mu­nity; we don’t pri­ori­tise Bayesian rea­son­ing when try­ing to im­prove our think­ing or tend to use quan­ti­ta­tive mod­els as of­ten; we place much more weight on the im­por­tance of un­der­stand­ing in­di­vi­d­ual psy­chol­ogy, group dy­nam­ics and global in­cen­tives and power struc­tures. And, al­though im­prov­ing sci­en­tific progress is the kind of high-risk, high-re­ward bet some EAs do pri­ori­tise, Lev­er­age’s plan for it looks very differ­ent.

In an ideal world, neigh­bour­ing groups like this might have spurred each other on and been a way to pro­duc­tively challeng­ing each oth­ers’ ideas. But in our not-so-ideal re­al­ity, real-life in­ter­ac­tions can be messy, and even well-in­ten­tioned com­mu­ni­ca­tion some­times fails.

Since cen­tral­is­ing in Lev­er­age 2.0, we’ve been think­ing about our plans and re­al­is­ing we need to do a lot more to com­mu­ni­cate our work and en­gage with ex­ter­nal groups effec­tively. In re­flect­ing on this, we ended up think­ing a lot more about the mis­takes we made in the past.

Lev­er­age, there­fore, wants to apol­o­gise for in­stances where we’ve caused dam­age, lay out some of the mis­takes we have made and set out what neigh­bour­ing pro­jects can ex­pect from us mov­ing for­ward.

We also want to make clear our re­la­tion­ship to the EA com­mu­nity to­day. While his­tor­i­cally we were in­volved in EA move­ment-build­ing among our other world im­prove­ment pro­jects, and we con­tinue to sup­port any com­mu­ni­ties try­ing to make the world bet­ter, the EA com­mu­nity is not part of our cur­rent fo­cus. This means that while our work may be of in­ter­est to some peo­ple here, we may work with peo­ple in the EA com­mu­nity, and we are broadly sup­port­ive of the work the EA com­mu­nity is do­ing, nei­ther Lev­er­age nor Paradigm is di­rectly in­volved in try­ing to build or pro­mote EA. I ex­pect staff that are also part of the EA com­mu­nity (such as my­self) will con­tinue to be in­ter­ested in sup­port­ing EA pro­jects in our spare time, but this won’t be a fo­cus for Lev­er­age or Paradigm as or­gani­sa­tions.


3.2 Con­cerns about Lev­er­age and what to ex­pect from us mov­ing forward

The main mis­takes Lev­er­age has made with re­gards to our re­la­tion­ship to the EA com­mu­nity, which I will try to ad­dress are:

  1. our ap­proach to co­or­di­na­tion with other or­gani­sa­tions,

  2. how we com­mu­ni­cate about our work,

  3. our at­ti­tude to­ward PR and reputation

  4. some of our in­ter­per­sonal in­ter­ac­tions with in­di­vi­d­u­als.

There’s also the ques­tion of how to as­sess Lev­er­age’s im­pact, which I will dis­cuss in the fi­nal part of this sec­tion.


3.2.1. Co­or­di­na­tion with other organisations

In the early days, Lev­er­age 1.0 had very differ­ent views on PR and move­ment-build­ing from oth­ers in the EA com­mu­nity. Lev­er­age staff were ex­cited to get more peo­ple in­volved in figur­ing out how to un­der­stand the world and donat­ing to press­ing prob­lems in global de­vel­op­ment. While the opinions of in­di­vi­d­ual Lev­er­age mem­bers differed in many ways, it would be fair to say that as a group, we tended to think con­cerns about brand­ing and risks aris­ing from grow­ing the move­ment too quickly were overblown. Th­ese differ­ences meant Lev­er­age had early strate­gic dis­agree­ments with or­gani­sa­tions in the EA com­mu­nity and our co­or­di­na­tion at­tempts were of­ten clumsy and naive.

