If you think a typical EA cause has very high impact, it seems quite plausible that you can have even higher impact by working one level of “meta” up -- working not on that cause directly, but instead working on getting more people to work on that cause.
For example, while the impact of a donation to the Against Malaria Foundationseems quite large, it should be even more impactful to donate to Charity Science, Giving What We Can, The Life You Can Save, or Raising for Effective Giving, all of which claim to be able to move many dollars to AMF for every dollar donated to them. Likewise, if you think an individual EA is quite valuable because of the impact they’ll have, you may want to invest your time and money not in having a direct impact, but in producing more EAs!
However, while I agree with this logic, I’m nervous about going too far. As Dylan Matthews says, “if you take meta-charity too far, you get a movement that's really good at expanding itself but not necessarily good at actually helping people”. This is what leads to what Matthews called “[d]oing good through aggressive self-promotion” or what I’m calling “the meta trap”. While some meta-projects may have the highest impact in expectation, there are higher-order reasons to want to avoid giving all your resources to meta orgs.
Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.
Which is more probable? (1) Linda is a bank teller or (2) Linda is a bank teller and is active in the feminist movement.
When asked by Tversky and Khaneman, the majority of people picked #2. However, this isn’t possible, since the probability of two events both occurring cannot be greater than the probability of one of those two events occurring.
This is called the conjuction fallacy, and it is a classic bias of human rationality. However, it’s also a classic bias of meta-charity.
-
When you chain different probabilities together, every additional step in the chain will, in almost every case, weaken it. This is also true with chaining together steps of meta-charity together -- while you’re getting higher returns in expected value, you’re also reducing the chance the impact will actually occur.
Consider someone who is considering donating to fund the salary of a staff member to work full-time finding volunteer mentors to advise new EA chapters at various universities. These EA chapters will in turn bring more college students into EA, and these new EAs will then graduate and will all earn-to-give for GiveWell top charities. (While a bit silly-sounding, this plan is so realistic in EA, I’ve actually funded a form of it.)
This plan could have quite a high impact. While donating to AMF all our lives is great, if we can spend our effort to get two people to donate to AMF instead of us, we’ve doubled our impact. If we can spend our effort creating an entire college group to get dozens of people to donate AMF, so much more impact! And then we can expand an entire network of college groups! And then we can become even more efficient in expanding this network. So much impact!
However, we’ve also now constructed a meta-chain that is five steps removed from the actual impact. There’s a lot that can now go wrong in this chain -- the chapters could get set up successfully but fail to get enough people to donate, the chapters could fail to get set up at all due to problems unrelated to the mentoring, the mentors themselves could fail to be better than if the full-time staff member just advised full-time instead, and the staff member could end up being really bad at recruiting volunteer mentors.
This doesn’t mean the chapter chain doesn’t have high expected value or that it’s not worth doing. It just means that it’s risky, and I’m nervous that as the levels of meta scale up, the additional risk taken on by introducing ways to break the chain might be much greater than the additional leverage taken by introducing another meta step.
I do think meta-charities are worth pursuing and I fund them myself. But for every time I think about how good of an opportunity the connections facilitated at EA Global are, I also worry about whether the new EAs brought into the movement really are going to create more counterfactual impact than the considerable cost of the conference.
When I was in college, I once joked about a fictitious club called “Club Club” with the only purpose of perpetuating the club. Every Club Club meeting would be about how to advertise Club Club, how to recruit more Club Club members, and how to better retain the members that Club Club already had. Club Club wouldn’t actually do anything. On days where I’m especially grumpy, I worry that EA may become that.
The problem is that if we spread the idea that meta-orgs are the highest impact opportunity too well, we risk the creation of a meta-movement to spread the meta-movement and nothing else. Once meta-orgs get to the point where it’s all about EAs helping other EAs to help EAs, we’ve gotten to the point where there’s serious risk that actual impact won’t occur.
Consider that plan again where we get someone to full-time find chapter advisors for setting up lots of EA chapters. Now imagine that instead of advocating to the college students that they earn to give for GiveWell charities we suggest that this chapter building project is really the best possible thing to be doing, so they should get involved in, donate to, and volunteer for it. Now we’ve got chapters developing chapters to perpetuate developing more chapters. But what does this actually accomplish? We might as well have them working to set up Club Club.
EA is made up of quite a few different object-level causes and it can be hard to figure out which one is the best. Is it global poverty? Existential risk reduction? Animal welfare? Or something else?
