TLDR: Funding gaps should be talked about in more specific ways, e.g., considering specific cause areas, organizational scale, and funder diversity. When broken down this way, there are likely far more gaps than many EAs imagine. Our model should look more like the table below than a simple “whether EA has strong funding or not” aggregate.
Funding gaps are a well-discussed topic in the EA movement and between funders of all sizes. The concept overall is highly relevant and useful when making decisions. However, I think the bulk of communication about funding gaps is quite unspecific and unrefined, and even carefully communicated content can lead to confusion in overall discourse. A claim like “EA has a funding gap” or “EA does not have a funding gap” is too unspecific a heuristic and can lead to a lot of confusion about the state of funding, as some strong charities receive no funding despite a consensus that the given area is considered “funding flooded.” Working with a number of early-stage charities going through the Charity Entrepreneurship Incubation Program has really given me a more nuanced sense of where specific gaps are and how important this information can be.
Cause area variation
Funding gaps quite clearly change across cause area, and it could be true overall that AI has very few funding gaps while mental health has many. But I think the level of nuance could even go a lot deeper than that. Corporate campaigns in animal welfare might be well-funded while, at the same time, funding in vegan outreach is quite limited. Funding differences according to location can also be radical, with a project in London getting far more funding than the same project based out of Abuja. Most funders and EAs recognize this, although it's sometimes forgotten in the broader discourse. A less-considered factor is that organizations might have other differentiating factors outside of cause area that affect their funding options. This possibility is what I want to delve into more in-depth for this post.
Organizational size variation
A factor that is not often talked about is that the size (and correlating factors like age) of an organization can be just as influential to funding availability as cause or intervention area. For example, there are areas with large governmental funders who only consider organizations of a certain size and age or have requirements that would be near impossible for a smaller organization to fulfill. Governments are not the only entities with these restrictions – large funders are often looking for large opportunities. Some funders are uniquely keen on very large megaprojects but, of course, relatively few of them launch from scratch, with most experiencing a slow build-up of resources and capacity.
Example: global poverty
To take global poverty as an example, GiveWell is typically looking for organizations that can absorb considerable amounts of funding in the near future (e.g., 10 million per year or more). This strategy is well thought out. GiveWell wants to move a huge amount of funding and only has time to evaluate a certain number of organizations to the level of depth they require in order to recommend them. It would take 10x as much evaluation time to consider 10 organizations that can each only absorb 1 million per year. Although this strategy makes sense in GiveWell’s case, it can have a strange effect on the ecosystem. It may result in small projects not getting funding because of the idea that GiveWell would already have funded strong projects in the area (without a donor necessarily knowing whether a charity was ruled out for cost-effectiveness or simply size). It also means there are considerable impacts that may be left on the table from new organizations who have not gotten to that scale or small-scale projects that could be highly cost-effective but do not have the capacity to grow to that size. This means that even in an area that has funding overhangs, there could be and often are still highly promising projects that are funding-limited. For example, my current sense of the poverty funding space based on organizational size is radically different than the impression one might get if they only consider the large-scale funding column.
Differentiated size can, in some ways, be very exciting. Being the funder in an area that has a weak seed stage but a strong large-scale funding stage could allow a given donation to be catalytic and get an organization to the size it needs to be to access other pools of support. It could also mean there are highly impactful opportunities (e.g., at higher levels of cost-effectiveness or with a different risk profile) that could result in more impact per $, even considering only the more direct impacts compared to the well-known impacts of larger-scale organizations.
I think global poverty is a good example of an area that has highly differentiated funding availability depending on organization size, both inside (connecting to GiveWell) and outside (connecting to large-scale foundations and governmental actors) of the EA movement.
Example: animal welfare
Let’s look at a different area that, although much smaller, is balanced much better across organization size: the animal welfare movement.
Even though the total sum of money going into animal welfare is considerably smaller (particularly at the high end), due to its balance, a new well-performing animal welfare organization does not have as clear a valley of death for getting to a certain size. There are, however, funding pathways fairly clearly divided. Even though there are certain size requirements that Open Philanthropy, for example, might have in order to fund animals, places like the Animal Welfare Fund allow smaller projects to ramp up to a size where they could be considered by larger funding bodies down the road. A result of this is that when considering projects founded by CE, although our largest projects end up being poverty initiatives, far more of them struggle in the early stages (including the ones that eventually get to scale) relative to our animal-focused organizations.
