Enjoyed the episode? Want to listen later? Subscribe by searching 80,000 Hours wherever you get your podcasts, or click one of the buttons below:

We’re a bit less constrained by kind of generally interested, talented people, and a bit more constrained by either people who have very particular skills that are needed, such as AI technical safety, or grantmaker skill sets — the kinds of things we list on our priority problems.

Benjamin Todd

In the last ’80k team chat’ with Ben Todd and Arden Koehler, we discussed what effective altruism is and isn’t, and how to argue for it. In this episode we turn now to what the effective altruism community most needs.

According to Ben, we can think of the effective altruism movement as having gone through several stages, categorised by what kind of resource has been most able to unlock more progress on important issues (i.e. by what’s the ‘bottleneck’). Plausibly, these stages are common for other social movements as well.

  • Needing money: In the first stage, when effective altruism was just getting going, more money (to do things like pay staff and put on events) was the main bottleneck to making progress.
  • Needing talent: In the second stage, we especially needed more talented people being willing to work on whatever seemed most pressing.
  • Needing specific skills and capacity: In the third stage, which Ben thinks we’re in now, the main bottlenecks are organizational capacity, infrastructure, and management to help train people up, as well as specialist skills that people can put to work now.

What’s next? Perhaps needing coordination — the ability to make sure people keep working efficiently and effectively together as the community grows.

The 2020 Effective Altruism Survey just opened. If you’re involved with the effective altruism community, or sympathetic to its ideas, it’s a great thing to fill out.

Ben and I also cover the career implications of those stages, as well as the ability to save money and the possibility that someone else would do your job in your absence.

Get this episode by subscribing to our podcast on the world’s most pressing problems and how to solve them: type 80,000 Hours into your podcasting app. Or read the transcript below.

Producer: Keiran Harris.
Audio mastering: Ben Cordell.
Transcriptions: Zakee Ulhaq.

Key points

Bottlenecks for the world’s most pressing problems

It’s better not to think of things being generally talent constrained or not. It’s better to instead think in terms of the particular types of profiles of person that are constraining different problems. And then, yeah, we would say for most of the problems that we focus on, I mean, we partly selected them precisely for this, are very constrained by certain types of profile of person.

One we’ve given the example of in the past is in AI safety research. There’s lots of groups which have a lot of money, such as DeepMind and OpenAI and even nonprofits in the space have extra funding to hire people. And there’s lots of academic positions these days. So if you’re kind of good enough at doing that, that career path, there’s lots of groups that would fund you and it’s kind of, anyone above that threshold can usually get a job. So that seems like in that case it’s very constrained by someone who has the skills at a sufficient level to do AI technical safety research.

The outside institutions argument

This is the idea that if another person comes along who’s really motivated to work on these issues and has a really effectiveness-focused mindset but we don’t have any funding, that person can just go and work in governments. They could try and work and get funding in academia. They could go and work at another foundation and try to work on relevant issues from that foundation. They could go and work at a nonprofit.

So to some extent, there are already existing ways that you can get funding to work on these issues, at least to some extent. So this just kind of means like, well, even if very little funding was kind of directly aimed at these issues or in effective altruism we didn’t have much money, but if we had lots of great people, they could all still go and do useful things. So this is a kind of general reason why getting an extra person often seems kind of like that leads to more progress than getting an extra $100,000, which is enough to pay for one or two persons’ salaries.

The value of the most productive people

It seems like often the most skilled people in a career path or in an area have a lot more impact than the median. Let’s just take that as a given. Obviously that’s a big topic about how much people do actually differ in their productivity or output or impact in different fields.

But let’s just assume there are large differences. That can create talent constraints, if there’s some constraints on salaries, which I just kind of said, there are a bunch of constraints on salaries, like around culture and university, people being on fixed scales. And so when you do have some constraints like that, it kind of means that the most productive people get paid much less than they’re actually producing. And so then anyone extra like that is super valuable to you because they have this big, extra impact that you’re not paying for. And that kind of thing seems to create these big differences.

And it’s most obvious in something like academic research where often many researchers will be paid the exact same amounts, but it still seems like even those who are paid the same, one might be producing several times more papers than another.

And we interviewed a bunch of biomedical researchers about how much they felt talent versus funding constrained, and lots of them said, “Yeah, well I think the best researchers often produce way more,” and they didn’t say this, but effectively they have to pay them the same. So an extra person like that really contributes a lot more to their lab. So this is an argument why anyone who’s kind of unusually good at an area, we’re constrained by more people like that.

How the situation has evolved over time

Really early in the effective altruism [movement], say like in 2012, there was a bunch of interested people, but none of us really had much money. So it really did seem like an extra $100,000 would have… I mean, when we first started working at 80,000 Hours, our salary was like £15,000 a year. So an extra $100,000 could have paid four of our salaries. So that really seemed like, “Wow, we’re like very funding constrained.”

It seemed like money was much harder to come by, so therefore much more useful. But then we kind of went into the second stage where then there was Open Philanthropy and there’s other billionaires who were somewhat interested. And so like Reid Hoffman’s donated to Global Priorities Institute and a bunch of wealthy people got interested in effective altruism. Some people went to earn to give and they did well, and then they started funding organizations in the community, organizations working on these problems.

And so then although money is still a bottleneck and is always useful, it became a bit less so. And then instead there was this kind of phase when here was lots of money and not as many people around to do things. And so then there was a sense in which we were pretty constrained by maybe just this kind of general junior talented people stage. But then it seems like in the last couple of years, there has been a shift where, if you’re just fresh out of university and you don’t have any particular skills or training in these areas, it’s maybe even a little bit harder to get a job than it would have been, say, in 2016.

And so now in this third stage, we’re a bit less constrained by kind of generally interested, talented people, and a bit more constrained by either people who have very particular skills that are needed, such as we used the AI technical safety example earlier, or grantmaker skill sets, the kinds of things we list on our priority problems. Or maybe we’re more constrained now by what you might want to call an organizational bottleneck, which is the ability to figure out who’s interested. So there’s a kind of searching/vetting bottleneck, and figure out who would be able to contribute and then train them, manage them. And even just have things that lots of people could do.

Replaceability

One thing that can happen is that you might just be better at the job than the person who would have done it otherwise. And it just seems like empirically the case to me that often in hiring rounds, like for organizations in these areas, even among the top couple of candidates, there do seem to sometimes be quite large differences in how much impact they would be expected to have at the organization. And so if we think the top candidate might have like … It wouldn’t be crazy that they would have twice as much impact as the marginal candidate. In which case, at worst case replaceability would only reduce their impact by 50%.

I think the bigger point is it seems like there are often pretty big differences, even when there’s been lots of applicants. This is actually exactly what we would expect if we think people skills is again this heavy tailed or log normal distribution, you’d expect the differences among the very top people are actually larger than the differences among the median people. And so a job application round can be really competitive in the sense that hundreds of people applied, but that doesn’t necessarily mean that the top two candidates were the same.

Transcript

Rob’s intro [00:00:00]

Robert Wiblin: Hi listeners, this is the 80,000 Hours Podcast, where we have unusually in-depth conversations about the world’s most pressing problems and what can be done to solve them. I’m Rob Wiblin, Director of Research at 80,000 Hours.

Today we’re bringing you the final installment of a series of 3 conversations between Arden Koehler, and our CEO, Ben Todd.

In their last episode, they discussed what effective altruism is and isn’t, and how to argue for it. In this episode they turn to what the effective altruism community most needs.

According to Ben, we can think of the effective altruism movement as having gone through several stages:

  • In the first stage, more money (to do things like pay staff and put on events) was the main bottleneck to making progress.
  • In the second, it was talented people willing to work on whatever seemed most pressing.
  • And in the third stage, which Ben thinks we’re in now, the main bottlenecks might be organizational capacity, infrastructure, and management to help train people up, as well as specialist skills that people can put to work now.

Arden and Ben also cover the career implications of those stages, as well as the ability to save money and the possibility that someone else would do your job in your absence.

They focused on analysing these concepts within the context of the effective altruism community and its ideas, though they do think they apply more broadly.

Just a warning that this conversation includes more jargon than usual – so if you’re not familiar with the effective altruism community and its ideas, it might not be the episode for you.

But given that this is an episode focused on the effective altruism community, this is the perfect opportunity to let you know that the 2020 Effective Altruism Survey just opened.

If you’re involved with the effective altruism community, or sympathetic to its ideas, it’s a great thing to fill out.

It’s something of a census for the ‘effective altruism’ community, helping to figure out what people believe about how to do the most good, and what they’re up to.

All the questions are optional but it asks about who you are, how you got involved, how helpful you’ve found the community, your views on what’s most impactful, and a few other things.

So if you’d like to make sure your views, projects and experiences are included, click through the link in the show notes.

Oh and just a reminder that even though he’s our CEO, these are chat episodes so the things Ben says shouldn’t be taken as ‘official 80,000 Hours’ positions or anything like that.

Alright, without further ado, I bring you Arden and Ben.

Bottlenecks for the world’s most pressing problems [00:02:16]

Arden Koehler: Hi listeners. I’m Arden. I’m a researcher at 80,000 Hours.

Benjamin Todd: Hi, and I’m Ben, the CEO of 80,000 Hours.

Arden Koehler: So Ben has thought a lot about what kinds of careers, what kinds of work people can do to make the most difference they can on the world’s most pressing problems. Back in 2015, Ben, you wrote an article saying that you thought a lot of the most pressing problems in the world were talent constrained as opposed to money constrained. So what did you mean by that and how have things changed, if at all?

Benjamin Todd: Yeah, let’s, I guess, maybe zoom out a bit and talk about what is a talent constraint. You can think that different global problems could be held back by different bottlenecks. Or you can think if you add a unit of some type of resource to a problem, how much progress and extra value gets created. Then you can kind of ask, would an extra unit of money or an extra unit of a person working on the problem, or maybe even more abstract things like maybe an area is constrained by ideas or political capital, or you can think about all these types of resources, and because they can’t be perfectly converted into each other, you can end up with a situation where a problem is more constrained by one than another.

Benjamin Todd: So if we want to have a big impact, we want to be trying to ask which resources are most constraining different group of problems, and then we can try and get those resources to those problems.

Arden Koehler: Just to clarify, when you talk about resources constraining different problems, you mean like solving those problems is like we can’t solve the problem because we don’t have enough of a certain kind of resource.

Benjamin Todd: Yep.

Arden Koehler: Okay. So talent constraint. What does that mean in particular?

