Track record and tractability
Historically, some of the highest impact individuals in the EA movement and across the broader world have been people who founded effective charities. The difference between an average charity and the top charities is likely very large. Estimates for how large this difference is range from 10 to 1,000 times more impactful. But even if we look at the estimated difference between GiveWell’s top charities or between ACE’s top charities, we can see the range is extremely wide. The impact of the individuals who have founded the charities at the very top (e.g. THL and AMF) have been massive. This is not even mentioning the founding of more meta-organizations like GiveWell, who has directed millions of dollars to high impact organizations.
There is a perception that founding high impact charities is near impossible or unlikely to happen without the founders having decades of experience and specifically related credentials. However, if we look at the historical evidence it seems this perception is misguided. The following charities were founded by people with less than 4 years experience in a closely related field but had an explicit focus on doing the most good. This is not a complete list, as I do not know the history of many of the other top charities in the EA movement:
- Giving What We Can
- 80,000 Hours
- Charity Science Health
- Good Food Institute
- Fortify Health
This is a fairly impressive list, especially since a good portion of them had absolutely zero experience in the field before starting it.
Although track record gives us a sense of some tractability, asking whether this can be deliberately replicated is still reasonable. However, I think the evidence weighs up favourably for this as well. If we look at the most recent three charities on this list (all founded within the last 2 years), Charity Science Health has gotten two GiveWell incubation grants and has signed up over 180,000 people to their program. Good Food Institute is both ACE recommended and funded by the Open Philanthropy Project and is seen as a leader in its field. Fortify Health was founded less than 6 months ago and is now doing very well and is on track to potentially becoming a top GiveWell charity. These need to be compared to the relatively few failed direct projects founded by EAs to get a sense of what the odds are of a new charity becoming high impact.
Another factor to consider with tractability is the relatively high levels of support the EA movement can give to young projects. Fortify Health for example was legally housed, funded, advised and supported by Charity Science. Similarly, Charity Science Health was given lots of support from other more established charities. Very few charity founders have the strength of community that EA charity founders have which allow them to research a project before founding it or get funding before establishing themselves fully.
A big reason why the founding of so many effective organizations was and still is possible is to due with how the charity market works. Sadly, it's fairly rare for a charity to be established at the start with the explicit goal of being high impact from an EA perspective. A huge number of charities are started out of personal passion or being personally affected by a cause, and although these can end up being high impact, their average effects are much lower than a nonprofit started from a research and impact-focused mindset. This inefficiency makes it much easier to start a high impact charity than an equivalently successful for-profit endeavor. Many organizations that have done research into charitable areas, such as GiveWell and Charity Science, have found non-trivial gaps, even in fairly research-focused areas like global poverty.
It's also worth noting that it can be high impact to start multiple charities in similar areas but with somewhat different focuses or country targets. GiveWell, for example, has recommended several deworming charities. Even more dramatically, it would be easy for someone to say, “Charity Navigator already exists. Why do we need GiveWell?” when clearly the quality and organizational focuses are sufficiently different that founding GiveWell is likely one of the highest impact charities ever started. Furthermore, Elie and Holden might have joined Charity Navigator and tried to change them from the inside. However, that would have been a lot less efficient than them starting their own organization. From what I have seen, people find it difficult to impossible to transform an existing institution, and this is all the more acute the larger it is. Unless you can get a position near the top, starting your own organization allows you to move much more resources towards more cost-effective or evidence-based methods. This will also true in many other areas.
Although EAs have been fairly aggressive on founding meta-charities, relatively few direct charities have been founded by the EA movement, despite the relatively high success levels of the direct charities that have been founded by EA-minded individuals. Aside from the ones listed above, there’s also Evidence Action, New Incentives, The Humane League, and MIRI.
Founding an organization is a powerful way of getting more resources into a high impact direction. As an individual, even a very high talent one, you are limited to 40-60 hours a week of work and generally have an earning potential of under $500,000. However, both these numbers pale in comparison to general organizational scales. Even a smaller organization with say 5 staff and a one million dollar budget, greatly increases your ability to make major progress on an issue. If you hire and fundraise exclusively from EAs, you have to compare your counterfactual to the other place they would have donated/worked at, which sometimes is very high impact and other times less so. This also leads to one of the major assets of charity entrepreneurship. If you succeed in founding a charity that’s better than the current top, you can act as a multiplier to all of the donations going to the present best by shifting them to your newly enacted intervention. If somebody starts a charity that is 1.5 times better than AMF, that will multiply future would-be donations to AMF by 50%, which at tens of millions of dollars per year is huge. However, even if you don’t manage to oust the prevailing org of your area, often within an organization you can have your funding and jobs filled by non-EAs, which greatly alleviates counterfactual concerns.
