Joan Gass: How to Build a High-Impact Career in International Development

In this aca­demic ses­sion, Joan Gass, who holds a Master’s in Public Ad­minis­tra­tion from the Har­vard Kennedy School, ar­gues that mem­bers of the EA com­mu­nity who want to work in global de­vel­op­ment should adopt a “ven­ture cap­i­tal” ap­proach, look­ing for op­por­tu­ni­ties to make bets on the cause ar­eas with the high­est ex­pected value. In par­tic­u­lar, she con­tends that fos­ter­ing eco­nomic growth in emerg­ing mar­kets and build­ing state ca­pa­bil­ity are high-value ar­eas, while dis­cussing other ar­eas in RCT re­search and so­cial en­trepreneur­ship that are rel­a­tively low in value and should be avoided.

Below is a tran­script of Joan’s talk, which we’ve lightly ed­ited for clar­ity. You may also watch it on YouTube or read it on effec­tivealtru­

The Talk

When I started at the Kennedy School, I was in an in­ter­na­tional de­vel­op­ment-spe­cific policy pro­gram. I ap­proached our pro­gram fa­cil­i­ta­tor at the be­gin­ning of the pro­gram and started talk­ing to him about ca­reer de­ci­sions. And this is what he said to me:
He had some re­ally strong thoughts about ca­reer op­tions and what he thought were the high­est-im­pact things to do.

Some caveats be­fore I be­gin this:

I wrote this pa­per for my global de­vel­op­ment the­sis and did it over a three-to-four month pe­riod. It is speci­fi­cally for stu­dents who are think­ing about ca­reers in global de­vel­op­ment or pub­lic policy. As a re­sult, it doesn’t cover the full spec­trum of what we typ­i­cally talk about within effec­tive al­tru­ism. In par­tic­u­lar, it doesn’t take into con­sid­er­a­tion longter­mist think­ing or an­i­mal welfare. It’s just tar­geted to­ward an au­di­ence of my peers within the pro­gram that I was work­ing in. Also, I don’t think it’s ex­haus­tive. I chose a few ex­am­ples of cause ar­eas within global de­vel­op­ment to illus­trate a method­ol­ogy that I was in­ter­ested in ex­plor­ing, but [those don’t re­flect] an ex­haus­tive search.

So why does it mat­ter what one chooses to fo­cus on within global de­vel­op­ment?

I think we’ve seen some ex­am­ples of things that peo­ple have done within the global de­vel­op­ment space that have been com­plete home runs. [Nor­man Bor­laug] in­vented a key strain of wheat dur­ing the Green Revolu­tion Revolu­tion that saved be­tween 10 mil­lion and 1 billion lives. Elimi­nat­ing smal­l­pox averted 1.5 to 2 mil­lion deaths per year. And there’s some re­search that shows that four out of five Haiti­ans who have emerged from ex­treme poverty have done so be­cause they im­mi­grated out of Haiti. Those are some ex­am­ples of “home runs” from a policy per­spec­tive or global de­vel­op­ment per­spec­tive.

At the same time, there have been ac­tions that have been re­ally nega­tive for the world. For ex­am­ple, Brazil had a slow­down in the 1980s that re­sulted in a cur­rent net loss of $7.5 trillion in value.

When think­ing about cause pri­ori­ti­za­tion, I took the clas­sic effec­tive al­tru­ist [ap­proach of con­sid­er­ing] im­pact, ne­glect­ed­ness, and tractabil­ity, and tried to ad­just it us­ing met­rics ap­pli­ca­ble to the global de­vel­op­ment space. I’ll run through how I thought about [each area] in the con­text of my the­sis.


Ne­glect­ed­ness is about the peo­ple and re­sources cur­rently [de­voted to] an is­sue. I used prox­ies re­lated to World Bank fund­ing, with some mod­ifi­ca­tions, to think about how much money was go­ing to­ward cer­tain spaces. I also did a sur­vey of [Kennedy School] alumni who had been out of my pro­gram for a few years to think about how many peo­ple were work­ing on a given is­sue.