We think we then later took too ad­ver­sar­ial an ap­proach in dis­agree­ments with neigh­bour­ing or­gani­sa­tions. For in­stance, Lev­er­age lead­er­ship con­cluded that other or­gani­sa­tions were not go­ing to pri­ori­tise EA move­ment-build­ing ad­e­quately. In­stead of en­gag­ing in di­alogue about the differ­ences, Lev­er­age took unilat­eral ac­tion to try to build the EA move­ment, run­ning con­fer­ences and al­ly­ing with pro-move­ment growth EAs.

Some­times unilat­eral ac­tion is nec­es­sary to tackle en­trenched pow­ers and in­cen­tives that are un­re­spon­sive to con­cerns, but this should not be taken lightly. It’s im­por­tant to se­ri­ously con­sider all the po­ten­tial con­se­quences and take such ac­tion only as a last re­sort when you have ex­hausted other op­tions. While Lev­er­age tried to weigh the con­se­quences of our planned ac­tivi­ties and as­sess the re­al­is­tic chances of co­or­di­na­tion with other or­gani­sa­tions, we’ve con­cluded that we should have con­tinued to reach out and try to en­gage in col­lab­o­ra­tion and di­alogue, even though ear­lier at­tempts at this failed.

Mov­ing for­ward, EA or­gani­sa­tions can ex­pect both Lev­er­age and Paradigm to:

  1. Not run EA move­ment-build­ing pro­jects. Work­ing di­rectly on grow­ing EA, is not our com­par­a­tive ad­van­tage nor any longer part of our focus

  2. Reach out ear­lier if we end up plan­ning ini­ti­a­tives that might im­pact the EA com­mu­nity and en­gage more in di­alogue over strate­gic disagreements

  3. Where the EA com­mu­nity would find this helpful, do more to sup­port their work. For ex­am­ple, con­nect­ing in­di­vi­d­u­als we meet to the EA com­mu­nity and pro­vid­ing re­sources and train­ing if re­quested by mem­bers of the EA com­mu­nity. We wouldn’t ex­pect to do this un­der an EA brand, we’d just be sup­port­ing pro­jects we thought were good for the world.

  4. Be more re­spon­sive to con­cerns from EA com­mu­nity mem­bers or or­gani­sa­tional lead­ers. We’ve re­ceived a range of feed­back over the years. Some of this feed­back was not always con­struc­tive or was hard to en­gage in di­alogue around (e.g. anony­mous posters). How­ever, we re­al­ise that we could have worked harder to bridge mi­s­un­der­stand­ings and work out which feed­back was con­struc­tive so that we could in­cor­po­rate that.

If there are fur­ther sug­ges­tions peo­ple have here, please feel free to add them in the com­ments or to email me (larissa@lev­er­agere­search.org ).


3.2.2. Com­mu­ni­ca­tion about our work

We know that it hasn’t been easy to un­der­stand Lev­er­age’s work in the past. Com­mu­ni­ca­tion about Lev­er­age has of­ten been spo­radic, hard for ex­ter­nal au­di­ences to un­der­stand, and ac­cess to our ma­te­ri­als re­stricted, lead­ing Lev­er­age to be shrouded in mys­tery. This con­fu­sion then con­tributes to con­ver­sa­tions be­ing less pro­duc­tive and our im­pact be­ing less clear.

As I men­tioned in the his­tory of Lev­er­age, there was a trade-off in time spent con­duct­ing re­search ver­sus time spent com­mu­ni­cat­ing it. As we didn’t in­vest time early on in com­mu­ni­cat­ing about our work effec­tively, it only be­came harder over time as we built up our mod­els and on­tolo­gies. While of­ten this was the right trade-off for Lev­er­age 1.0 where the fo­cus was ad­vanc­ing our ideas, some­times it wasn’t and in ei­ther case, this makes the job of com­mu­ni­cat­ing our work mov­ing for­ward challeng­ing[1].