Benjamin Todd: If a problem is talent constrained, that would mean that getting an extra year of labor on that problem would have an unusually big impact compared to one of the other resources. We most naturally compare it to money because money is easily quantifiable. One way to make the comparison very concrete would be, okay, so imagine one scenario is you earn to give and then donate to the problem. Or another scenario is that you work on the problem. And then the question is which one has the bigger impact. If it’s the working one, then it’s quite natural to say that it’s talent constrained. Yeah, that’s not a perfect definition, but it gives us a rough idea of what we might be talking about.

Arden Koehler: That’s kind of assuming that somehow you’re just as good at earning to give or something as you would be at doing the direct work on the problem.

Benjamin Todd: Yeah. This is a big problem with introducing the term talent constraints, is there is no single talent out there. Everyone has different skills and strengths, and so it’s better to think of that as, “Well, there’s all kinds of different skills that people have, and different problems or constraints most by different skill sets or profiles of people rather than kind of talent in general.”

Arden Koehler: Okay.

Benjamin Todd: Yeah. I mean, economists make models where there’s labor and capital and those are the only two inputs, but in the real world it’s way more messy than that.

Arden Koehler: Okay. So when you said that you thought the most pressing problems were more talent constrained, or I guess maybe constrained by the people being able to contribute certain kinds of skills, can you give a specific example of when you think that was the case?

Benjamin Todd: What I’m trying to say is it’s better not to think of things being like generally talent constrained or not. It’s better to instead think in terms of the particular types of profiles of person that are constraining different problems. And then, yeah, we would say for most of the problems that we focus on, I mean, we partly selected them precisely for this, are very constrained by certain types of profile of person.

Arden Koehler: Right, because obviously our role is to try to get people with those skills working on those problems or get people to develop those skills and work on those problems.

Benjamin Todd: Yes. One we’ve given the example of in the past is in AI safety research. There’s lots of groups which have a lot of money, such as DeepMind and OpenAI and even nonprofits in the space have extra funding to hire people. And there’s lots of academic positions these days. So if you’re kind of good enough at doing that, that career path, there’s lots of groups that would fund you and it’s kind of, anyone above that threshold can usually get a job. So that seems like in that case it’s very constrained by someone who has the skills at a sufficient level to do AI technical safety research.

Arden Koehler: So in the case of AI safety, you said before they’re not sort of perfectly fungible money and talent. Can you just give us a sense of why can’t DeepMind just spend money to train people or recruit the best people in the world or whatever in order to get people to staff it safety teams?

Benjamin Todd: Yeah. It’s a good question. I think one big factor is that it’s pretty hard to do research into something unless you really care about the thing. And yeah, like partly, you need to be really interested in the topic. But then I think in particularly with something like AI safety research, people generally have the sense that someone who really, really cares about actually figuring out what will really help to make things safe, will be able to pick much more effective research topics than someone who’s just like, “Well, I’m like intellectually interested in this thing and it’s kind of in the area and I’m just going to kind of work on that.”

Benjamin Todd: And then the trouble is, by having more money, you can’t kind of magically make people suddenly deeply care about existential risk or like AI alignment. I mean, you can a little bit, because there’s going to be people kind of on the margin and maybe they switch if it’s easier to get a job in the area.

Arden Koehler: Or like you spend some money on educational campaigns to try to convince people that this is a really important issue.

Benjamin Todd: Yes. Yeah. And then, or you could run like an AI fellowship, which Open Philanthropy has done. But generally these things take a while. So you can end up in a situation where there’s a temporary mismatch between how much funding is available and how many people have the right profile to take the thing. And in that period, it would be constrained by that skill profile.

Arden Koehler: I guess it sounds like this is partly because we think some of these issues are urgent, right? If we had all the time in the world, maybe the money would be really fungible, or we could spend in order to get the right portfolio of people later because you’d have a long time to get people interested in this stuff, get people educated in various ways. But because we want people to be working on this stuff in the next few years, it’s much harder.

Benjamin Todd: Yes. The less time you have, the more you can get differences between different types of resources.

Arden Koehler: Okay. So then what are the arguments that many of the most pressing problems are talent constrained in particular?

Benjamin Todd: Yeah. All the problems that we focus on. Yeah, I think we should probably stop using the word talent constraints because as I was saying, it’s too general. It’s better to instead talk about very specific needs, like specific types of personal skills or whatever the… however you wants to cache it out, the different problems need. And yeah, I think this has caused quite a bit of confusion.

Benjamin Todd: One example of that is that when people think about talent constraints, they tend to think of just like, “Oh, someone who’s like generally talented, but doesn’t have any kind of particular skills or particularly unusual thing about them”. They’re just kind of like a generalist useful person. And there’s actually been a thing recently where I’d say many of the problems we highlight are not constrained by young generalists, they’re more constrained by people with a particular bit of expertise and also people who are unusually motivated by these issues and have a kind of unusually effective altruist mindset and a couple of other things that are quite rare in general.

Arden Koehler: Yeah. I guess the example that you gave earlier of an AI technical safety researcher, it’s like, I like to think that I have some talents, but it’s not like I can just turn that into AI technical safety research.

Benjamin Todd: Yes, exactly. One thing recently, just as an example where I saw this coming up is Open Philanthropy, when they did their recent recruitment round said, “In my view, perhaps the biggest overall lesson was that the pool of available and interested talent is quite strong. More than a hundred applicants had very strong resumes and seemed quite aligned with our mission.” And so then a lot of people took that to mean, “Okay, Open Philanthropy is not talent constrained anymore.”

Arden Koehler: So that “I” there was like the person running the recruiting round?

Benjamin Todd: Yeah. Sorry, this was Luke Muelhauser who wrote a blog post about the recruitment round.

Arden Koehler: Okay.

Benjamin Todd: And personally I would say it’s more accurate to say they’re not particularly constrained by finding people who have a strong resume who seemed quite aligned with their mission, but they are still constrained by someone who can just kind of like hit the ground running as a researcher. And some evidence for this is that they trialed 12 people, I think for like three to six months. But of those, they only hired five. And they’re like a multi-billion dollar foundation. So they clearly have the funds to hire more people if they found people above the bar.

Arden Koehler: So they might’ve hired many more of those 12 if they’d felt like they could really hit the ground running.

Benjamin Todd: Yeah. There’s an issue of it taking a lot of resources to train someone. Yeah. I mean, I don’t know their reasons. But, I mean, it seems like they could have easily hired an extra one person if they had found someone who they thought would be like, wouldn’t require lots of training or would it be unusually good at the role.

Arden Koehler: Yeah. So I guess in that quote, it seems like they’re saying that, in fact, talent is not the problem. They say like, “Well, impressive resumes and people who are really aligned. So it must be something else. And if we think it’s not funding, then it’s probably these particular skills or maybe just the ability to start right away.”

Benjamin Todd: Yeah. A couple of different things which might be the thing. One is management capacity. One is training capacity. One is ability to filter through all these people and find out the ones who would be really, really exceptional. And then another is just like, there’s a big difference between being pretty good and really, really good.

Benjamin Todd: The way I would characterize it is they’re not very constrained by kind of people with strong resumes who seem pretty aligned, but they are still very constrained by people who can hit the ground running immediately or unusually effective compared to existing staff members. And yeah, maybe some of those other bottlenecks as well, like management capacity.

Arden Koehler: Okay. I’ll try to stop using the term talent constraint. So now with that clarification out of the way …

Benjamin Todd: Yeah, maybe … I don’t know, it’s worth doing one other clarification, which is like many people read this and were like, “Oh, so staff at Open Phil are now replaceable.” And we’ll get on to replaceability later, but I think that we can’t conclude that from the evidence, because to know that we’d need to know, well, lots of things, but one would be just how big were the differences between the people they did hire. Like of the five people they did hire, how productive they are compared to the sixth person they could have hired but didn’t? Maybe there could still be a big difference between those. Just because there was lots of applicants doesn’t really tell us. And yeah. And Rob Wiblin has this blog post that we can link to about how even if loads of people apply to a role, you can actually still have really big differences between the top couple of applicants.

Arden Koehler: Yeah. I mean, actually, if anything, the fact that only five of the 12 were offered the role seems to suggest that they did think there were relatively big differences, assuming that they had the funding to hire more.

Benjamin Todd: Yes.

The importance of particular skills [00:13:12]

Arden Koehler: Okay. So you’ve given us some sort of evidence or some reason to think that this does not mean that Open Phil does not have skill bottlenecks. So what’s the positive case for thinking that many of the most pressing problems or the problems that we focus on the most are in fact constrained by particular skills?

Benjamin Todd: Yeah. So there’s a couple of different arguments, and to some degree, yeah, the situation within each problem is different. But yeah, I’m going to be trying to generally talk about the problems that we highlight most on our key ideas page, where it’s reducing existential risks, AI safety, biorisk, global priorities research, effective altruism movement building, and improving institutional decision making, and to some extent, other longtermist issues like we have this longer list on the page now.

Benjamin Todd: So I think one quite general argument that could apply to many people, I call the outside institutions argument. This is the idea that if another person comes along who’s really motivated to work on these issues and has a really effectiveness-focused mindset but we don’t have any funding, that person can just go and work in governments. They could try and work and get funding in academia. They could go and work at another foundation and try to work on relevant issues from that foundation. They could go and work at a nonprofit.

Benjamin Todd: So to some extent, there are already existing ways that you can get funding to work on these issues, at least to some extent. So this just kind of means like, well, even if very little funding was kind of directly aimed at these issues or in effective altruism we didn’t have much money, but if we had lots of great people, they could all still go and do useful things. So this is a kind of general reason why getting an extra person often seems kind of like that leads to more progress than getting an extra $100,000, which is enough to pay for one or two persons’ salaries.

Arden Koehler: Okay. Sorry, two clarifications on this. It seems like one of the reasons that we prioritize the issues that we do is that they seem neglected, right? So that puts at least a bit of a limit on how strong this kind of argument can be, right? Because if it were the case that you could just go into government and work on AI safety in their AI safety division, that would suggest that it was less neglected than it is. And it seems like we tend to pick problems where it’s actually kind of hard to find…

Benjamin Todd: Well, so these outside institution ones, these are places where unless you care about the issue, nothing would happen. So yeah, you can go into academia and do some random research, or you could really try to get … You could, to some extent, get a grant and work on something that you think is really important. So an extra person who really cares about the impact of their research can go and do that stuff already.

Arden Koehler: Okay. So maybe it’s like harder than it would be if it wasn’t a generally neglected issue. Maybe it will be harder to get funding because people won’t be as interested. But it’s at least possible to go into these other places and in fact, make them into a place where you can work on this kind of issue.