When think­ing about im­pact, I looked to see how much im­prove­ment in hu­man well-be­ing there might be if a prob­lem were solved.

When think­ing about tractabil­ity — which is, I think, one of the fuzzier ar­eas to mea­sure — I con­ducted in­ter­views with ex­perts, looked at his­tor­i­cal ex­am­ples, and, when I was still un­sure, did a thresh­old anal­y­sis. I’ll give an ex­am­ple of that later. [It in­volved de­ter­min­ing] what per­cent con­fi­dence I had in think­ing an area was a good one to bet on. Then, I tried to see if I my anal­y­sis was enough to put me above or be­low that thresh­old.

What recom­men­da­tions [did I ar­rive at]?

Of the uni­verse of op­tions that I looked at, I ended up recom­mend­ing peo­ple fo­cus on:

* New modal­ities to foster eco­nomic pro­duc­tivity
* New modal­ities or ways to de­velop state ca­pa­bil­ities
* Global catas­trophic risks, par­tic­u­larly pan­demic pre­pared­ness
* Meta EA re­search on cause pri­ori­ti­za­tion within global de­vel­op­ment

I’m go­ing to go in-depth on eco­nomic pro­duc­tivity, so I’ll skip over it for now and talk briefly about the last two to just give you a fla­vor of what I did for my the­sis.

One of the rea­sons why I em­pha­size pan­demic pre­pared­ness is that gen­er­ally, pan­demics dis­pro­por­tionately im­pact de­vel­op­ing coun­tries. Think of how Ebola im­pacted Libe­ria. Also, there’s a de­cent amount of ev­i­dence and pre­dic­tions in this area. One that I cited in my the­sis is Larry Sum­mers’ talk about how the an­nual cost of ex­pected pan­demics in the next few years, in terms of nega­tive im­pli­ca­tions on hu­man well-be­ing, is com­pa­rable to a range of out­comes you could see from cli­mate change. There’s a pretty limited amount of fund­ing cur­rently go­ing into this space, par­tic­u­larly in re­gards to the “tail risk” sce­nar­ios that we of­ten talk about within EA.

Plus, I think there’s a lot of po­ten­tial for [pan­demic pre­pared­ness] to be tractable. A few days af­ter the Open Philan­thropy Pro­ject helped fund a Blue Rib­bon Study Panel on biodefense, live an­thrax was ac­ci­den­tally shipped across the coun­try.

In terms of meta re­search [on effec­tive al­tru­ism], there are some ar­eas that I would have loved to in­ves­ti­gate if I’d had more time. Th­ese in­clude:

* The “80,000 Hours ”ap­proach to global de­vel­op­ment
* Promis­ing cause ar­eas like im­mi­gra­tion, im­prov­ing se­cu­rity in low-in­come ar­eas (theft and in­se­cu­rity are among the top drivers of suffer­ing in low-in­come ur­ban ar­eas), and gen­der-based vi­o­lence (gen­der is one of the biggest drivers of vi­o­lence around the world)

There is definitely a lot more re­search to be done if folks are in­ter­ested.

What about ar­eas that I ei­ther [felt only mod­er­ately ex­cited] about or didn’t end up recom­mend­ing? I ended up in the “medium” range with global health, which might be sur­pris­ing for an EA au­di­ence. I think ad­dress­ing global health prob­lems is clearly highly im­pact­ful. There’s a lot of work left to do. But it’s not quite as ne­glected from a fund­ing per­spec­tive. So, maybe an ad­di­tional marginal per­son who has in­no­va­tive ideas might make slightly less of a differ­ence than in some other cause ar­eas that are high-im­pact, tractable, and more ne­glected.

I think global health is still a pretty solid bet; it de­pends on your po­si­tion. It’s also clearly very high in terms of tractabil­ity, be­cause we have a lot of his­tor­i­cal ex­am­ples of wins in this space. I recom­mend you look at global health, par­tic­u­larly if you’re more risk-averse in your ca­reer.

What about ar­eas that I didn’t recom­mend in my the­sis?