We also made more gen­eral com­mu­ni­ca­tions er­rors. In par­tic­u­lar, we of­ten made promises to provide fur­ther up­dates on var­i­ous top­ics on par­tic­u­lar time frames but then rarely posted the promised up­dates at all, let alone by the stated dead­line. In my ex­pe­rience, this is a com­mon com­mu­ni­ca­tions mis­take but an eas­ily avoid­able one[2].

The main com­mu­ni­ca­tion im­prove­ments peo­ple can ex­pect from Lev­er­age Re­search mov­ing for­ward are:

  1. Clear in­for­ma­tion about who we are and what we do on our web­site, in­clud­ing in­for­ma­tion about our work and a team page

  2. New re­search re­lated to psy­chol­ogy and early stage sci­ence to be shared on the Lev­er­age web­site mov­ing forward

  3. Some of our con­tent re­lat­ing to train­ing tech­niques and self-im­prove­ment to be available on the Paradigm Academy web­site (al­though likely not un­til next year)

  4. Con­tent that doesn’t fit un­der ei­ther or­gani­sa­tion to be shared as part of Ge­off’s es­say se­ries on the his­tory of Lev­er­age or pub­lished by in­di­vi­d­u­als.

  5. More small events hosted by our staff in the Bay that give peo­ple who are in­ter­ested in our work the op­por­tu­nity to meet us and ask questions

  6. A re­spon­sive point per­son to whom you can di­rect in­quiries about Lev­er­age’s work (me!)

  7. We will seek to com­mu­ni­cate more ac­cu­rately and clearly about our fu­ture work and up­dates, in­clud­ing our un­cer­tainty. If peo­ple are in­ter­ested in par­tic­u­lar up­dates they can reach out to me[2].

We ex­pect it’s un­likely that much of our in­creased ex­ter­nal com­mu­ni­ca­tion will be in EA chan­nels (like the EA Fo­rum) as much of it may not be di­rectly rele­vant to EA, but we will share re­search through our own chan­nels for those who are in­ter­ested.

Paradigm Academy is not cur­rently fo­cused on ex­ter­nal com­mu­ni­ca­tion as much as Lev­er­age is, so I don’t have up­dates on their com­mu­ni­ca­tion plans. If you have ques­tions about Paradigm, please con­tact Mindy.


3.2.3. PR and reputation

Lev­er­age 1.0 has his­tor­i­cally un­der­val­ued rep­u­ta­tion and PR and in­stead fo­cused more sin­gle-mind­edly on the achieve­ment of its goals. This con­tributed to our lack of fo­cus on com­mu­ni­cat­ing about our work which, in turn, dam­aged our rep­u­ta­tion in the EA com­mu­nity.

The EA com­mu­nity has had a great deal of suc­cess with things like bring­ing in large fun­ders and work­ing with gov­ern­ments and other in­sti­tu­tions. This suc­cess has been crit­i­cal to spread­ing EA ideas, im­prov­ing policy, tack­ling dis­eases and sav­ing an­i­mals from fac­tory farms. Much of this suc­cess is at­tributable to the ways the EA com­mu­nity has care­fully man­aged its rep­u­ta­tion. We think that the EA com­mu­nity is do­ing in­cred­ibly im­por­tant work, and we don’t want to jeop­ar­dise that.

While we now bet­ter un­der­stand the con­sid­er­a­tions other EA or­gani­sa­tions had when we first dis­agreed about move­ment-build­ing strat­egy, our pri­mary fo­cus is re­search. Our re­search fo­cus (early stage sci­ence) of­ten in­volves work­ing with ideas and the­o­ries that are untested, un­usual and mi­s­un­der­stood in the main­stream.

There’s a del­i­cate line to walk here. The world has sig­nifi­cant prob­lems, and it may re­quire rev­olu­tion­ary new ideas to solve those prob­lems. En­ter­tain­ing un­usual per­spec­tives and ex­plor­ing ne­glected ar­eas is vi­tal to gen­er­at­ing new ideas and of­ten re­quires a unique cul­ture. But this can also lead to miss­ing cru­cial con­ven­tional wis­dom, putting off po­ten­tial al­lies and can be an easy ex­cuse for poor com­mu­ni­ca­tion.