Laying out more information
So, this view, broken up by cause area and organization size, adds a lot of nuance to the discussion of what the funding gap is for different organizations. However, I still think it’s pretty inexact and there is more information to consider. Another big factor that connects to funding availability is the diversity of funders in a given space. Particularly, funders with different views and intuitions. One funder with 10 million per year of available funding would almost definitely make different calls than 10 funders with 1 million each, even if all agree broadly with the principles of EA and want to maximize impact. Doing good is messy, and intelligent and reasonable people can disagree on what the most impactful thing to do is. Sometimes an organization will be extremely clear on which areas they cover and which they do not (GiveWell does a good job of this, for example), but every funder has a certain worldview as well as logistical and ethical assumptions (whether stated or not) that affect their donations. So, let's add some information onto our chart here connecting to the number of funders with unique, but still fairly EA, views in a given area. An estimated number of unique funders is now added in brackets to each cell below.
Unsurprisingly, areas with more limited funding will have fewer differentiated funders. More surprisingly, an area can have a very large volume of total funding but relatively few unique effectiveness-minded views. Although animal organizations might have had an easier time scaling up, once they are large, if they have a different view than one of the relatively few funders it could change their situation dramatically. Poverty, on the other hand, has a pretty healthy ecosystem, even when limiting it to just those with semi-EA mindsets. Diverse funding sources also make an ecosystem less likely to be affected by a historical fluke (e.g., if Open Philanthropy is the only/main actor in a space and recommends two areas but only finds a program officer for one of them, this could majorly affect the funding in the space as a whole).
Broader but not deeper
This information, breaking up gaps by cause, organization size, and funder diversity, is about as deep as I personally went into this issue. I think there are other factors that could really change the probability of funding, such as disconnected capacity (for example, if Open Philanthropy was not able to find a good program officer for air quality, there would have been far less funding in the space) or geographic factors.
For a few areas, I did do a full map that I might post at a later time covering how much room for more funding there is and what specific sources give funding vs. what charities report having RFMF. I did, however, broaden it and get some other folks to check the numbers. A little over half a dozen people commented on these numbers and suggested changes to my estimates. Many of them were funders who had a sense of the ecosystem they were in, while the rest were charity CEOs working in the relevant spaces. However, I expect these numbers are far from perfect. I think that in reality there could be more or fewer sources of funding, depending both on the information known (there are some big funders known to only a handful of people in the EA movement) and the definitions used (what counts as impact-focused?). Still, I do think this is a decent estimation of what a new charity would know when getting started with some moderately strong connections in the EA movement (a group I know very well).
I think that taking the same landscape and breaking it up in other novel ways (e.g., not by cause but by evaluation process) could also lead to finding some impactful funding gaps.
I think there is a bit of a habit in the EA movement of not posting something unless you have an extremely high level of confidence in each number, but I think it's better to be imprecisely correct vs. precisely incorrect. I also see moving to a model closer to this level of nuance, even with imperfect data (which, sadly, will always be the case for information like this), as a very positive step forward. Happy to see information in comments or DMs that could suggest updates on this model and, of course, I expect this model to change over time as new funders come on or current funders change their priorities.
Note: I also might write a follow-up post on how the EA movement could facilitate possible solutions to move forward on key “red” gaps in this model.
If anyone has any neartermist community building ideas, I'd be happy to evaluate them at any scale (under $500K to $3M+). I'm on the EA Infrastructure Fund and helping fund more neartermist ideas is one of my biggest projects for the fund. You can contact me at firstname.lastname@example.org to discuss further (though note that my grantmaking on the EAIF is not a part of my work at Rethink Priorities).
Additionally, I'd be happy to discuss with anyone who wants seed funding in global poverty, neartermist EA community building, mental health, family planning, wild animal suffering, biorisk, climate, or broad policy and see how I can get them started.
Thank you for this comment!
I agree it's important to get more detailed about the types of funding gaps. I tried to make a similar point regarding talent here.
I appreciate some of my posts have probably contributed to this problem, though I also think it's important to get the message out that there's a lot more money available in general. I've struggled to strike a good balance.
One additional thing I dislike about the funding gap language is that it makes it sound like the gaps are all or nothing, when in reality there's usually just a smooth diminishing returns curve.
Something that can help with is always specifying what 'bar' the gap is relative to e.g. "there's a not much funding gap at the 20x cash transfers level, but there's a large gap at the 5x transfers level." This language also makes it clear which cause area you're talking about.