Benjamin Todd: Yes.

Arden Koehler: Okay. And then the other clarification, or the other question I had was, this is an argument for just thinking that getting people motivated by the sort of principles or interested in working on these issues is particularly valuable relative to the amount, a salary that you might pay somebody to work on these issues, right? It’s not like something that specifically has to do with the particular skill bottlenecks?

Benjamin Todd: Well, I think it does mean that just like if you magically could create some extra money or an extra person who’s motivated, then it’s better to have the person, which is one way of just seeing what we mean by skill constrained versus money constrained.

Arden Koehler: I see. I guess I was thinking, it seems very general. So that might be just called labor constrained or something.

Benjamin Todd: Yeah. So although I think in general people should avoid the term talent constraints.

Arden Koehler: Yeah.

Benjamin Todd: There is still kind of a sense in which you can talk about it in a very general way sometimes with some of the arguments. But it’s still good when you’re talking about… when you’re starting to plan your career to try to make it more specific.

Arden Koehler: Okay. So maybe some of the arguments are for there being these more general constraints on just like … Or like it being high priority to get more people interested in working on these issues. And then some of them might be for more particular skills that we need people to develop.

Benjamin Todd: Yes.

Arden Koehler: Okay, cool. So what’s another argument?

Benjamin Todd: Yeah. This one’s a bit, just a bit of an aside, but there’s almost a kind of converse argument to what I just said, where like, if one way we’re doing the comparison is earning to give versus working directly on the issue, with the earning to give strategy, going to earn money, well there, you’re just competing with lots of people who aren’t altruistically motivated, because everyone wants money to some degree. So that’s the kind of reason to think that it should be quite hard to have an influence via earning money and then donating it. Whereas that argument doesn’t obviously apply to some things you could do with your time, where they’re only really of interest to people who want to have an impact. And so there should be much less competition to get those roles.

Benjamin Todd: So maybe a really paradigmatic role would be some boring government bureaucrat job that no one really wants. There’s not much status in it. There’s not much money in it. But it seems like government bureaucrats do often oversee these really big budgets and have quite a big influence.

Arden Koehler: Okay. How’s that the converse to the other thing?

Benjamin Todd: I guess I was thinking like, well, if you’re already motivated, you can go and get funding from somewhere else easily. Whereas if you instead try to earn to give, then it’s not easy because you’re competing with everyone. I don’t know if it’s exactly the converse, but it seems like there’s some kind of spectrum of things there which fit together somehow.

Arden Koehler: Okay. Okay. So that’s, again, another argument that maybe doing good in general is sort of generally talent or labor skill constrained. Are there particular arguments that say like, “These top problems that we’re working on in particular at this moment in history are especially talent or labor or skill constraint?”

Benjamin Todd: Yes.

Arden Koehler: Constrained. Sorry.

Benjamin Todd: Yeah. I think a big one there is what I call the overhang argument. This is kind of the idea that effective altruism for a while is like a small number of people interested in working on these issues or kind of taking a broadly effective altruist approach. And then basically, at some point… I forget the exact dates, but like around 2016 I suppose, maybe a bit earlier, Open Philanthropy really started granting to these areas in a big way. So it was like the amounts of money being given each year suddenly started to rise a lot. And suddenly there was billions of dollars that was kind of interested in these approaches, but the number of people didn’t immediately grow tenfold as well. So then this kind of created this situation where there was a mismatch between the two.

Benjamin Todd: And that kind of really alludes to what were you just saying right at the start with AI technical safety research and how if you suddenly got way more of one resource rather than another, it can create a bottleneck for that, the other resource for a while. This is because there’s complementarities between different resources.

Arden Koehler: Okay. So this is like maybe a particular time, or at least in 2016 or whenever exactly this happened, there was this big overhang. Do you think that overhang is reduced at all with time?

Benjamin Todd: Yeah, it’s actually, it’s a pretty difficult question. So you have to try to estimate how much money is available to the community and money generally pursuing effective altruist style things is growing compared to how much the number of people interested in doing it is growing. And we don’t have great data on either of those. My guess is the number of people is growing at something like 30% a year.

Arden Koehler: So it’s people who are like, people who will identify as effective altruists or members of the effective altruism community?

Benjamin Todd: Yeah, the thing we have the best data for is just people who say they are really keen on the fact of altruism and willing to switch their careers in a kind of broadly effective altruist way, which unfortunately is not… There’s many other people who are interested in working on the issues that we highlight besides people interested in effective altruism.

Arden Koehler: Well that sounds fortunate to me.

Benjamin Todd: Yes, but it means that we don’t have data on those ones, so it’s a bit harder to tell.

Arden Koehler: So maybe over 30%. I mean, hopefully when we talk about these issues, people who don’t want to wear the label of being part of the effective altruism community might still be interested or motivated to start working on these problems, so.

Benjamin Todd: Well, yeah. And in fact, there just are lots of people like that. But yeah, and just 30%, if we say it’s now been like four years, that is threefold growth in four years. Yeah, I think the amount of money available has grown some, but because there’s now billions of dollars potentially, you’d have to find like extra billionaires to really grow it substantially or really get like a lot of people. So I guess I think–

Arden Koehler: Who are earning to give?

Benjamin Todd: Yeah. So yeah, getting 10,000 people to take the Giving What We Can Pledge, or something like that. But that hasn’t happened. So I think probably the number of skilled people has caught up some with the money. So the overhang is less than it used to be.

Arden Koehler: But do you think there is still some overhang?

Benjamin Todd: There still seems to be some overhang in the sense that Open Philanthropy and maybe some other donors would be willing to give a lot more if they could find exciting opportunities. Yeah, there’s lots of smaller opportunities that are cool and they kind of wish they could give to you, but they don’t have time to assess them all. But if someone could really come along and is like, “Okay, I’ve got an amazing plan to spend $100 million,” or let’s call it $10 million, they would probably seriously consider that.

Benjamin Todd: So in that sense, there’s still this big overhang because we could be spending much more than we are. But though yeah, you can kind of see that someone able to actually do that would have to have the skills needed to be an entrepreneur who could run a big organization and do something that all the funders think is effective, which is quite a high, high bar.

Arden Koehler: Yeah. You mentioned they maybe don’t have time to assess all of these opportunities and we’re talking about Open Phil here, but it seems like there are other grant makers in the community and it seems like this is just sort of funny combo of like, well, it is a shortage of talent-labor-skill to be able to have people assessing these opportunities, but the effect is it makes money flow less freely. So it’s kind of interesting. It sort of like mimics a shortage of funds, but it’s really a shortage of skill or labor.

Benjamin Todd: Yeah, it’s a certain type of skill. One way of seeing it is because in the overhang argument the fundamental thing driving it is complementarities between money and various types of labor. That as the overhang gets bigger, the types of labor that become the most valuable are the types of labor that are the most complimentary with funding. So, for instance, someone who’s able to be a grant maker, that skillset becomes really valuable the more money there is around. Someone who is able to be an entrepreneur and really kind of deploy a lot of resources or a manager becomes more valuable. To some extent, research talent becomes more useful because research is… Like their discoveries are applied over a larger amounts of resources.

Arden Koehler: Wait, so why does a manager become more valuable? That feels unintuitive to me.

Benjamin Todd: Well, yeah, I’m not totally sure, but the idea is something like, if you really want to scale up quickly and suddenly hire loads of people, then, in practice you won’t be able to do it because you don’t have enough managers. You’ve got the funding, but you can’t turn that funding magically into people doing the right thing. You need someone to hire them, train them, manage them.

Arden Koehler: Okay, that makes sense.

Arden Koehler: You also said this thing about, it’s quite a high bar to be able to impress all of these fund managers or grant makers. And that also seems interesting because you could think that a natural response to the overhang would be to lower the bar for funding projects. Does it seem like that’s happened? And if not, why not?

Benjamin Todd: I think it has happened a bit in that it seems easier to raise a bunch of money for an effective altruism style thing than before 2015. Another thing is salaries have been raised a bit, which I think is also a sensible response because if you’re saying, “Well, we’ve got loads of money,” and no one’s doing anything, a pretty natural thing is just to pay people more. I think people often think raising salaries will help more than it does, but it does help a bit.

Arden Koehler: And that’s because what we were talking about earlier. You can’t just get people who have the skills that you need to suddenly appear by offering a high salary.

Benjamin Todd: Yeah. I mean, I think there’s actually a couple of other factors. One is that people, yeah, people are motivated by many different things, only one of which is money. And so you can increase salaries a lot, but only get somewhat better people, not hugely better people. And in particular, the people who care most about doing good, tend to care the least about money. It’s not a perfect match, but there seems to be some relationship there. So it kind of helps even less than you might first hoped compared to. If we were in a much more money-driven domain like in finance, you’d expect that more money will mean more people pretty directly, because that’s kind of why people are doing it. But it’s not really the case within research or social impact, non-profits, or things like that.

Benjamin Todd: Another effect is because of culture, you can’t hire a new person and pay them three times more than all your existing staff, because people will feel like that’s unfair. So actually, if you want to raise salaries, you tend to have to raise salaries across all of your staff. So that actually means the marginal cost of hiring those higher salary people is way more than their salary. It’s like their salary plus the increase across everyone else.

Arden Koehler: Yeah. Interesting.

Benjamin Todd: Also in some areas you can’t raise salaries, like academic institutes are typically on a fixed university salary scale.

Arden Koehler: Right. It’s not like an effective altruism nonprofit is in charge of the Oxford salaries for the various people doing global priorities research.

Benjamin Todd: Yeah. Lots of stuff happens in organizations that can’t. I mean, there’s other things as well, like PR issues where if you were paying loads in a nonprofit, even there, it will be seen badly.

Arden Koehler: Yeah.

The value of the most productive people [00:26:45]

Arden Koehler: Okay. So are there any other arguments for thinking that the problems we prioritize most highly are constrained in this way?

Benjamin Todd: Yeah. So I think there’s a couple more. And like I should say, none of these are kind of decisive by themselves, but they kind of add up. So this one seems less clear to me, but it seems like often the most skilled people in a career path or in an area have a lot more impact than the median. Let’s just take that as a given. Obviously that’s a big topic about how much people do actually differ in their productivity or output or impact in different fields.

Benjamin Todd: But let’s just assume there are large differences. That can create talent constraints, if there’s some constraints on salaries, which I just kind of said, there are a bunch of constraints on salaries, like around culture and university, people being on fixed scales. And so when you do have some constraints like that, it kind of means that the most productive people get paid much less than they’re actually producing. And so then anyone extra like that is super valuable to you because they have this big, extra impact that you’re not paying for. And that kind of thing seems to create these big differences.