Somehow, meta-work became it’s own cause in this list, but I think that’s a mistake. Meta-work isn’t a cause, it’s a meta-cause, and it’s supposed to make the above actual-causes go better. To understand the meta-impact that meta-work has, it’s important to understand the object-level causes and have opinions on which one is best.
However, I feel like far too often people (including myself) hide behind donations to meta-charity as a feel-good way to support the EA movement as a whole without doing the hard work of figuring out which object-level causes are the best. Unless you’re funding cause prioritization research or hoping to bring in EAs who will shed more light on the question of which cause is the best, this seems like a big risk. It avoids learning opportunities and discussions we could be having about what the best causes actually are, which also pushes the entire movement forward.
Abraham Lincoln is purported to have said “Give me six hours to chop down a tree and I will spend the first four sharpening the axe.” I think this is generally a good philosophy to follow. But at some point you have to swing the axe and actually chop down the tree. If you have six hours to chop down the tree and you spend all six hours sharpening the axe, you’re obviously doing it wrong.
The problem we face with meta tasks is that we don’t really know how much time we have, and we have to allocate this unknown amount of time to axe sharpening (meta-work) and tree chopping (actual impact). At what point should we start chopping? I’m nervous that we may get so carried away with meta-work we forget to actually chop at some point.
GiveWell is considered a meta-org, but they focus on direct research about which cause is best. Historically, they have not focused much resources on outreach or marketing. Instead, they just focused on doing a very good job on their research and delivering high-quality recommendations. In turn, they attracted many donors, including a big foundation. As GiveWell says, “Much of our most valuable publicity and promotion has come from enthusiastic people who actively sought us out” and that they “have generally felt that improving the quality of [their] research, and [the] existing audience’s understanding of it, has been the most important factor in [their] growth”.
Perhaps counter-intuitively, doing really well on object-level stuff could also be one of the best things we can do to grow a quality movement. People aren’t attracted to marketing, they're attracted to people doing a good job. Marketing is only useful in so far as it draws attention to good work.
To be clear, I don’t think the EA movement is in a meta trap yet. I think we’re doing good work and making a lot of progress on important, object-level issues. But I want to be careful. I think the meta trap is a real problem.
Here are two steps I think would work to defuse it:
1.) The more steps away from impact an EA plan is, the more additional scrutiny it should get. The idea of a meta-meta-org may sound unusual, but many EA plans are actually that. This doesn’t mean they’re wrongheaded -- I just think they warrant extra skepticism. Are we really getting extra impact from each step? Or are we just introducing a lot of risk by bringing in another chain that might collapse the whole thing?
2.) More EAs should have a substantial stake in object-level impact. Right now I’m aiming at donating 50% of my pool to the best meta-projects I know and spending the other 50% on direct impact through GiveWell’s top charities. I don’t know if 50% is the correct number, but I hope this will set an example of what I want the movement as a whole to do.
I think Jeff Kaufman at EA Global put it best:
So yes, we should do this, we should put substantial effort into growing the movement. But this isn't the only thing we should do. We can't have an entirely meta movement that goes grow, grow, grow, build growth capacity, bring in people to bring in people, bigger and bigger, and then shift focus? Turn your giant optimized-for-growth movement into an optimized-for-helping one? Not going to work.
We need to do things that help people alongside growing the movement, and personally I try to divide my efforts 50-50. As I argued above, for the doing-good-now portion I think global poverty is our best shot. This isn't settled—EA is all about being open to the best options for helping others, whatever those causes happen to be—but today I think the best you can do to help people now is donate to GiveWell's top charities.“
-
I'm more positive to meta-level work and don't really recognize the risk that EA will fall into a meta-trap. That said, this is an interesting post on an important matter.
Generally speaking, I would guess that humans have a bias against meta-level work, and in favour of object-level work - not the least since the latter usually have more visible and easily understandable effects. They don't spend enough time on various indirect ways of improving our productivitiy, but prefer to just "start getting things done". It was not the least due to this bias that it took such long time before we saw any substantial progress, I'd say.
After the Scientific Revolution, we have seen very substantial progress. We've also seen an enormous explosion in the amount of "meta-work". Whereas people before mostly had object-level jobs (e.g. food production), nowaydays most people work on meta-level work (e.g. improvement of food production, improvement of the cognitive skills of those working on food production, etc).