Accessibility point (relevant for all Forum posts):
I have deuteranopia (a common form of red-green colour-blindness), and can't really see the different colours in your "limited to very strong" graphs, which makes evaluating them a bit harder and more cognitive effort (I basically have to rely entirely on the text). It's also quite distracting to have what looks to me like subtly different shades of the same colour.
~5% of the population have some form of colour-blindness (~1/12 men, ~1/200 women). I would really appreciate if the colours could please be selected from a colour palette like this one :) Thanks!
I think this doesn't really work super well, primarily because knowledge about funding gaps is pretty anti-inductive. While there is some institutional momentum, my current read is that if one of the large funders notices a funding gap in an ontology as simple as the one proposed here, they start moving into the space and will try to close it. E.g. while OpenPhil used to primarily only fund larger organizations, this has now changed, and OpenPhil is making many more grants to individuals and small institutions.
I think there are many funding gaps, but I expect the ontology outlined in this post to not hold up very well at actually describing them. If you can describe funding gaps as simple as this, then I expect this to change quite quickly after someone notices.
Isn't this fine? It seems like the goal of talking about funding gaps is to have someone fill them, so it's not a problem if the spreadsheet causes someone to fill them. Or are you saying that the large funders have much better information than someone with a spreadsheet, so the market for funding is basically efficient and the spreadsheet wouldn't find anything useful?
Yeah, my model is that this ontology already seems very well within the type of consideration that I expect to be covered by existing funders, and that the current frontier of undiscovered considerations looks substantially more messier or counterintuitive than this.
What do the numbers in brackets mean? Including this information more prominently would make this post a lot more skim-able.
The categories for biorisk look very wrong to me (I don't think there's a funding gap).
Keen to hear about any data on this topic, James is right it is the number of ~EA funders with unique perspectives.
LTFF, OP, SFF, FTXF etc. are all keen to fund bio stuff. If they don't do so in practice, it's because nobody pitches them with good proposals, not because they're not interested. Also, some of the bio grants aren't public.
I think "good proposals" should be disambiguated a bit here. There's a range of possible options* for what you might mean.
I mention this because I personally found it hard to parse your comment, and I expect I'm more familiar with the EA funding landscape than the average reader.
I meant pretty much any of the possible interpretations of goodness, though not the literal interpretation of 'proposal'.
Thanks! I don't have strong evidence for this, but I definitely have a strong prior that we'll miss good grants, evaluated from the POV of benevolent impartial agents with perfect clairvoyance. The world just doesn't seem that fundamentally predictable to me.
I disagree with this, not because we're particularly good at predicting which projects get successful, but because funders have been very generous with money lately (e.g., EAIF and LTFF have had pretty high acceptance rates), which makes it pretty unlikely that we'll miss one.
Oh yeah that's a really good point.
Maybe Joey can clear it up but I believe it's the number of funders in that bucket, as an indication of funder diversity.
Great post, Joey!
Something you touch upon very briefly and that I believe could uncover other important funding gaps is analyzing according to risk profile and the corresponding risk appetite of the supply side of funding. In my experience, risk appetite seems to be dispositional in the funder and not very changeable, leading some people to want to take huge risks and others as little risk as possible.
As a result, we might expect that organizations have more or less access to funding depending on different risk factors:
As a broad example, it seems that people donating to GH&D charities often give to GW charities and are looking for charities for which there is a lot of empirical evidence of the good being done. Many giving to longtermist organizations are convinced by the underlying logic and rationale and are happy to be more speculative, giving to organizations that tackle important problems in way that could reasonable turn out to have no impact of the assumptions don't hold, but could result in massive impact if they do. But there seem to me to be (anecdotally) far fewer people who 1) give to GH&D and 2) are happy to be much more speculative in this way just described, which would lead me to think that charities or other high impact funding opportunities fitting this profile would be comparatively neglected.
Strong upvoted for the idea of breaking down funding in this way - chart is super useful and easy to understand!
(Wrote this earlier; just submitting it as a comment now before discussing this post with EA Austin.)
I didn't find the chart easy to understand. Like Jonas, I couldn't figure out what the numbers in brackets mean.
Additionally, I couldn't tell if the limited / middling / strong classificiations were just intuitive judgment calls based on Joey's experience, or if they were supposed to represent something objective.