Par­tic­u­larly for peo­ple in my pro­gram [at the Kennedy School], I recom­mended think­ing care­fully be­fore go­ing into so­cial en­trepreneur­ship. There are a lot of defi­ni­tions [of so­cial en­trepreneur­ship]; [for my pur­poses] it means in­di­vi­d­u­als try­ing to sell a product that has a dou­ble bot­tom line [i.e., is both prof­itable and has a pos­i­tive so­cial im­pact]. I think that in situ­a­tions in which you dis­place a gov­ern­ment from pro­vid­ing a pub­lic good, that can have longer-term [nega­tive] im­pli­ca­tions.

One ex­am­ple of this is a so­cial en­ter­prise in In­dia that pro­vided pri­vate, clean wa­ter for rich house­holds in ur­ban ar­eas. That took away pub­lic de­mand for sys­tems change to bring clean wa­ter to the en­tire neigh­bor­hood, in­clud­ing to lower-in­come house­holds. There­fore, I have con­cerns with this model when en­trepreneurs are tar­get­ing monopoly-pro­vided ser­vices.

In gen­eral, I think there are also con­cerns within my pro­gram that, for folks who are rigor­ously trained in eco­nomics and data anal­y­sis, so­cial en­trepreneur­ship [may not give them a] com­par­a­tive ad­van­tage or [re­flect] what they’re uniquely po­si­tioned to do. I think we might have overem­pha­sized it a bit [in the past] due to its pres­tige and sta­tus. I did some anal­y­sis of Echo­ing Green Fel­lows’ like­li­hood of suc­cess. When I con­sid­ered their im­pact 10 years fol­low­ing their fel­low­ships, it was lower than a lot of my peers ex­pected. (Echo­ing Green is a ven­ture firm that funds so­cial en­trepreneurs.) For those rea­sons, I’m a lit­tle bit hes­i­tant, on av­er­age, to recom­mend so­cial en­trepreneur­ship.

Another [po­ten­tially con­tentious ques­tion cen­ters on] whether run­ning ad­di­tional RCTs is benefi­cial. [For my the­sis], I was in­ter­ested in not just the value of RCTs in gen­eral, but the value of an ad­di­tional marginal per­son work­ing in this space. Of course, in any given space, there’s var­i­ance. I think there’s prob­a­bly re­ally high value for some­one to do a lon­gi­tu­di­nal study around the long-term effects of de­worm­ing. [EA-al­igned fun­ders in global de­vel­op­ment would likely] pay a lot of money for that, and I would re­ally love for some­one to do that. But for an av­er­age per­son en­ter­ing this space, run­ning an av­er­age RCT, what do we think the ad­di­tional value is?

I came to the con­clu­sion that the value of re­search might not be su­per-strong for RCTs be­cause of con­sid­er­a­tions re­lated to ex­ter­nal val­idity. We might get higher-value re­search on some macro-level is­sues and from sys­tems change work or di­ag­nos­ing ad­minis­tra­tive and poli­ti­cal con­straints. I could take 20 min­utes to talk about this point, but I’m just go­ing to ac­knowl­edge that it’s a hot topic within EA.

One way to sum­ma­rize this is to say that within the global de­vel­op­ment com­mu­nity, we’ve thought a lot about ser­vice de­liv­ery that’s not at scale.


For ex­am­ple, what do we think about vo­ca­tional train­ing, cash trans­fers, or ac­tivi­ties within an NGO con­text? A lot of what I ended up recom­mend­ing in my the­sis is [re­lated to the idea] that there’s value in think­ing about how we move from ser­vice de­liv­ery that’s not at scale to ser­vice de­liv­ery at scale. How do we iter­ate to make ser­vice de­liv­ery at scale hap­pen in the first place? And how do we fa­cil­i­tate broader eco­nomic growth?