Given this ten­sion and be­cause we sup­port the work that EA is do­ing, we will do more to en­sure that the way we’re per­ceived doesn’t nega­tively im­pact the EA com­mu­nity. In par­tic­u­lar, Lev­er­age and Paradigm will:

  1. Be more open to feed­back on ways we might be ad­versely af­fect­ing the EA community

  2. Ac­tively seek more ad­vice on our PR and com­mu­ni­cate with EA or­gani­sa­tions ear­lier where we think our work might im­pact other groups

  3. Col­lab­o­rate with EA or­gani­sa­tions if we want to do things like pre­sent more con­tro­ver­sial ideas at EA events and take more se­ri­ously how any­one par­ti­ci­pat­ing in an EA space might be taken by oth­ers to rep­re­sent EA in some way re­gard­less of whether or not they are an EA or­gani­sa­tion.


3.2.4. In­ter­per­sonal interactions

I know some in­di­vi­d­u­als have had in­ter­ac­tions with some mem­bers of staff at Lev­er­age in the past where they’ve felt dis­missed, put down or un­com­fortable. Where this has been the case, it has un­der­stand­ably coloured some peo­ple’s im­pres­sion of Lev­er­age as a whole, and we want to apol­o­gise for nega­tive ex­pe­riences peo­ple have had.

The kinds of con­cerns I’ve heard most fre­quently in­clude:

  1. Peo­ple feel­ing like we were judg­ing them on whether or not they were wor­thy of col­lab­o­rat­ing with, or feel­ing like we were as­sess­ing them on whether or not they would be use­ful in­stead of car­ing about them as a person

  2. Lev­er­age staff ask­ing weird and prob­ing ques­tions which might feel par­tic­u­larly un­safe in con­texts like in­ter­views or when dis­cussing psy­chol­ogy re­search

  3. Lev­er­age staff be­ing over­con­fi­dent when pre­sent­ing their ideas

  4. Lev­er­age staff be­ing dis­mis­sive of other peo­ple’s plans, pro­jects or ideas

  5. Lev­er­age staff gen­er­ally be­ing weird.

Firstly, we want to apol­o­gise for any in­ter­ac­tions peo­ple have had with Lev­er­age 1.0 staff that have made them feel un­com­fortable, judged, looked down on or for times peo­ple thought we were treat­ing them as in­stru­men­tal. We do not see peo­ple this way nor want to make them feel like that. A big part of our work is about un­der­stand­ing peo­ple, car­ing about their prob­lems and sup­port­ing their growth as in­di­vi­d­u­als. We want to help peo­ple in de­vel­op­ing their ideas and their pro­jects, not just be­cause the world needs more peo­ple work­ing to­gether to do good, but also be­cause we gen­uinely care about peo­ple.

Mov­ing for­ward, we want to do a much bet­ter job of ex­plain­ing our ideas and giv­ing a much more ac­cu­rate im­pres­sion of how un­cer­tain we are. I can’t promise you won’t end up dis­cussing weird ideas when in­ter­act­ing with us, but we want this to be en­gag­ing, not off-putting. Feel free to give us feed­back di­rectly when talk­ing to us in the fu­ture or, if you pre­fer, you can email me (larissa@lev­er­agere­search.org) or fill in this form with feed­back or con­cerns.

To wrap up this sec­tion, I will share a cou­ple of thoughts that re­late to some of the con­cerns I’ve heard. I don’t want to make ex­cuses for peo­ple be­ing un­friendly or mak­ing oth­ers feel bad in in­ter­ac­tions, but this might help peo­ple un­der­stand Lev­er­age bet­ter.