I couldn't tell whether limited / middling / strong was supposed to be a measure of availability of funding relative to demand, availability of funding in absolute terms, a measure of how easy it is for someone looking to get funding in this area for a project of a given expected cost-effectivess (and if so whether the cost-effectiveness standard is different or the same for each cause area, or what the standard was at all), or something else.
The combination of the table seeming fairly vague and the fact that some people in the comments strongly disagreed with some of the classifications (e.g. for biorisk) makes me concerned that the table is not as informative as I initially assumed it would be (given that the post had 197 karma when I saw it), leading me to be concerned that some people might assume the classifications in the table mean more than they do. I.e. I'm afraid that this post won't do a good job adding the nuance to the funding gap discussion that it intended to.
Thank you for posting! Is the above based only on EA-aligned charities? Some cause areas (e.g. Mental Health) may lend themselves to for-profit or social enterprise models. I'm not sure how you'd include those
Indeed this is only considering nonprofit funding sources. I think the data would be quite different if also considering for-profit options.
This chart gives me the impression that farmed animal welfare and large global poverty charities are very well funded, just like AI safety and longtermist EA community-building. But I thought that global poverty and farmed animal welfare did have significant funding gaps, whereas the effect of AI safety donations is more like funging with Open Philanthropy. Could you clarify the situation here?
My understanding is that there is still more money within farmed animal welfare and global poverty than opportunities for funding.
For example, GiveWell is aiming to move $1B/yr by 2025 and doesn't currently have enough opportunities identified to cover that. Open Philanthropy also has many millions of global development cash to spend and are still working to identify areas to fund.
Less clear in animal advocacy but my sense is that OP and the EA Animal Welfare Fund are still mainly bottlenecked by getting good funding proposals and that good opportunities generally can get funded ...though maybe I'm missing something - I'm not on the EA Animal Welfare Fund and I don't have access to all their decision making.
Anyone who does have a funding gap in these cause areas is welcome to correct me either in the comments here or via email (email@example.com) and I'll try to explore why it isn't getting funded.
My understanding of GiveWell's post "We aim to cost-effectively direct around $1 billion annually by 2025" is that there are significant funding gaps, but they will probably drop the funding bar from 8× the cost-effectiveness of cash transfers to 5–8× (roughly speaking).
It sounds like there is significant room for more funding here (which Open Phil's neartermist budget wouldn't be enough to completely cover for very long). This seems like a very different funding situation than how the EA Infrastructure Fund and Long-Term Future Fund are generously granting funding to good proposals. I think it's important for the table to distinguish these two scenarios.
For farmed animal welfare, as per the title: ”We need more nuance regarding funding gaps”, I think it is indeed more nuanced than “there is more money than opps for funding in farmed animal welfare.”
Quickly consider the likes of:
On the other hand:
So both opportunity and funding bottlenecks apply at the movement level for farmed animal welfare. But, the nuance is that they really apply to quite differing extents to different parts of the movement.
Thank you for the insightful post!
Declining Marginal Impact of Donations
"For anyone deploying capital, nothing recedes like success.
Back in 1951, when I was attending Ben Graham's class at Columbia, an idea giving me a $10,000 gain improved my investment performance for the year by 100%. Today, an idea producing $500 million adds 1% to our performance.
It's no wonder that my annual results in the 1950s were better by nearly 30% than my annual gains in any subsequent decade."
—Warren Buffett, 1997 Letter to Shareholders
I think you would agree with Buffett on small-scale being more neglected and higher return :)
My updates from your post:
Why do you consider 1-3 funders as "limited" funding in mental health but "very strong" in long-termist EA community building? It might be worth clarifying.
Does biorisk include nuclear risk? If so, I would mention it. If not, I would add nuclear as a cause area.
Updates for the table's" data:
-the largest funder is quitting nuclear safety
-FTX Future Fund is a new megadonor in AI, biorisk, nuclear, policy, and EA community building.
I love "Doing good is messy"—I will add it to my Anki quotes.
thanks so much for sharing these thoughts, Joey. Really clearly laid out and so important to be able to understand the nuances.
Very good description of the problem. It aligns with what I see in the meta funding space.
More info always seems better, but maybe it's not useful here?
My thinking is that perhaps all the gaps worth filling are already well known and being addressed roughly as soon as they become overdetermined. Other gaps maybe aren't worth addressing because the expected value of doing so is low. More info might help identify the marginal gap, but if there's something like a power law distribution of gaps in terms of expected value of filling them then we've likely already identified all the best ones to fill and the rest are the long tail where differences don't matter much and people should fill based on other criteria.