I’m ad­vo­cat­ing for some­thing that might look more like a hits-based ap­proach to global de­vel­op­ment [ver­sus one fo­cused on mak­ing marginal im­prove­ments, as we’ve done in the past].
In this di­a­gram [see slide above], there are two over­lap­ping graphs. We have a risk-re­turn anal­y­sis show­ing that [higher-risk] ac­tivi­ties can have a higher re­turn — but also po­ten­tially a higher var­i­ance of suc­cess. I’ve ar­gued that in the global de­vel­op­ment space, there’s been a lot of value in think­ing about bet­ter bets in the micro space. There’s a de­cent amount of ev­i­dence show­ing that, on av­er­age, some of the benefits we thought ac­tivi­ties in microfi­nance [would bring] aren’t as strong as those from GiveDirectly [which gives cash di­rectly to peo­ple liv­ing in ex­treme poverty to use as they see fit]. We’ve definitely pushed the effi­ciency fron­tier for­ward.

But I’m also ar­gu­ing that there might be cer­tain macro things, like eco­nomic growth, that have a wide var­i­ance of out­comes. Some won’t work at all. And some we might knock out of the park. If we fo­cus on these higher var­i­ances, we might see a higher re­turn over­all.

I’m go­ing to go a bit more in-depth for one or two min­utes.
As I was think­ing about eco­nomic growth, I was an­a­lyz­ing the im­pact at stake. If you look at ac­cel­er­a­tions in eco­nomic growth — times when coun­tries’ economies have re­ally taken off com­pared to what we ex­pected them to do — [those pe­ri­ods have] have cre­ated $20 trillion in to­tal value. To put that in con­text, that’s the cur­rent an­nual GDP of Nige­ria mul­ti­plied by 60 — a re­ally sig­nifi­cant amount of value. Imag­ine if we could figure out how to foster or ex­tend these growth ac­cel­er­a­tions. This is an area where ques­tions around ne­glect­ed­ness are de­bated. There’s a de­cent amount of gen­eral World Bank fund­ing that’s go­ing into [pro­mot­ing] eco­nomic growth, but a pretty limited amount sup­ports more novel meth­ods. Those are the ar­eas that I’m much more in­ter­ested in.

There are huge de­bates within eco­nomics about whether or not this is tractable. Is it even pos­si­ble to foster eco­nomic growth? In my the­sis, I tried to provide some ex­am­ples show­ing why I thought it was tractable and why I am re­ally ex­cited about new meth­ods for fos­ter­ing eco­nomic growth.
This fuzzy di­a­gram [see slide above] is an ex­am­ple of a growth di­ag­nos­tic. It shows ways in which you can try to iden­tify the bind­ing con­straints or bot­tle­neck for eco­nomic growth to take off in Nepal. Sev­eral differ­ent folks within the gov­ern­ment and the pri­vate sec­tor have said that once they had this di­ag­nos­tic and were aware of the challenges, they were able to fo­cus their poli­ti­cal cap­i­tal, ad­minis­tra­tive cap­i­tal, and other re­sources in [re­sponse]. And I think this is an ex­am­ple of a method­ol­ogy that could be po­ten­tially re­ally high-im­pact.

Another con­ver­sa­tion I had was around whether we are con­fi­dent enough in any given [at­tempt to spur] eco­nomic growth to make it worth­while. I did some anal­y­sis of how much money it would take to try out one of these ap­proaches [rel­a­tive to the amount of] value we ex­pected to get from it. What would the differ­ence be be­tween try­ing one of these new meth­ods and just giv­ing money di­rectly to [peo­ple in need]?
My ex­am­ple use case was 0.1% growth in GDP in Ethiopia last­ing five years. The anal­y­sis showed that if there is a 2.5% chance that we can get some­thing right in terms of eco­nomic growth, then it’s worth a $5 mil­lion in­vest­ment, be­cause that’s the equiv­a­lent amount of value we would get from donat­ing $193 mil­lion to GiveWell. So then my ques­tion is: Do we think there’s a 2.5% chance that we could ac­tu­ally achieve this out­come? We could do fur­ther re­search around case stud­ies and ex­pert in­ter­views to an­swer that.