Arden Koehler: Because the incentives aren’t quite lining up. You’re not really able to reward the thing that is in fact the most valuable to you.

Benjamin Todd: Yes. And it’s most obvious in something like academic research where often many researchers will be paid the exact same amounts, but it still seems like even those who are paid the same, one might be producing several times more papers than another.

Arden Koehler: That’s what book sales are for.

Benjamin Todd: And we interviewed a bunch of biomedical researchers about how much they felt talent versus funding constrained, and lots of them said, “Yeah, well I think the best researchers often produce way more,” and they didn’t say this, but effectively they have to pay them the same. So an extra person like that really contributes a lot more to their lab. So this is an argument why anyone who’s kind of unusually good at an area, we’re constrained by more people like that.

Arden Koehler: Yeah. So I guess this is, in some sense, another argument that shows that money can’t quite be turned into this kind of specialist labor as easily as a first-year economics class might make you think or something.

Benjamin Todd: Yeah. The other general type of argument is just simply what people in these different issues think. And so we’ve tried to do these surveys of people who work in these different problems that we work on. And sometimes we just ask them, “Do you feel more talent constrained than funding constrained?” And typically people say they feel more talent constrained. Yeah.

Benjamin Todd: And another question we asked was like, we tried to get different organizations to estimate the value of retaining their most recent hire in dollars. And then typically people gave very large figures for those, hundreds of thousands of dollars per year, which was more than most of those people could have donated if they’d earned to give instead, which is suggestive of there being talent constraints. Though again, you have to be very clear about what that data actually shows. It shows that the people who are able to get jobs in those areas were worth a lot in terms of money, but that might be a very unusual type of person.

Benjamin Todd: Also, it’s saying that existing employees are very valuable, but they’ve already been through the hiring and training process. We did ask about their most recent hires. So it’s not that they’ve been trained for years and years, but the organizations already paid a lot of costs to get them to that point.

Arden Koehler: So those will probably be worked in, or it’s a lot of work to hire, so they will not want to lose this person and have to replace them, even independent of everything about whether it’s…

Benjamin Todd: So they should say that those people are more valuable to the organization than a random new person who they haven’t assessed and trained yet.

Arden Koehler: But it seems like the figures that people gave were even higher than you would expect, even working that stuff in.

Benjamin Todd: Yeah, I think so. I mean, it’s hard to know. Hiring and training people does cost a lot of senior management time, so that could also… That’s effectively very expensive as well. So it could have a significant effect on.

Arden Koehler: Yeah. Although, I mean, if that’s what’s driving it, then in some sense that’s another form of this sort of talent or labor constraint. It’s like, “We don’t have enough senior management time in order to make this happen. If only we had some more people at that level, then maybe it’d be easier for us to do this hiring.”

Benjamin Todd: Yes. But then yeah, that’s like a very particular type of person, someone able to be a manager in one of the important non-profits in one of these areas.

Arden Koehler: Yeah. And I mean actually maybe able to it was a little misleading because in this particular case, it’s like, not only are they able to, they have the experience, they actually usually are managers. So some of this just maybe will take time.

How the situation has evolved over time [00:31:16]

Benjamin Todd: Yes. Yeah. Maybe this is now a good point to kind of step back a bit and I think it’s quite useful to think about how the situation’s evolved over time as another way of seeing the situation. So you can kind of think really early in the effective altruism, say like in 2012, there was a bunch of interested people, but none of us really had much money. So it really did seem like an extra $100,000 would have… I mean, when we first started working at 80,000 Hours, our salary was like £15,000 a year. So an extra $100,000 could have paid four of our salaries. So that really seemed like, “Wow, we’re like very funding constrained.”

Arden Koehler: And in particular you might’ve hired an extra four people if you’d had that extra 100,000.

Benjamin Todd: Yeah. It’s hard to know in our particular case exactly what would have happened, but, yeah.

Arden Koehler: Something along those lines.

Benjamin Todd: Yeah. It seemed like money was much harder to come by, so therefore much more useful. But then yeah, as I was saying, we kind of went into the second stage where then there was like Open Philanthropy and there’s other billionaires who were somewhat interested. And so like Reid Hoffman’s donated to Global Priorities Institute and a bunch of wealthy people got interested in effective altruism. Some people went to earn to give and they did well, and then they started funding organizations in the community, organizations working on these problems.

Benjamin Todd: And so then although money is still a bottleneck and is always useful, it became a bit less so. And then instead there was this kind of phase when yeah, there was lots of money and not as many people around to do things. And so then there was a sense in which, yeah, we were pretty constrained by maybe just this kind of general junior talented people stage. But then it seems like in the last couple of years, there has been a shift where, if you’re just fresh out of university and you don’t have any particular skills or training in these areas, it’s maybe even a little bit harder to get a job than it would have been, say, in 2016.

Benjamin Todd: And so now in this third stage, we’re a bit less constrained by kind of generally interested, talented people, and a bit more constrained by either people who have very particular skills that are needed, such as we used the AI technical safety example earlier, or grantmaker skill sets, the kinds of things we list on our priority problems. Or maybe we’re more constrained now by what you might want to call an organizational bottleneck, which is ability to figure out who’s interested. So there’s a kind of searching/vetting bottleneck, and figure out who would be able to contribute and then train them, manage them. And even just have things that lots of people could do.

Benjamin Todd: An example that sometimes gets brought up is Teach For America. So Teach For America, I don’t know how many thousands of people do it every year, but it’s like some significant fraction of talented college grads. The equivalent in the UK, I think, it’s called Teach First. It’s the biggest graduate employer in the UK. And I think it’s something like 5% of graduates do it, or at least, I don’t know, at Oxford, it was something like 5%.

Arden Koehler: So they’re very good at putting people to work.

Benjamin Todd: Yeah. But there’s no kind of effective altruism equivalent of Teach For America, something that can just absorb a thousand people and train them and get them doing things that seem really useful for the top problems.

Arden Koehler: So it sounds almost like this is a kind of movement building, but I guess oftentimes movement building involves just introducing people to the ideas and things like that. But maybe this is something more particular, like taking people interested in the ideas to really being able to contribute. And maybe that involves actually having a bunch of roles available for people, but maybe it also involves helping give people a bunch of skills.

Benjamin Todd: Yeah. There’s almost another type of overhang as well, which is just number of people who’ve generally heard of effective altruism and are generally interested in these issues is much larger than in the past. Whereas in the past, just kind of having heard of effective altruism already made you this unusual person. Now it seems like the bottleneck is more going from people who are generally interested in, to like trained an actually in a really effective role. And it’s like that later stage that seems like it’s been the bottleneck in the last couple of years.

Arden Koehler: Yeah. Interesting. So it’s like, there’s funding available and there’s people available, but somehow, we, in the effective altruism community, and in particular in these professional organizations, we haven’t yet figured out how to get those things to work together to actually get progress made, or at least on a large scale.

Benjamin Todd: Yeah, I guess within the main organizations. And I wouldn’t want to overstate it because in some ways, things have got better. Like even the fact that Open Philanthropy did this hiring around and hired like junior people they were going to train up, they weren’t doing that a few years ago. Or there’s things like the Research Scholars Program which is aiming to do this.

Arden Koehler: This is at FHI and it takes people who are prospective graduate students and helps them figure out what they want to do.

Benjamin Todd: Yeah. In particular, like to do with research in these problem areas.

Arden Koehler: So yeah, I guess maybe one complication here is that it feels most easy to imagine this organizational capacity bottleneck or something, in the case of like, “Well, organizations that have the effective altruism label, aren’t big enough and don’t have enough managers to basically be able to hire these people.” But then I guess since we think so many people can make such a big, positive difference working in areas besides effective altruism organizations, in government, in research, what’s the equivalent of this capacity bottleneck for those cases?

Benjamin Todd: Well, I was almost wondering if I should emphasize now that what I’ve been talking about is always just a matter of degree about which bottleneck seems like the very most pressing right now, but always additional organizational capacity, talented people, funding, they’re always useful and there’s always good things to do with those things. So I’m not saying that all those other things are just not useful at all. And you’re giving some really good examples of, “Well, if you are just a generally talented person, maybe it’s a bit harder to get some of these jobs, particularly at the nonprofits that are most central to the community right now than it was, say, in 2015.”

Benjamin Todd: But that doesn’t mean there’s nothing useful to do. You can go and train up in academia or start focusing on some kind of research there. There’s many, many hundreds or even thousands of people could go and work in government and policy positions. Yeah, you could go and work at some other nonprofits that are relevant to these issues, but not labeled as effective altruist organizations. And so yeah, having extra talented people is still really useful. It’s just, exactly what you might focus on, would be a bit different.

Arden Koehler: Yeah. I was thinking maybe one answer to the question of, what’s the analog of organizational capacity for those other areas, it might be guidance or something. Of course, this is something that 80,000 Hours is trying to provide, but figuring out what are the best roles in those other institutions and people having support or community when they’re in those other institutions so that they feel good about it and feel motivated for the long haul. Those could be sort of the equivalent. And if we got those, then it’d be easier for people to put their skills to work.

Benjamin Todd: Yes. And I think one thing that does make it hard to go and do those other things is it often requires more independence, because you might be going out alone or it might feel like that. And so yeah, in a sense that’s like another type of organizational bottleneck, is like, could someone form a really good community of people that are all trying to work in a certain area of policy together, and that would help them all do that more easily.

Arden Koehler: Yeah. I mean, I think working at an effective altruist organization, you and I are super lucky because we get to talk to people about the things that we care about all day, and talk to people who share our values. I think it’s really hard for people who are like, “I really care about these things, but I’m going to go out into the wild and work in a department of a government where nobody else will care about the same things I care about.” But I guess, if there was some way to make that less true and make those communities more supportive, then that would make it a bit more attractive and easier for people.

Benjamin Todd: Yes. And I mean, this is starting to change a bit. There are lots of other people interested in these ideas, doing those things, who will be up for chatting to you.

Arden Koehler: Cool. Okay. So I guess this is the… You said you felt like–

Benjamin Todd: Maybe now we’re in the third stage of effective altruism.

Arden Koehler: Yeah. So first it was no money. Then it was like…

Benjamin Todd: Generally junior people.

Arden Koehler: Need people to do stuff in general. Now it’s this more organizational constraints, or people with specific skills. What’s coming next, do you think?