In addition to the general force multiplier and chance at being very high impact, charities have a chance of becoming very large. The impact of even a marginally better, but very high budget, charity can be extremely large. For example, if Oxfam was 1% more cost-effectiveness focused, it could save an enormous number of additional lives. Founding charities now gives EAs the opportunity to become the next generation of very large nonprofits. Particularly with initial funding sources such as the EA movement and Open Philanthropy Project, it’s not impossible that a charity founded by EAs could grow to be a field leader, which would have immense impact on the world.
Flow through effects
In addition to the case for average impact (e.g. a 10% chance of founding a GiveWell recommended charity) and a case for potentially very large hits (e.g. a 0.001% chance of founding the next Oxfam), there is a strong case that founding direct charities has strong movement building effects. It gives EA a very concrete achievement it can point at as an example of EAs doing something clearly good and high impact. Appearing more action-focused can have major benefits and help offset the perception of EAs as exclusively being focused on philosophy and theoretical concerns. This can draw more people into the EA movement, particularly people that want to see actionable, counterfactually caused accomplishments before connecting to a movement.
There is also the inspirational effect on others to start a similar organization. Evidence Action was started with deliberate reference to evidence and cost-effectiveness and they got recommended by GiveWell, which inspired us to start Charity Science Health, which in turn inspired Fortify Health. I suspect that at a certain point there will be diminishing returns on this particular aspect, but at the moment it’s still very high. The more people who succeed the more people will see that this wasn’t a one-off fluke, but rather something that can be repeated.
An additional effect is if you think value drift is a possible risk, establishing career and social capital in the charity sector is a way to increase the odds of long term altruism. If your CV is in the social sector, even if you value drift, you’re likely to stay in the area because it’s easier to get jobs there.
See previous writing on the value of CE and the expected value of a top charity for more reasons on the impact of charity entrepreneurship. We are also going to write more on this topic soon, including the pros and cons of EAs founding charities and an announcement post for a related organization on this topic.
I appreciate the write up and think founding charities could be a really effective thing to do.
I do wonder if this might be an overly rosey picture for a couple of reasons.
Are there any stories of EAs failing to start charities? If there aren't, that would be a bit strange and I'd want to know why there were no failures. If there are, what happened and why didn't they work? I'm a bit worried about a survivorship effect making it falsely look like starting charities is easy. (On a somewhat releated note, your post may prompt me to finally write up something about my own unsuccessful attempt to start a start up)
One is that some of the charities you mention are offshoots/sister charities of each other - GWWC and 80k, Charity Science Health and Fortify Health. This suggests to me it might be easier to found a second charity than a first one. OPP and GiveWell also fit this mold.
Including AMF is, in some sense a bit odd, because it wasn't (I gather) founded with the intention of being the most effective charity. I say it's odd because, if it hadn't existed, the EA world would have found another charity that it deemed to be the most effective. Unless AMF thought they would be the most effective, they sort of 'got lucky' in that regard.
It's also worth noting that Animal Charity Evaluators started as an 80,000 Hours project and that the Good Food Institute was the brainchild of the Mercy for Animals leadership team.
More precisely, it was started by a student who came to volunteer in Oxford one summer, had the idea and then created it over that summer and afterwards as his brainchild, fundraising to start it as a staffed-up charity, etc. CEA hosted a number of students who came to do volunteer work over summers and other free periods. So while it was labelled an 80,000 Hours project, it's appropriate to use it as an example of someone with little relevant experience starting a charity.
Would be keen to hear your story as I am working to develop better models around what makes projects have success (particularly nonprofits, but I think all data can be helpful).
1) I think this is fair. I have another post in the works on something along these lines. Super long story short though, a lot of the failures are small projects or at an earlier stage vs more like full scale charities. I think that is a problem/concern in its own right, and I think a pretty good case can be made that established charities should be shut down and considered failures more often.
2) I do think a case can be made that second charities are easier to start than first ones (although I would put Fortify Health as quite distinctive from CSH, as my involvement was quite modest in terms of hours). I still think however, there are lots of examples of first time successes.
3) My understanding of AMF from talking to them is that when making the decision that eventually lead to them choosing bednets, Rob M considered that it had to be 1) really big problems 2) really need help 3) might be fixable, as well as some other connected criteria like not tons of other people working on it. From my understanding, quite a few different interventions were considered (e.g. TB, freshwater, landmines). I do not get the sense it was like GiveWell-style shallow reports, but the concept of doing more good was definitely a big part of the decision making.