Fi­nally, what do we do with all of this? And at the very end of my the­sis, I tried to recom­mend what peo­ple who have gone through this anal­y­sis might do next. Maybe you’ve ar­rived at similar con­clu­sions — or maybe you’ve used the same method­ol­ogy and ar­rived at differ­ent con­clu­sions. [Re­gard­less], I think that there’s po­ten­tially a four-step pro­cess that you could use.
Step 1: Develop your im­pact hi­er­ar­chy. Try to be cause-neu­tral. Think about us­ing some­thing like im­pact, ne­glect­ed­ness, and tractabil­ity to iden­tify what what you think the most im­por­tant prob­lems are to work on. Rank them.

I think that in the global de­vel­op­ment space, you should take your na­tion­al­ity into ac­count be­cause that gives you spe­cific lev­er­age on cer­tain is­sues, es­pe­cially if you’re think­ing about gov­ern­ment work.

This is just an illus­tra­tive ex­am­ple — not my par­tic­u­lar cause pri­ori­ti­za­tion — of what an im­pact hi­er­ar­chy could look like.

Step 2: Map out your per­sonal fit. For ex­am­ple, this hy­po­thet­i­cal per­son in my ex­am­ple is from In­dia and re­ally ex­cited about microfi­nance there. It’s some­thing that is a core com­pe­tency and that they’ve done be­fore. Maybe they think biose­cu­rity is re­ally im­por­tant, too, but they would need to be in the U.S. to do that. They want to be close to fam­ily, so it’s out­side of their per­sonal fit. Per­sonal fit, lo­ca­tion, skills, and in­ter­ests could over­lap with some­one’s pri­ori­ti­za­tion of a few causes — or they might nott.

Step 3: Test the high­est-im­pact op­tion that over­laps with your per­sonal fit. Or maybe if you’re in grad school, or have some free time and run­way, you might test an op­tion that’s just out­side your bound­ary. Think about ar­eas you could and couldn’t stretch to. All of these steps al­low you to get more data on where you might be a good fit.

Step 4: The last thing that I think you can do is back­wards-en­g­ineer how to have a lot of in­fluence in a given area. You can look at peo­ple who have been re­ally suc­cess­ful and in­fluen­tial in that field and fol­low [a similar] ca­reer path. Ideally, you have a va­ri­ety of ca­reer paths to choose from, and then you can think about what your next steps might be to get to a place where you will have a max­i­mum amount of in­fluence in that field.

Also, you can down­load my pa­per.
A warn­ing: It is 50 pages long, but I also in­cluded a four-page policy overview. I would re­ally love com­ments and feed­back. This was my at­tempt to take some method­olo­gies that we talk about within effec­tive al­tru­ism and ap­ply them to a field in which I have back­ground knowl­edge and the op­por­tu­nity to re­flect on it. But I think there’s a lot of work still to be done in this area.

James Snow­den [Moder­a­tor]: Thanks so much, Joan, for that in­ter­est­ing talk. I’ll just give a very brief sum­mary of what I took from the talk and then I’ll ask you some ques­tions.

Joan an­a­lyzed a num­ber of differ­ent op­tions for smart grad­u­ates work­ing in in­ter­na­tional de­vel­op­ment us­ing a frame­work mea­sur­ing im­por­tance, ne­glect­ed­ness, and tractabil­ity.

She con­cluded that four ar­eas look par­tic­u­larly promis­ing:

1. New ways to foster eco­nomic pro­duc­tivity
2. New ways to de­velop state ca­pa­bil­ities
3. Global catas­trophic risks (in par­tic­u­lar, pan­demic pre­pared­ness)
4. Meta re­search on tal­ent in global health and de­vel­op­ment

Joan also out­lined a prac­ti­cal, four-step pro­cess for choos­ing a ca­reer:

1. Deter­mine which ar­eas you think are most im­por­tant to work on
2. Iden­tify your own per­sonal fit
3. Challenge your­self to test the high­est-im­pact op­tion, stretch­ing within the bound­aries of your per­sonal fit (which I think is very im­por­tant)
4. Develop a plan to reach max­i­mum in­fluence on the is­sue

Joan, I think you’ve done a re­ally great job of sur­vey­ing a num­ber of promis­ing ar­eas and de­vel­op­ing prac­ti­cal heuris­tics for mak­ing good de­ci­sions about your ca­reer. My first ques­tion is: If you were some­one sit­ting in this room who wanted to build on your re­search and make it more ro­bust or ex­pand the scope, what kind of ques­tions would you be ask­ing? And what would be the most prac­ti­cal way to do that?