Benjamin Todd: I’m not sure. So one option could be that we just go back to one of the earlier stages. So you could imagine if the number of people keeps growing, but the amount of funding doesn’t grow as fast, then eventually we could go back towards being funding constrained, or the other thing that could happen, if we managed to scale up these areas a lot so that it’s easier to absorb lots of people who don’t already have those specific skills, then it could go back to being constrained by generally talented people.

Arden Koehler: I guess that would be kind of a good thing. I could imagine sort of this cycle. But of course, remembering, like you said, that this is always just, what’s the most valuable thing on the margin. It could be going through the cycle, but getting bigger and bigger and bigger until it’s like, “Well, we’re funding constrained because we don’t have an absolute ton of money to pay all of the people that are so talented and have the skills and that we have the capacity to employ.” And that would be good.

Benjamin Todd: Yeah, and we should talk later in the episode about how to kind of respond to this as an individual. Another thing that could become the bottleneck is, as more and more people get interested in taking these kinds of approaches, it becomes harder and harder to coordinate them. And so maybe we’ll end up thinking coordination is the main bottleneck. And one example of that is, as more and more people are involved, it becomes harder to share information between them. And so maybe having a really good infrastructure to share information becomes the key bottleneck. And, I mean, to some extent, this has already happened, which is one reason why we wanted to set up the job board. Or that’s almost actually more the organizational bottleneck’s thing because when there’s only 10 jobs, you don’t need a job board, but now there’s hundreds of jobs and it becomes really valuable for someone to save everyone else time by collating them all into one.

Arden Koehler: To be clear, I think many of these jobs would have existed before. We just didn’t know that they would be good things for people to do. Because a lot of things on the job board are not at effective altruist organizations. They’re in government.

Benjamin Todd: That’s true, though there are also just way more jobs in general.

Arden Koehler: Yeah. But I guess the coordination thing sort of shades into the organizational capacity thing, because it’s like we don’t have this way of coordinating people, or something like that.

Benjamin Todd: Yeah. There could be other types of examples, like different groups starting to compete over the same resources, or one group doing a thing that they think is really good, but someone else thinks is a disaster and they could have, in theory, compromised and done something that both of them think is good, but they’re not able to do that because we don’t have the infrastructure.

Saving resources for later [00:42:01]

Arden Koehler: So you’ve given some arguments that seem relatively convincing for why the particular issues that we focus on right now at 80,000 Hours, so existential risks from artificial intelligence, biorisks, extreme climate change, institutional decision-making, building effective altruism… Am I forgetting any?

Benjamin Todd: Global priorities research.

Arden Koehler: Global priorities research, yes. And nuclear security.

Benjamin Todd: And a bunch of the other, what we call like our other promising problems.

Arden Koehler: Although I feel like we’re less sure that those are in particular constrained by the same resources because we just haven’t looked into them very much.

Benjamin Todd: I feel like they’re mostly constrained by someone who’s able to either research or pioneer the area, and someone like that who comes along would get funding, but there’s not many people able to do that.

Arden Koehler: Okay.

Benjamin Todd: Yeah. Or they could go and get a job in academia researching the topic or something like that.

Arden Koehler: So even if all those issues are constrained in the way that you say, it seems like one could argue that actually the ways of doing the most good are still funding constrained, because maybe they think that, in fact, the thing that does the most good, is just saving resources for later. And we can save money maybe more easily. I don’t know. I mean, you could argue with this. You can invest money more easily perhaps than investing in movement building. Maybe it’s more efficient. In fact, we could really put all of those funds to amazing use right now if we just invested them. What do you think about that?

Benjamin Todd: So if we put a lot of weight on patient longtermism…

Arden Koehler: Which we talked about in a previous episode of this little series.

Benjamin Todd: Though we’re not supposed to refer back… Well, we have a blog post about it, explaining it recently.

Arden Koehler: It’s basically the idea that the best opportunities for doing good might be far in the future.

Benjamin Todd: Yes. Good definition. So if we put a lot of weights on that, well then I think the most pressing priorities are, again, movement building or global priorities research, which I think are both more constrained by various skill bottlenecks than funding.

Arden Koehler: Really? So you think they’d be better to spend funds on global priorities research than on just investing them to grow for the next 300 years or so?

Benjamin Todd: Well, the reason why we’re doing the movement building at all is because we think it’s higher returns than just investing in the stock market.

Arden Koehler: Okay, that’s helpful.

Benjamin Todd: Once those opportunities are used up, then it may well be then the best thing on the margin is to save money and then it would come down to… Well, an additional person can earn a bunch of money and save it. So yeah, I guess I’m not sure exactly whether you’d say that was like… I guess it would just be a question of whether having more people leads to more money than having more money.

Arden Koehler: Well, I mean maybe the reason it gets confusing in this case is that if the best opportunities for impact are really far in the future, then resources are more fungible. We were talking at the beginning, it takes time to train people up. But if you have that time, then you can really turn money into talent or skills.

Benjamin Todd: Yes. I think that’s true. And I think also just the option of saving does complicate some of the arguments that I just mentioned. So the funding overhang arguments, well, you could instead just save the money and wait for the bottleneck to go away, and you don’t want to do that if you’re a more urgent longtermist because that waiting, that delay, has a big negative. But if you’re quite on the patient end of things and you’re happy to wait, so–

Arden Koehler: Or, of course, if you aren’t a longtermist at all. It seems like if you’re not a longtermist, then probably you’re going to be less inclined to put that money away.

Benjamin Todd: Well, yeah. Then you just get into the kind of what we call the now versus later trade-off in whatever that other problem is that you’re focused on. And then you have to figure it out there, and it’s not obvious to me that, say, people really focus on global poverty wouldn’t be better served by investing the money and donating it in a couple of decades than giving it now.

Arden Koehler: Okay. Yeah. So I guess that just depends on specifics of how quickly are people getting better off, what’s the rate of return on investment, and maybe we don’t want to get into that here, but you’re pointing out that it’s an option. It might be the right thing to do, even if that’s your focus.

Benjamin Todd: Yeah. But I think it is true that the overhanging argument seems much less pressing if you’re happy to wait.

Arden Koehler: Another argument for thinking that the project of doing as much good as you can, is still funding constrained, is if you thought that we were wrong about what the most pressing problems were, and in fact that we don’t know what the most pressing problem is. So maybe, then you would think, “Well, even though you can’t turn money into labor or skilled labor super quickly, it’s still easier to try to pay some new person to come in with a brand new skill set to work on some new problem that you only heard of last year, than it is to convert somebody who’s an expert in problem A into an expert in problem B.” I mean, I don’t know if that’s right, but it seems like something somebody could think.

Benjamin Todd: Well, I mean, I think if we don’t know what the most pressing problem is, then again, we should focus on global priorities research, which takes us back to that.

Arden Koehler: Okay.

Benjamin Todd: Or we should invest in broad interventions. So things that make society generally better at tackling whatever global challenges turn out to be the most pressing ones. So maybe reducing great power conflicts is an example of something like that. Or improving institutional decision-making is another one we talk about. And then it comes down to which ones those are the most constrained by.

Arden Koehler: Okay.

Benjamin Todd: But I thought you were just going to make the general point that the degree of funding versus scale constraints in different problems is different. And so if you think global health is the most pressing issue, I think there’s a much stronger case for it being funding constrained, than the ones we talk about the most, which is basically that, if you look at GiveWell’s top charities, I haven’t checked the figures recently, but I think last giving season, it seemed like there was on track to be about $100 million funding gap, which they didn’t expect it was going to be filled by even Open Philanthropy or their donors. And so having an extra hundred million dollars actually per year would be super useful to just fill that gap.

Arden Koehler: So that means that these organizations that GiveWell is recommending, they do, in fact have the organizational capacity and the people to use these funds.

Benjamin Todd: Yes.

Arden Koehler: Unlike we were talking about is maybe the case in the effective altruism community that’s focused on these longtermist issues.

Benjamin Todd: Yeah. Or just the issues that we most focus on. And yeah, what’s kind of going on there is that it turns out there’s these really scalable global health interventions, like giving out insecticide-treated malaria bed nets, where, to some degree, you can just turn $5 into a bed net and you can just do more and more of that.

Arden Koehler: At least to a certain point. I mean, presumably they would run out of infrastructure eventually if we just dumped money on them.

Benjamin Todd: Yeah, yeah. But roughly the top charities have this hundred million dollar gap, so we’re a long way from that. And then after that, you could still do GiveDirectly and that could potentially absorb much more money again. I should say, I think an extra talented person could go and work on something even better than earning to give to those charities. They could go and try and work in an international aid department in government where they could go and work at a big foundation that focuses on these issues. They could go and work in a think tank that’s working on development policy. And I think many of these could potentially be much higher leverage than earning to give and spending it on malaria nets.

Benjamin Todd: But it does at least show that there is this big use of money that seems like at least very good on the scale of things, even if it’s maybe not the very best thing we could do about global development.

Arden Koehler: So I guess some people might think that the best thing that we can do for the long-term future is just to broadly improve positive values through certain kinds of advocacy. And that seems like maybe a case in which it might be funding constrained. It seems like–

Benjamin Todd: I don’t think so.

Arden Koehler: Really? Okay, yeah.

Benjamin Todd: Social advocacy is one of the hardest things to just pay for and get.

Arden Koehler: I thought it might be because it’s like a numbers game or something, like it’s all about just getting enough people to be talking about this. And so that feels more like you can pay for it than a research breakthrough or something.

Benjamin Todd: No. I think if we think about big social shifts in the past, like civil rights or gay marriage, although the advocates doing that do need some funding, in practice they didn’t have that much funding and it’s much more a case of someone being super inspiring and figuring out a great message and then kind of gradually spreading it across society.

Benjamin Todd: So I would say that’s very constrained by people who are really good at advocacy.

Arden Koehler: Yeah, okay.

Benjamin Todd: I mean, maybe you could imagine, could we design a Facebook ad that could just get people to become altruistic and put hundreds of millions of dollars into it. But I think kind of basically, no.

Arden Koehler: Or maybe it would be more constrained by a brilliant creative mind to create such a–

Benjamin Todd: Well, yeah, coming up with something like that would be a very big achievement.

Arden Koehler: Yeah. Interesting. Okay. Cool. All right. Is there anything else we want to say on this sort of issue of–

Benjamin Todd: Well, you did just give me one idea where…

Arden Koehler: Sure.

Benjamin Todd: I mean the kind of thing that might be a longtermist thing that’s closest to buying lots of malaria bed nets could just be buying loads of carbon permits and hopefully something a bit more leverage than that, but just… There are various climate change things we can do that could absorb billions of dollars, and that reduces climate tail risks and climate change is probably a risk factor.