One of the ad­di­tional ad­verse effects of our poor pub­lic com­mu­ni­ca­tion is that when Lev­er­age staff have in­ter­acted with peo­ple, they of­ten didn’t un­der­stand our work and had a lot of ques­tions and con­cerns about it. While this was un­der­stand­able, I think it some­times led staff to feel at­tacked which I sus­pect, in some cases, they han­dled poorly, be­com­ing defen­sive and per­haps even with­draw­ing from en­gag­ing with peo­ple in neigh­bour­ing com­mu­ni­ties. If you don’t build up re­la­tion­ships and dis­cuss up­dates to your think­ing in­fer­en­tial dis­tance builds up, and it be­comes easy to see some dis­tant, amor­phous or­gani­sa­tion rather than a col­lec­tion of peo­ple.

I think Lev­er­age also strug­gles with the same challenge the EA com­mu­nity faces when it comes to man­ag­ing both truth-seek­ing and in­di­vi­d­ual wellbe­ing. On the whole, I be­lieve lead­ers in the EA com­mu­nity do a great job of challeng­ing seem­ingly mis­taken ideas with cu­ri­os­ity and kind­ness. I’m sure we at Lev­er­age can do bet­ter on this di­men­sion.

Speak­ing purely from my per­sonal ex­pe­rience, I’ve found the Lev­er­age and Paradigm staff to be very wel­com­ing and em­pa­thetic. My ex­pe­rience of the Lev­er­age cul­ture is one where it feels ex­cep­tion­ally safe to ex­press ideas and be wrong. This sense of safety has benefited my abil­ity to de­velop my mod­els and in­de­pen­dent think­ing. It’s also a place with a strong fo­cus on un­der­stand­ing peo­ple and car­ing about helping them im­prove. I want to en­sure that more peo­ple have this ex­pe­rience when in­ter­act­ing with Lev­er­age.

Mov­ing for­ward, I hope the greater fo­cus on ex­ter­nal en­gage­ment at Lev­er­age 2.0 will give rise to more op­por­tu­ni­ties for peo­ple to have dis­cus­sions and gen­er­ally hang out with the staff at Lev­er­age. I’ll be hon­est, you should still ex­pect us to be pretty weird some­times—but we’re also very friendly.


3.2.5. Lev­er­age’s impact

The fi­nal con­cern I’ll dis­cuss is around what im­pact Lev­er­age has had and whether it has been a good use of re­sources. I don’t think there are mis­takes Lev­er­age has made here be­yond the ones dis­cussed above (e.g. com­mu­ni­ca­tion). In­stead, I think this is just a difficult ques­tion.

If the con­cern is about whether there has been rigor­ous cost-effec­tive­ness anal­y­sis of Lev­er­age’s work, the an­swer is no and, to be hon­est, I don’t think that frame­work makes sense for as­sess­ing this kind of re­search.

If the ques­tion is in­stead, “was Lev­er­age 1.0 gen­er­ally a good use of re­sources?”, my hon­est an­swer is that I don’t know.

Assess­ing the value of re­search, es­pe­cially un­pub­lished re­search out­puts, is a com­plex prob­lem. Sev­eral peo­ple in the EA com­mu­nity have faced this challenge when eval­u­at­ing or­gani­sa­tions like MIRI, FHI, and GPI or eval­u­at­ing ca­reer op­por­tu­ni­ties in AI strat­egy and gov­er­nance (which of­ten in­volves un­pub­lished re­search). My in­tu­ition is that most peo­ple with­out ac­cess to in­sider in­for­ma­tion about the re­search or­gani­sa­tion or with­out the tech­ni­cal abil­ity to as­sess the re­search should con­clude that they don’t know whether a par­tic­u­lar re­search or­gani­sa­tion is a good use of re­sources and, where they need to make some calls, heav­ily defer to those who do have that in­for­ma­tion and abil­ity. There­fore, I sus­pect most peo­ple should similarly con­clude they don’t know if Lev­er­age was a good use of re­sources, given the lack of pub­lished re­search or ex­ter­nal signs of cred­i­bil­ity.