Joan: I think that there are ways that the method­ol­ogy could be more ro­bust. I used a few prox­ies that I men­tioned at the be­gin­ning of the talk — for ex­am­ple, World Bank fund­ing as a proxy for all global de­vel­op­ment fund­ing. You could do a more pre­cise es­ti­mate of that. And I was look­ing speci­fi­cally at grad­u­ates of my pro­gram. Those grad­u­ates aren’t fully rep­re­sen­ta­tive of the full set of grad­u­ates and ex­pe­rienced re­searchers in the field of de­vel­op­ment policy. So there’s definitely more method­olog­i­cal depth [that some­one could add].

I also think there are a lot of ad­di­tional ar­eas that I didn’t have the scope or ca­pac­ity to ex­plore — not only more ap­proaches to how peo­ple might do this meta re­search, but on top­ics that I think would be par­tic­u­larly in­ter­est­ing to ex­plore fur­ther, like mi­gra­tion and vi­o­lence.

James: You men­tioned a few case stud­ies as ex­am­ples of some ar­eas you were look­ing into. I was par­tic­u­larly in­ter­ested in the case stud­ies on growth and whether there are ex­am­ples of par­tic­u­lar or­ga­ni­za­tions that might’ve played a causal role in the story there, which might help us think about where we could have an im­pact.

Joan: In prin­ci­ple, I think there’s a lot more high-value, in­ter­est­ing re­search to be done on growth. What causes growth epi­sodes to take off? How does a frag­ile state or a con­flict zone be­come more sta­ble?

I don’t know if we’re ever go­ing to be able to show [with com­plete con­fi­dence] that an in­di­vi­d­ual or or­ga­ni­za­tion [caused] growth. But some in­ter­est­ing ex­am­ples that I was look­ing at (and would have con­tinued to look at if I’d had time) in­clude the Korean Devel­op­ment In­sti­tute. That is a re­ally in­ter­est­ing ex­am­ple. It’s a think tank that helped Korea de­velop a growth strat­egy. And in the 1980s, there was a similar or­ga­ni­za­tion in In­done­sia, where a group of PhD economists helped ad­vise the gov­ern­ment. Some were [In­done­si­ans who already worked] in the gov­ern­ment. So, I think we have a cou­ple of his­tor­i­cal ex­am­ples that at least point to smart folks who were think­ing about eco­nomic growth policy and pro­vided in­put at the right time. That could be promis­ing [to ex­plore].

James: One last ques­tion from me be­fore I hand it over to the au­di­ence. You men­tioned some new ap­proaches [to en­abling] states’ ca­pa­bil­ities and eco­nomic growth. You sep­a­rated those from older ap­proaches. I’m cu­ri­ous about ex­am­ples of par­tic­u­larly promis­ing new ap­proaches, maybe from the last 10 years, which we might not know about and which you think could be par­tic­u­larly good ar­eas to work in.

Joan: I think there’s value in both [older and newer ap­proaches]. Within the eco­nomic growth space, I was par­tic­u­larly in­ter­ested in this “growth di­ag­nos­tic” ap­proach — this idea of not just do­ing what­ever pro­motes growth [and is fea­si­ble]. Let’s tar­get the things that we think are likely to pro­mote the _most_ growth. I think maybe that’s why some of this work has had a mixed track record in the past. We could try tak­ing a new ap­proach to state ca­pa­bil­ities, gov­er­nance, and the de­liv­ery mechanism. I think there’s a lot of re­ally in­ter­est­ing work around fa­cil­i­tated emer­gence [fo­cus­ing on prob­lems in­stead of solu­tions and, in­stead of fol­low­ing a set plan, us­ing a pro­cess that al­lows for iter­a­tion and learn­ing].