Arden Koehler: Yeah. That’s really interesting because climate change is one of the better-funded areas that we are interested in. And so it’s interesting that you feel like that’s one place where more money could do a lot more good.

Benjamin Todd: Well, I’m not saying that money would do more good there.

Arden Koehler: Right, sorry.

Benjamin Todd: I actually think money does more good in the other things. I’m just saying that it could absorb a bunch of money and it would do a somewhat useful thing, though not like the most useful thing.

Arden Koehler: Yeah. Sorry. I was trying to say it seems more funding constrained than some of the other areas, which is not the same thing as saying that money could do more good there.

Benjamin Todd: Well, yeah, it’s more funding relative to talent/skill constrained. I mean, I’m not even sure I would stand behind that.

Arden Koehler: But it is plausible.

Benjamin Todd: I’m just saying it could absorb lots of funding and it would help.

Arden Koehler: Cool. Okay. So hopefully this hasn’t been too confusing. I feel like it got a little bit messy, when the entire point was to clarify things, but is there any way we can sort of sum up the main things that you’ve learned thinking about this?

Benjamin Todd: Oh yeah, some of the things we’ve covered in the conversation. The one thing would just be, as you can see, it’s fairly complicated and sometimes confusing. But as a quick summary, we talked about how each global problem can be constrained by different resources. And so, as people who wants to do good, we want to figure out for the ones we’re focused on, which things would be the most helpful. And we gave a bunch of arguments for why the ones that we’re focused on seem to be… Right now we kind of call it the third stage effective altruism, where they’re more constrained by certain skillsets and organizational constraints than in the past. Though, as I’ve said, all of these resources can be used to have more impact, even within our niche longtermist problems. Additional funding, you could give it to one of the EA funds, the EA Long-Term Future Fund or the EA Community Fund. And they seem to often fund good stuff, including us.

Arden Koehler: Hopefully good.

Benjamin Todd: So, yeah, we’re just talking about what things have been most pressing at different stages.

Replaceability [00:53:27]

Arden Koehler: Okay, so I’m wondering how this is related to the concept of replaceability. So, actually replaceability was one of the first concepts that I associated with effective altruism. I heard people talking about, “Well if you go work for a nonprofit versus taking some high paying job and donating your salary, in the case where you’re working for the nonprofit, if you didn’t work for them, somebody else would do similar work. Whereas in the case of earning to give in a high paying role, the next person who would have done that work would have just kept the money and not given it away. So you were a lot less replaceable in the earning to give role.” So that was one point in its favor. And I remember being like, “Wow, that’s interesting.” So how’s your thinking on that analysis evolved over time and how is it related to this concept of funding versus different kinds of constraints?

Benjamin Todd: Yeah, so I still think all else equal that argument you made is an additional reason in favor of earning to give compared to another nonprofit job. But an extreme version of the argument where it’s like, “Well, you wouldn’t really have any impact working in the nonprofit because someone would do it otherwise” isn’t true. And that can often mean that it’s higher impact to work in a nonprofit rather than to earn to give. And one way to see that is the whole discussion we’ve just had about funding versus skill constraints. Those arguments that I was making all have taken account of replaceability already, or at least that’s the hope. So I’ve basically been trying to show that often you can do more by working directly on one of these issues than earning to give or a similar amount of money would do. And yeah, if those arguments are correct, then that’s showing that, in this case, that argument that you made isn’t holding, at least for people who are able to take these jobs.

Arden Koehler: Okay, so walk us through this a little bit more. Why wouldn’t somebody just be replaceable in this extreme sense in a job like working at a nonprofit?

Benjamin Todd: So yeah, later we should outline just how maybe we might analyze replaceability in general. But just to that specific point, one thing that can happen is just you might just be better at the job than the person who would have done it otherwise. And it just seems like empirically the case to me that often in hiring rounds, like for organizations in these areas, even among the top couple of candidates, there do seem to sometimes be quite large differences in how much impact they would be expected to have at the organization. And so if we think the top candidate might have like … It wouldn’t be crazy that they would have twice as much impact as the marginal candidate. In which case, at worst case replaceability would only reduce their impact by 50%.

Arden Koehler: So this is like, if you get the job that shows, maybe, or that suggests that, in fact, you would be better at it than the next candidate, because the people in the organization have weighed up the evidence and decided you seem like the best person for it. In which case, your impact is everything on top of what the next person would have done, which could be substantial.

Benjamin Todd: Yeah. I think the bigger point is it seems like there are often pretty big differences, even when there’s been lots of applicants. This is actually exactly what we would expect if we think people skills is again this heavy tailed or log normal distribution, you’d expect the differences among the very top people are actually larger than the differences among the median people. And so a job application round can be really competitive in the sense that hundreds of people applied, but that doesn’t necessarily mean that the top two candidates were the same.

Arden Koehler: Yeah. So I think we mentioned Rob’s post earlier, he goes into the math for why, in fact, actually, if there’s more applicants that could mean you’re less replaceable, which is maybe a bit counterintuitive. Listeners can check that out, we’ll link to it.

Benjamin Todd: Okay and then the second point is that suppose you get the job and someone else would have taken it otherwise. Well, if that person is also concerned to work on these issues, they’re going to go and do some other thing that helps. So you’ve actually freed up that person to go and do something else. And so that impact should also be accounted for.

Arden Koehler: So by taking the job, I make it so that they don’t have to do the work and they can do something else that’s useful.

Benjamin Todd: Yes. So that doesn’t apply if you’re taking a job that someone who doesn’t care about impact could have done otherwise, but it does count if you are considering working in a job that lots of altruistic people who care about these issues do work on.

Arden Koehler: Yeah. So I wonder if it’s like if you’re thinking about replaceability, maybe the main thing to avoid is doing a job that somebody who doesn’t care at all about impact could do just as well as you, in terms of having a positive impact with their work.

Benjamin Todd: Yes, I think that is one other implication, probably. Just also one other interesting point is the two things I just said, actually, they balance each other out. So if I am definitely replaceable and someone would have definitely done the job otherwise, then now there’s 100% of an extra person doing something else that’s good. Other cases I’m not replaceable and no one would have taken the job otherwise, in which case I do 100% of the impact. And so, in either case, I’m still doing some good. Do you see what I mean?

Arden Koehler: Interesting. So if there are other people around that are just as skilled as you are in the same way, then on the one hand… If there are those other people around, then it’s better to free them up, and if there aren’t those other people around then it’s like–

Benjamin Todd: Then you’re not replaceable.

Arden Koehler: You’re not replaceable at all. Yeah, interesting. Okay, so I was going to like–

Benjamin Todd: And yeah, that’s an intuitive idea in a way, because all else equal, we should expect there’s like more talented people wanting to work on pressing problems, and so more’s going to get done.

Arden Koehler: That does sound common sense.

Benjamin Todd: Yeah, because otherwise it’s just like no one would be able to do anything because it would always be being replaced, but it’s clearly not how the world works.

Arden Koehler: Progress in the world is sometimes made. Yeah, so one way of pushing back on the second point is if the issue that we were pointing to before, you were pointing to before, about organizational capacity is really bad, then it might not be the case that you really are freeing up somebody to go do great work. So imagine there’s only a hundred opportunities for doing great work in the world, this is getting very fanciful.

Benjamin Todd: Yeah, or within our top problems within AI safety or something.

Arden Koehler: Within problems we think are the most pressing, then by taking the job, the other person isn’t really freed up. If I’m the hundredth person to take it or to take one of these jobs, then yeah, they can’t do anything. And so it seems like maybe the replaceability argument is a little bit stronger while we are having this bottleneck around organizational capacity.

Benjamin Todd: Yeah. So, to some extent, you just need to get into the details of the situation. And yeah, I agree there could be a case where the person you free up just can’t do something valuable because there aren’t any other good roles left. In which case you would again become a bit more… You’d be more replaceable. Though there is one lower bound to bear in mind, which is like someone could always earn to give and that’s actually a marginal thing that lots of people could do.

Arden Koehler: Where that could mean even taking a moderately paid job and donating 10%.

Benjamin Todd: Yes. Though I suppose if we think it’s that then… I mean, if that thing is much less impactful than actually getting the job, then it would again make you more replaceable.

Arden Koehler: I guess I was just trying to make the claim feel more plausible that this is something that anyone could do, because it’s not the case that anyone could be a quantitative trader that makes a ton of money a year.

Benjamin Todd: No, so yeah, unfortunately earning to give just gets associated with really, really high paying roles. But the way I prefer to think about it is you’re earning to give whenever you just earn more than you would have and donate extra because of that.

Arden Koehler: What if you just earned the same amount that you would have, but donate more than you would have?

Benjamin Todd: I guess I don’t quite think of that as earning to give, but it’s still a good thing today.

Arden Koehler: Okay. Possibly nitpicking. Okay, so it seems like the replaceability argument could be a bit stronger, but it’s still not going to be that strong, even if we’re in this specific world.

Benjamin Todd: Well, yeah. I mean I think a general message with replaceability is just it’s really complicated and you have to think about the details of the specific case. You can’t just make a general argument like, “Well, someone else would have done the job otherwise so therefore you have no impact.” And yeah, depending on the details, it could go one way or the other, you could be more or less flexible.

Arden Koehler: Does it seem like there’s a general thing though where it’s still the case that earning to give is always going to be the least replaceable until everyone starts donating a bunch of their income?

Benjamin Todd: Yeah. No, I mean, like I said right at the start, I do think, all else equal, the argument you made right at the start is a reason in favor of earning to give. But yeah, I think another big point I would want to make, because I think it’s a relatively small correction, because maybe we’re talking about it reduces your impact by 50% or something, but we tend to be dealing with so much bigger uncertainties than 50% that, that really just isn’t a big issue. We should instead be focusing more on just how high impact is the organization? How good is your personal fit? Will you get good career capital from it?

Arden Koehler: And you think that all of those are more… I mean, sorry, I’m just a little confused. So you’re thinking all of those factors could have a bigger effect on your impact than whether you’re 50% replaceable?

Benjamin Todd: Yeah, I think either they’re bigger or it’s more obvious which… More career capital is better, that’s pretty obvious, whereas with replaceability, it’s really complicated to work out what effect it’s having in a particular case. So once you take account of both the uncertainty and the smaller size of the effect, it often just doesn’t seem like the most useful thing to be thinking about when choosing a career, which is why when we are actually taking people through how to compare options, we don’t really foreground replaceability as a key thing to think about. Though the way it gets factored in is we try to recommend things that people do that are more skilled constrained than funding constrained. And in a sense, replaceability is being factored into that already through that.