If you did want to try and make some head­way on this ques­tion in the Lev­er­age case, my sug­ges­tion would be to try and think about:

  1. Whether or not you be­lieve so­cial sci­ence re­search into un­der­stand­ing peo­ple and so­cieties seems es­pe­cially cru­cial for world improvement

  2. The de­gree to which it looks as though Lev­er­age 1.0 was suc­cess­ful in bet­ter un­der­stand­ing peo­ple and so­cieties through its re­search.

From my per­spec­tive, this kind of so­cial sci­ence re­search does seem both im­por­tant to many plans for sig­nifi­cantly im­prov­ing the world and gen­er­ally use­ful. How­ever, I ex­pect a lot of dis­agree­ments about the tractabil­ity of the area and its im­por­tance rel­a­tive to other things.

When it comes to how suc­cess­ful Lev­er­age 1.0 was in its re­search, read­ers can gain some in­for­ma­tion by read­ing Ge­off’s es­say se­ries, eval­u­at­ing work we pub­lish in the fu­ture, and as­sess­ing how use­ful the train­ing tech­niques we have de­vel­oped are.

From the in­side, I’m op­ti­mistic that Lev­er­age 1.0 was a good use of re­sources, but with lots of un­cer­tainty, es­pe­cially around how to as­sess re­search in gen­eral. I’ve found Lev­er­age’s in­tro­spec­tion tools and peo­ple mod­els to be di­rectly use­ful, which makes me think their psy­chol­ogy work is promis­ing. It also ap­pears to me that they have a vast amount of high qual­ity, in­ter­nal ma­te­rial on un­der­stand­ing the mind and on meth­ods for mak­ing progress on challeng­ing re­search ques­tions. Many of the staff I’ve in­ter­acted with at Lev­er­age seem to have de­tailed mod­els of Lev­er­age 1.0 re­search ar­eas and seem to have de­vel­oped im­pres­sive skills over their time at Lev­er­age 1.0.

For most peo­ple, I don’t think my in­sider view should sub­stan­tially change their opinion. In­stead, if as­sess­ing Lev­er­age’s im­pact is of par­tic­u­lar in­ter­est to you, I’d sug­gest look­ing for more pub­li­cally-ac­cessible signs of Lev­er­age’s suc­cess or failure in the fu­ture and us­ing that to in­form whether Lev­er­age’s past work was use­ful.


3.3 Feed­back on how we’re doing

Since I will be tak­ing the lead on Lev­er­age’s en­gage­ment with other com­mu­ni­ties, I want to end this post by strongly en­courag­ing read­ers to reach out to me if you no­tice ways that we can do bet­ter on any of these di­men­sions mov­ing for­ward.

You can email me at larissa@lev­er­agere­search.org or fill in this form.

The form asks for, but does not re­quire, a name. In gen­eral, we do pre­fer non-anony­mous feed­back, but we are open to re­ceiv­ing anony­mous in­put if it is con­struc­tive. Your feed­back will provide me with in­for­ma­tion on whether or not Lev­er­age does im­prove on the di­men­sions we’ve laid out here and will help me to work out how we can do bet­ter.

Similarly, if you have ques­tions about any­thing I’ve not been able to cover here or feed­back on this post, please feel free to add it in the com­ments.


Foot­notes:

[1] Edited in re­sponse to feed­back in the com­ments here. Pre­vi­ously this sen­tence read:

While of­ten this was the right trade-off for Lev­er­age 1.0 where the fo­cus was ad­vanc­ing our ideas, this makes the job of com­mu­ni­cat­ing our work mov­ing for­ward challeng­ing.

[2] Th­ese two ad­di­tions were made af­ter the post had been pub­lished in re­sponse to email feed­back I re­ceived point­ing out that I’d for­got­ten to men­tion our past promises to provide up­dates that we didn’t fulfil.