His­tor­i­cally, there have been a lot of peo­ple who [travel to a de­vel­op­ing area] and con­duct a week­end or week-long train­ing [meant to pro­mote growth-friendly policy]. Then they’ll leave and ev­ery­thing isn’t fixed. I think there are a lot of re­ally in­ter­est­ing ques­tions around how to foster lo­cal tal­ent de­ci­sion-mak­ing and build on ca­pac­i­ties that already ex­ist in a longer-term, sus­tain­able way. There’s also a re­ally in­ter­est­ing ques­tion of get­ting smart, ca­pa­ble peo­ple who are na­tion­als of their coun­try fur­ther in­volved in civil ser­vice and mak­ing a ca­reer that’s re­ward­ing for them. Those would be ar­eas that I would fur­ther ex­plore.

James: Got it. Great. We have a few ques­tions from the au­di­ence. One au­di­ence mem­ber asks: If you’re us­ing cli­mate change as a cost com­par­i­son across cause ar­eas, is cli­mate change it­self also worth con­sid­er­ing as a cause area?

Joan: Yeah. Ab­solutely. I have a pretty small para­graph in my the­sis in which I ask: Why isn’t cli­mate change in this bracket? I think the im­pact is com­pa­rable to pan­demics based on the re­search that I saw. But cli­mate change is rel­a­tively less ne­glected in terms of fund­ing. I think there are peo­ple who would dis­agree with that. Some peo­ple would say that the tail-end risks of cli­mate change are much, much worse than pan­demics. I think there’s a lot of re­ally in­ter­est­ing re­search that should be done on cli­mate change. For ex­am­ple, what are the most effec­tive lev­ers? There’s a lit­tle bit from Pro­ject Draw­down on pri­ori­tiz­ing in­ter­ven­tions, but from a policy per­spec­tive, I think that’s a re­ally fruit­ful ques­tion.

James: That leads into our next ques­tion: What was the re­sponse from your col­leagues at the [Kennedy School] to this frame­work? Were they mostly re­cep­tive or did you re­ceive a lot of crit­i­cism?

Joan: I think peo­ple were gen­er­ally re­cep­tive. Some of the con­clu­sions that I make are judg­ment calls that not ev­ery­body agrees with. And that af­fects tractabil­ity. I also think that there’s this in­ter­est­ing dy­namic that I saw more with the Kennedy School ca­reer group: By the time peo­ple get to policy school, some have already spe­cial­ized for sev­eral years in a par­tic­u­lar cause area within global de­vel­op­ment. They are open to the in­tel­lec­tual ex­er­cise of com­par­ing cause ar­eas, but maybe aren’t open to the per­sonal im­pli­ca­tions of chang­ing ar­eas within the field. Some­times I’d have con­ver­sa­tions [in which I said,] “Pre­tend that you are just start­ing over again. How would you feel about some of these ques­tions?”

James: Did you use a similar four-step pro­cess to figure out your own per­sonal ca­reer path and if so, where has it led you?

Joan: Yeah, that’s funny. Hon­estly, a lot of this was a per­sonal in­tel­lec­tual ex­er­cise for me. I came into the Kennedy School be­liev­ing that the top thing that I could do, es­pe­cially given my na­tion­al­ity as an Amer­i­can, was to scale up an ev­i­dence-based pro­gram. I made this pretty large re­vi­sion; I now be­lieve that some of the other paths I out­lined as top recom­men­da­tions might be higher im­pact.

And so, the the­sis was a jour­ney for me that al­lowed me to work through that and the im­pli­ca­tions for my ca­reer. Another way to phrase that is my im­pact hi­er­ar­chy was in flux. I came in with a cer­tain ver­sion of my im­pact hi­er­ar­chy and it changed over the course of my time at the Kennedy School.

That has led me to take the meta ap­proach. I’m in­ter­ested in EA com­mu­nity build­ing. So I think I took a step away from some of these [other op­tions] in part be­cause a lot of the meta ques­tions are su­per im­por­tant. I wish more folks were fo­cus­ing on them.