Benjamin Todd: And I would also say it is an extra argument in favor of earning to give. Though, yeah, it could also be an argument in favor of other… There are other jobs where you can do a cool thing that just wouldn’t be done by a non-altruistic person. So an example would be, yeah, well you could be an academic and you could focus your research more on high impact things. You could be a journalist and you could write about more high impact things than would have been written about. So obviously you won’t have perfect control over what you have to write, about you’ll have to do lots of boring stuff as well, but you’d still have some flexibility. And yeah, I think there’s like lots of roles like that where people who didn’t care about impacts wouldn’t have done it otherwise.

Arden Koehler: Yeah, okay. So like earning to give, this is a case where you’re converting resources that people would have spent on just other stuff into being spent on these top problems, like readership of a newspaper, if you’re a journalist writing about these problems or something related.

Benjamin Todd: Yes.

Arden Koehler: Yeah, cool. That’s interesting.

Benjamin Todd: We could go into a little bit more. Like if we did want to analyze replaceability for a particular job, how might we do it? And yeah, this is something that I’ve had all these very rough notes on for many years, but have never been able to quite get them to the point where they get released.

Arden Koehler: This is going to be a “Don’t try this at home, listeners. Don’t actually do it, but we’re going to do it here on the air”.

Benjamin Todd: A little bit. I mean, I think it helps to illustrate how complicated replaceability is, which maybe will discourage people from relying on it too heavily as an argument. So before we start, when we say how replaceable is someone, the way I think about it is you have the impacts that you tangibly seem to do in the role, and then you have your actual counterfactual impact having taken account of replaceability. And I like to think of it in terms of the actual impact divided by the tangible impact is how replaceable you are. That’s the replaceability factor.

Benjamin Todd: So if you’re a doctor and you do a hundred operations and you save a hundred lives, it seems like your tangible impact is, “I saved a hundred lives.” But actually, once we take into account replaceability, it’s going to be less than that. And so that’s your real counterfactual impact. Suppose that’s 10 lives, then in that case the replaceability factor is 10%. 10 divided by a hundred. And so then yeah, then we can say like, “Well, when you’re not replaceable at all, you’re having 100% of the impact that it seems like you have. And when you’re totally replaceable, you’re having zero impact, so then the replaceability factor is zero.” So then the question is how do we estimate this factor?

Arden Koehler: I think the reason that was confusing to me for a second was that you’re like, “When you’re totally replaceable, then the replaceable factor is zero.” The more replaceable you are, the lower the replaceability factor is.

Benjamin Todd: That is confusing, but the way it’s defined like that is because then you can multiply the replaceability factor by your tangible impact and get your true impact. Which is what we ultimately care about.

Arden Koehler: Okay. So remembering that it’s a bit backward or it’s a bit yeah. It’s upside down. Go on.

Benjamin Todd: So yeah. So then the question is how do we actually estimate this replaceability factor in different cases? And then I think there’s four main factors to consider. The first one is diminishing returns. So although yeah, adding an extra doctor at the margin does less good than the average doctor seems to do. Yeah, a random doctor is doing the average in their work. But by becoming a doctor, what you actually do is enable there to be, I think there’s around a hundred thousand doctors in the UK, so you allow there to be a hundred thousand and one doctors, supposing you fully increase the number by one. So then the question is how much does that extra doctor actually achieve? And that’s really the counterfactual impact of increasing the number of doctors by one, rather than where you seem to do directly in the job.

Arden Koehler: And that might be less than the average impact?

Benjamin Todd: Yeah. I mean, we should expect it to be way, way less, I think.

Arden Koehler: Just because in general you get diminishing marginal returns. The hundred thousand doctors are already treating the most severely in-need patients, and so you treat somebody a bit less severely in need. Obviously that’s a cartoonish way of putting it, but that’s the basic idea.

Benjamin Todd: Yeah. Though so of course it depends on the area. So I think medicine is something like, “Well, we already have loads of doctors. It’s a pretty established thing, so we should expect there to be pretty big diminishing returns.”

Arden Koehler: It’s a slightly weird time to be making this argument in the middle of the COVID crisis. But anyway, go on.

Benjamin Todd: Yes. So–

Arden Koehler: It might be a bit less diminishing at this moment, but yes.

Benjamin Todd: Yeah. Well, I mean maybe that just illustrates the point, is you have to actually get into the details of the case. And so one thing we could say is that generally large established issues, we should expect there to be more diminishing returns, whereas a really small startup, adding an extra person might not even have any diminishing returns, it might even be increasing returns. And that’s because extra people enables you to specialize, have economies of scale. And so generally when you’re adding people to an organization, generally you get a little bit of increasing returns at the start and then it starts to… It forms an S-curve and then starts to diminish later as you get bigger and you fill the whole market opportunity if it’s a for-profit company. So then yeah, you have to consider the particular cases and figure out.

Benjamin Todd: And then yeah, so that’s diminishing returns. And the second thing is supply and demand effects. So economics 101 would say, “Well, if we increase the supply of people willing and able to become doctors, then well, that would slightly drive down the salaries that you need to pay to get the number of doctors that you want. And then that should mean that all the hospitals hire slightly more doctors than they would have hired. And then that enables them to treat more patients than they would have treated.” And so that’s really hard to work out how big that effect is, because we need to know the supply and demand, the elasticity of firstly the doctor labor market.

Arden Koehler: But so if I’m thinking of becoming a doctor, does this imply that if I become a doctor, there’ll actually probably be more than one doctor added to the pool because it’s me plus whoever they can hire–

Benjamin Todd: Oh, no, no.

Arden Koehler: Okay, sorry.

Benjamin Todd: No, so it would be… It’s probably going to be less than one.

Arden Koehler: I see, because they’re going to spend some of that money on me and not somebody else. So it’s going to drive down the salaries but not enough to make up for that?

Benjamin Todd: So the naive thought would be like, “Well, by you becoming a doctor, doesn’t increase the number at all.” Sorry, if you become willing because you’re just taking someone else’s job. But then that’s not correct because by increasing the supply of doctors, we slightly decrease salaries of doctors. This is at least if it’s in a functioning market equilibrium, and then that means hospitals hire more and then they achieve slightly more. But in general, it’ll probably be less than one whole extra doctor’s worth. But where it lies between zero and one just depends on the supply and demand elasticities of the doctor labor market.

Benjamin Todd: And then also you need to consider the health production function. Because to generate health, we need both labor, which is doctors and it’s also nurses and other healthcare workers and then also capital to pay for medical equipment. All those things can be substituted a little bit for each other. So if doctors become cheaper than they’ll slightly use doctors more and that will enable them to produce more. But how big that effect is depends on how much is the medical system constrained by doctors compared to capital compared to nurses. And so that’s then more like the product market. But yeah, this is–

Arden Koehler: I am starting to become convinced that this is extremely complicated.

Benjamin Todd: Yeah. And maybe using the medical system is not a good example because that’s not a particularly for-profit style market where the considerations most often obviously apply.

Arden Koehler: But don’t we want a nonprofit style… I mean, most people who are thinking about replaceability are probably going to be thinking oftentimes about nonprofit work.

Benjamin Todd: Yes. I mean, I do think you can make analogs of these things for non-profits. Often when replaceability is discussed it’s in the case of like, “Oh, well it’s fine for me to do a harmful job in finance because someone else would have done it otherwise.” And so sometimes it is just regular for-profit jobs that people are discussing.

Arden Koehler: What do you think of that argument, by the way?

Benjamin Todd: Well, yeah, same thing, it doesn’t automatically go through for the reasons we said at the start, it’s like, you might not be perfectly replaceable so maybe if you were not replaceable at all, maybe you would be causing 100% of the harm, now maybe you caused 50% of the harm. But if that harm is big, you still need to not do that. At least, so you just have to empirically estimate how much harm is actually caused by an additional person in this job.

Arden Koehler: You might be better at causing harm than the next person. Right, okay. All right, sorry, go back to the complicated analysis of replaceability.

Benjamin Todd: So then yeah. So there’s a supply and demand effects. I mean, I think with those it will generally be a factor between zero and one. So we know you could be anywhere on the spectrum from fully replaceable to not replaceable, basically. Though, if I had to guess, I would guess that on average it’s a 0.5 correction.

Arden Koehler: That sounds suspiciously convenient. But go on.

Benjamin Todd: It’s like, yeah. I mean there do seem to be cases when you might be fully replaceable on these grounds. So an example of that is when there’s a quantity restriction on the number of people in that job. So in the UK, the number of barristers, which is a certain type of lawyer, there’s a fixed number which is determined by, I don’t know, the barrister guild or whoever that is. And it seems like because there’s this restriction then they have unusually good salaries and conditions and then that means that lots of people would be willing to do it, way more people than there’s jobs. And so then it seems like you become very replaceable in that.

Arden Koehler: That makes sense.

Benjamin Todd: Yeah, whereas on the other side, so that whole thing was like if the labor market is at equilibrium. But if you’re in a field that’s growing really fast and it’s not at equilibrium, you might be in a case where you’re just trying to hire people as fast as you can and you’re limited by your management time. And there you can end up in a situation where it’s threshold hiring. So just anyone who’s above the bar you hire and you keep doing that until you eventually hit your equilibrium. But you’re equilibrium point could be if you’re running a startup that, in theory, could have a thousand employees, but you only have 50 employees now, it might be many years before you hit that equilibrium point and you might be quite talent constrained in that time.

Arden Koehler: At which time people would be totally irreplaceable because you’re actually just adding another person.

Benjamin Todd: Yes.

Arden Koehler: Or I guess you’re just speeding up your–

Benjamin Todd: Well, you’re adding an extra person until you get to that equilibrium point. But that might be like several years, in which case, you’re not really replaceable in that job for those years, which is normally how long people take jobs for. The third one is differences in impact which aren’t captured by salary or by compensation in the job. So everything I was talking about there was some kind of market forces thing. But earlier in the episode we were talking about, well, academics, their salary and their other compensation doesn’t track that well how much impact they’re having. And, in fact, with doctors as well, their salaries are pretty fixed, especially in the UK, there’s a standard consultant salary and a standard GP salary. Whereas it seems like it could be possible that… I think there’s even some evidence about this, that some surgeons are just much better than other surgeons, and so they’re basically having this extra impact that they’re not being compensated for, and so there’s no reason to expect that. Yeah, that basically means if you’re one of those like extra good surgeons, then-

Arden Koehler: Something like you would expect, based on just the economics of it for people to… For like the next person to have a lower impact that better matches the salary.