Hence more people doing meta-work has historically been correlated with progress. Also, note that this development towards more meta-work has always been derided by populists who have claimed that meta-level work isn't "real work". To my mind, that is the real trap, which it's very important not to fall into.
I think there is a certain element of that kind of populism/bias to some of the criticisms of meta-level work. For instance, of course you have to stop working on movement growth at some point. That's not the question. The question is how to prioritize between movement growth and object-level work right now. And here I think that the haste consideration should be heavily weighted.
The most interesting point is the last one. Precisely because people are biased in favour of object-level work, object-level work can be very useful in recruiting. You can do an elevator pitch about earning to give in a way you can't about meta-level work whose rationale hinges on complex chains of reasoning. I think that the marginal value of this effect is rapidly diminishing, though - once you have a few examples of object-level work you can point to, additional examples won't help movement growth that much.
Also I'm not clear over how wide your definition of "meta-level work" is. Sometimes it seems like you're mostly talking about work on movement growth, but organizations like CFAR who don't work on movement growth are also counted as meta-organizations.
Finally, let me add that we should remember that it's not always clear how to distinguish between meta-level and object-level work.
Peter wrote:
I think the synthesis of your and Peter's points is, whether object-level or meta- work, how and how well we do something can be just as important as what we're trying to do in the first place. This distinction and consideration might be neglected by effective altruists when they're trying to figure out how to maximize impact.
Stefan (or, Dr. Schubert? You have a Ph.D., and I don't know what's the most appropriate way to address you in this setting), I commented that I think the current definition effective altruism uses for meta-level work is not granular enough, and this likely causes significant problems in our decision-making. I broke this down into "cause prioritization" and "movement growth". However, you make me realize there needs to be more distinctions than that. Here is my list for dinstinguishing meta-level work in effective altruism[1].
Cause Prioritization: Givewell; Open Philanthropy Project; Global Priorities Project
Charity Prioritization[2]: Givewell; Animal Charity Evaluators; Future of Life Institute
Foundational Research[3]: Foundational Research Institute; Future of Humanity Institute; Leverage Research; Center For Applied Rationality
Community Growth, Development, and Management: Center For Applied Rationality[4]; Giving What We Can; Raising for Effective Giving; The Life You Can Save; Effective Altruism Outreach; Effective Altruism Ventures; Effective Altruism Action; 80,000 Hours
Fundraising and Advocacy: Charity Science; Raising for Effective Giving; The Life You Can Save; Giving What We Can; Effective Altruism Ventures; Centre for Effective Altruism
Original/Applied Research[5]: Center For Applied Rationality; Charity Science; 80,000 Hours
Community Support[6]: CFAR; 80k; CEA; .impact
If you look at all this diversity and how putting all this in one cause makes for strange bedfellows across organizations, I concur with Peter meta-work, while important, isn't a cause in its own right, and it makes more sense to think of different meta-charities just filling different roles within effective altruism. Some meta-charities are specific to an object-level cause or two, and some are not.
[1] Most meta-charities seem to occupy more than one of these charities.
[2] Prioritizing between charities within a single cause, rather than prioritizing between causes.
[3] Research not specific to a single cause, but aimed at improving decision-making, cause prioritization, and cause selection as processes themselves. "Foundational research" is considered reseach not otherwise completed by the scientific community, and is more or less constrained to effective altruism. It can be considered original research in the physical sciences, social sciences, or philosophy.
[4] While not its primary focus, through their workshops CFAR has built an alumni network numbering in the hundreds and constantly growing, the full potential for which both within and outside of effective altruism seems yet untapped. Thus, I consider this an important passive role CFAR plays.
[5] Not the same as "foundational research". "Original research" constitutes applying existing research methods from other science or scholarship to new topics, such as Charity Science doing fundraising experiments, or 80,000 Hours doing careers impact reseach, which apply methods from social science in new ways.
[6] Organizations which play a role in facilitating or improving the effectiveness of individual effective altruists or other organizations.
Evan, (please say Stefan): nice work! I agree the "meta-level" category is too coarse-grained. At the same time, I also think there's a risk of becoming too fine-grained. You could have several different level of granularity, but the uppermost one should be relatively coarse-grained I think (perhaps 5-7 categories, but I'm not sure of the exact numbers), since otherwise it gets too hard to remember. Could one turn your six categories into four - movement growth (increased quantity of members), community support (improved productivity of individual members), prioritization and research? They would in turn have different sub-categories. Not quite sure of this proposal though.
It'd be great if you wrote something more systematic on this. Do include a comprehensive discussion of the reasons for your categorization, though. Criteria could include:
1) The categories should be intuitively clear. You don't want a single category encompassing what is intuitively very different things. 2) Not too many categories at the most coarse-grained level. 3) Each category must denote something sufficiently important. If, e.g. few people are working on a cause, it's better to define it as a sub-category of another cause area.
Yeah, those four categories work better. I'll just do that in the future. I'll also cite your criteria. I have notes corresponding to a more systematic breakdown of this, basically trying to cover all of effective altruism. The other foci I include are causes, and are:
i) Globay Poverty Reduction and Global Health: this includes charities recommended by Givewell and The Life You Can Save.
ii) Animal Stewardship: this includes charities recommended by Animal Charity Evaluations, and additionally Direct Action Everywhere and Faunalytics (organizations which identify with effectiveness as a criterion for doing good, but don't explicitly identify with effective altruism as a movement).
iii) Globally Catastrophic and Existential Risk Reduction: includes MIRI, CSER, FHI, etc.
iv) Policy and Economics: would include the Open Philanthropy Project, your Evidence-Based Policy Project, the Center for Global Development (which has received multiple grants for Good Ventures), EA Policy Analytics, GCRI, GPP, and the Open Borders Action Group.
v) Political and Systemic Change: a category like advocacy, but it looks very different from what EA orgs do. Direct Action Everywhere and Open Borders Action Group are two examples, who raise awareness in very public ways, through, e.g., protest action, more so than effective altruism organizations like Charity Science or Raising for Effective Giving who raise funds and advocate at only semi-public events or on a smaller-scale. These organizations are adjacent to effective altruism but not so core, as there has been dispute about whether the issues they're tackling and how they're tackling them constitute a tractable way of doing things. I may drop this category.
I'm thinking of keeping the division between "Foundational Research" and maybe "Scientific Research", though. While new, improving scientific research is something Open Phil has a good chance of funding in the future, as well as New Harvest and other organizations doing work in the natural sciences entering effective altruism more. I think that constitutes a category in itself. This laboratory work, or whether industrial or academic, likely doesn't have much in common with, e.g., FHI, Leverage Research and the Foundational Research Institute, which mostly do philosophy. Those three organizations seem to have much more common with each other.
However, "scientific research" might be an object-level cause. It's the one cause where it's very difficult to tell how we ought to think of it. For example, is developing a vaccine for ebola an object-level project, or is it a meta-project, with distributing the vaccine being the only true objective of the program? I don't know.
This is a very good start, and I think you're capable of doing this well. I'm thinking, though, that policy-work can be one method of reducing global poverty, whereas giving to charity is another one. So perhaps you want two dimensions - one for methods, and one for objectives.
Or alternatively, you might want to distinguish problem-oriented organizations, which use any method to solve a given problem (say X-risk) from method-oriented organizations which perfect a given method (say influencing the government) which they apply to lots of areas.
You seem to have a very comprehensive grasp of the different EA organizations, which is obviously very useful when you do this categorization work.
I definitely think this is true of humans generally, but I don't think this is true of EAs at all. EAs I see are generally way more excited about meta work than GiveWell top charities.
-
That's an interesting argument from history that I hadn't considered. Thanks.
-
I agree that it's a risk of people generally, but I very much doubt it would happen in EA. Right now I think the balance is clearly toward there being too much meta work than too little. But I guess I could be wrong about that.
-
Right. But I don't think much of the haste consideration, for the reasons I mention in this post.
-
I would include CFAR in my list of meta-orgs and I do think CFAR risks falling into a meta trap. I've been critical of CFAR's argument for impact in the past.
-
What do you mean? I agree it is unclear in some respects. For example, as Eric Herboso pointed out, cause prioritization is meta-level work but should probably be treated as object-level and doesn't face as many meta traps. Likewise, AMF is technically a meta-org in so far as they partner with actual on-the-ground orgs, but they should be considered object-level for this analysis.
I can't think of any orgs people think of as object-level that are more accurately characterized as meta-level.
I think it's true of EAs as well, although to a somewhat smaller extent. But I think we should try to get more data on this issue in order to solve it (see my other comment ).
Good. These are the kind of things I mean. I just mean that we should keep in mind that the distinction isn't sharp.