Benjamin Todd: Yeah, like if compensation was perfectly tracking impact, then organizations would hire extra people but let them have extra impact. And so there would be this efficiency there. But in fact it seems like sometimes people have way more impact than their compensation suggest, which is this additional thing that’s not captured by the replaceability analysis I was just saying.

Arden Koehler: Okay, so these are all the factors that go into–

Benjamin Todd: Well, there’s one more.

Arden Koehler: Oh, there’s one more, I’m sorry.

Benjamin Todd: So then there’s the spillover effects. So that was also what we were talking about earlier. Like if I take this job and then I free up someone else and they go and do something else, that’s good. Then we always need to take that into account when analyzing it. And that can be applied at many levels. That was just at the level of some couple of jobs, but you can also think of it in terms of industries. Like, if I become a doctor and then that means that could free someone else up who would have become a doctor otherwise, but then they go and do something else. And you’d want to think about the impact of that as well. And yeah, if you’re considering an organization or an industry that you think is unusually high impact compared to what people would normally do otherwise, then you can ignore this fourth factor. But sometimes there are important spillover effects such as the example we gave at the start.

Arden Koehler: Yeah. Or as long as you think there are other opportunities that are anywhere in the same ballpark as impactful as the thing that you’re doing, then this will be pretty important.

Benjamin Todd: Yeah. And the person who would have taken your place would do those things.

Arden Koehler: Right. Okay.

Benjamin Todd: So yeah, those are the four factors.

Arden Koehler: But for listeners who want to think about this in their own case, is the advice basically don’t worry too much about it. Or is there anything else to say?

Benjamin Todd: Well, yeah, I think we’ve covered many of the main consequences. So all else equal, it’s better to take jobs that altruists would need to take otherwise.

Arden Koehler: In order for you to have impact?

Benjamin Todd: The spillover, yeah. And then, yeah, in general, I think it’s really complicated and you’d need to do careful analysis before concluding anything about replaceability. And instead I’d encourage people mostly to focus on more robustly important things, like personal fit and how high impact the organization is and whether the cause is pressing and things like that. Though, yeah, I should say also if you do something with really good personal fit, then that is reducing the replaceability because it makes it more likely that you’re better than the person who would have taken your place otherwise, so it’s already partly captured by personal fit. And then the other way it comes into our advice is in the skills vs. funding constraint thing. Because if an area is unusually skill constrained, then that’s essentially a way of saying that people are less replaceable in that area. Yeah, and then we get into some of the arguments that we had earlier and when we try and get people in these areas to estimate how useful extra people are compared to funding, they’ve tried to take into account replaceability when making those estimates.

Arden Koehler: Okay, cool. So I guess stepping back a little bit from these details, it seems like a lot of this stuff that we’ve discussed in this conversation has been a mix of talking about the effective altruism community in particular, and maybe especially people who are focused on our problems that we think are the most pressing, but then sometimes we’ve talked about more general things. You’re talking about the market for doctors. Are there any differences between how the kinds of ideas that we’ve been discussing apply to the effective altruism community versus just everybody who is thinking about doing good versus everybody who’s just thinking about their career?

Benjamin Todd: I mean, all the general concepts apply whatever issue you want to work on. And so you can still try to think, is it more of a skill constraint? What are the skills constraints? Is it more funding constrained? How replaceable might I be? But it’s just that how the analysis turns out might be different. And we haven’t done the analysis on many other problems.

Arden Koehler: Okay, yeah. So maybe it’s individuated by the problems that you want to work on. And of course, if people are just thinking about their personal success and not thinking about doing good, I guess none of these things apply because you don’t care if you’re replaceable or not.

Benjamin Todd: Yeah.

Message for people who might feel at a loss [01:18:34]

Arden Koehler: Okay, so that’s replaceability. And before we were talking about talent/skill/labor constraints versus funding constraints, but all of this is at a relatively abstract level. So if listeners are looking to try to do as much good as they can with their careers, and they’re thinking about making a career decision, I can imagine some of this being a bit demoralizing. So especially the part about maybe we’re in this place right now where there’s these constraints on organizational capacity and it’s hard to find things for people to do who don’t already have these really valuable skills and people who feel like, “Wow, I don’t already have those really valuable skills,” might feel at a loss. What do you have to say to them about how they should conduct a job search in light of these dynamics?

Benjamin Todd: Yeah. So there’s quite a few different things to say here. So one thing is if the key bottlenecks are… Well, we said one is these particular skills that are needed in these problem areas. So given that, one thing you could do is focus on getting those skills. Focusing on career capital for a couple of years. And yeah, there have been lots of cases of people who’ve trained up in something outside one of these issues then come in without experience. So that’s one thing that maybe now seems more attractive than it did in the past when just these roles were often filled by really junior people.

Benjamin Todd: Another thing to say is that many people are overconfident in their chances of succeeding in different things, and many people are also underconfident. And we often speak to people I’m advising who are just like, “Oh, I could never get one of these jobs.That just seems so competitive.” But then they actually do get the job when they apply and they do really well. And so that’s a real shame if you’re in that case. So I think if in doubt, it is better to apply even to these things that you think you might not get. And you can see that there’s a little bit of an asymmetry where making a bunch of job applications maybe takes months or part-time over a year. But if you get it, it’s something you’re going to do for several years or maybe even many, many years beyond that.

Arden Koehler: Well, even if you only do it for several years, presumably it sets you up for something else after you do that.

Benjamin Todd: Yeah, exactly.

Arden Koehler: We talked about this a little bit with Michelle on the podcast with Michelle. So she talked about underconfidence and the importance of making lots of job applications.

Benjamin Todd: Yeah. And it’s not just lots of job applications, but it’s also being a little bit ambitious and going a little bit further than what you might think you’ll definitely get. Though also you also want to have some backup options. As well, in case you’re one of the overconfident people. So–

Arden Koehler: Or just unlucky.

Benjamin Todd: Well, yeah, exactly. So we sometimes hear people going around being like, “Well, yeah, I’ve applied to these three really competitive things. And if I didn’t get that, I’m going to do this other thing.” And the other thing is super competitive as well. And that’s not a solid plan instead of having a backup option as well. Yeah, maybe just another thing to say is a typical job application process has a one to maybe 10% acceptance rate. And so yeah, just for normal jobs, normally that means you need to apply like 10… You need to pursue like 10 to a hundred positions in order to have a high chance of getting a job.

Arden Koehler: If you’re the median candidate?

Benjamin Todd: Well, the average candidate.

Arden Koehler: Average candidate.

Benjamin Todd: And so, yeah, even if an area is already skill constrained, well maybe now the acceptance rate is 6% rather than 2%. So it’s three times higher chance. So it’s, in a sense, very skilled constrained, but you still need to make many applications before getting a job, for sure. And then yeah, one consequences of this is just there’s just not enough jobs within effective altruism branded organizations to give you a guaranteed chance of getting one. In the last effective altruism survey, for the very most engaged people in the community, which is maybe around 1000, 2000 people, I think it was around 76% said that the thing they wanted to go for was working at an effective altruism nonprofit. But it seems like currently there’s not quite enough jobs for everyone to do that.

Benjamin Todd: So it might well be great for them to pursue those positions, because they could be really high impact. So I wouldn’t necessarily say they’re making the wrong call to aim at that thing, but you should also bear in mind that if there’s not enough jobs, then some people will need to do something else. So it’s important to have a backup plan. And yeah, that would take us back to the conversation earlier about how there’s loads of things people could do within policy or academia or other non-profits. And also there’s always earning to give as well. And although we’ve been talking a lot about skill constraints, as I also mentioned, additional funding is still useful. So that’s still a solid option.

Arden Koehler: Yeah. Or earning to save. Yeah, as we talked about a little bit in the previous episode.

Benjamin Todd: Yeah, earning to save and also, yeah, focusing on career capital, personal developments, working with a great team where you can learn a lot, learning useful skills that are useful in lots of pressing problems.

Arden Koehler: Yeah. So maybe people like us at 80,000 Hours and people who are in the effective altruism community can try to focus a bit more on creating more capacity, both within these effective altruism branded organizations, but also as we were talking about earlier, creating/making it easier for people to succeed and flourish in these other institutions while doing a lot of good, and then the people who are taking those jobs can also try to take advantage of those opportunities, help contribute to them, help build those communities. And also just do the very brave work of going and working in an area that unfortunately doesn’t have a community yet.

Benjamin Todd: Yeah, totally.

Arden Koehler: Yeah, and we talk, or Michelle Hutchinson, our head of advising, talks about how people should approach job searches a bunch more in the podcast with her, me and Rob, so listeners can check that out if they want to learn more. All right, well thanks a lot for having this conversation with me, Ben, it’s been really interesting.

Benjamin Todd: Thanks for having me.

Rob’s outro [01:24:21]

Robert Wiblin: If you’d like to learn more about these topics, you should check out a couple of articles on our site.

One is: Think twice before talking about ‘talent gaps’ – clarifying nine misconceptions by Benjamin Todd.

And another is: How replaceable are the top candidates in large hiring rounds? Why the answer flips depending on the distribution of applicant ability – which I wrote.

We’ll link to both of those in the show notes.

Also a reminder that as I mentioned at the start of the episode, the 2020 Effective Altruism Survey just opened. If you’ve made it all the way to the end of this episode the survey is probably aimed at you.

If you’d like to make sure that the survey counts your views on what is most effective, experiences with the community, and what you’re working on, click through the link in the show notes.

If you found out about effective altruism because of this show, it’s especially valuable for you to register that so we can quantify our impact relative to other resources.

The 80,000 Hours Podcast is produced by Keiran Harris.

Audio mastering by Ben Cordell.

Full transcripts are available on our site and made by Zakee Ulhaq.

Thanks for joining, talk to you again soon.

About the show

The 80,000 Hours Podcast features unusually in-depth conversations about the world’s most pressing problems and how you can use your career to solve them. We invite guests pursuing a wide range of career paths - from academics and activists to entrepreneurs and policymakers - to analyse the case for working on different issues, and provide concrete ways to help.

The 80,000 Hours Podcast is produced and edited by Keiran Harris. Get in touch with feedback or guest suggestions by emailing [email protected]

Subscribe by searching for 80,000 Hours wherever you get podcasts, or click one of the buttons below:

If you're new, see the podcast homepage for ideas on where to start, or browse our full episode archive.