Should More Effective Altruist Organizations Fail?

Y Combinator is about the most successful business incubator to date. It’s hatched 37 companies with worth or sold for over $40 million in its’ few years of life, and has an acceptance rate of less than 6%. However, almost all of these companies fail. As Business Insider points out,“93% of the companies that get accepted by Y Combinator eventually fail”.

Though this is agnowledged as an unfortunate hurdle, it’s also widely believed that quite a bit of business failure is healthy to markets. Poor ideas and implementations get weeded out for better ones, and in the end the winners typically are quite a bit better than average. International capitalism is very harsh but results in quality products.

It would seem to me like the bar for Effective Altruist organizations is significantly higher in many ways than for these businesses. That is not because it’s difficult to fundraise, but because it’s difficult to be effective.

I want to clarify on the demands of an organization that itself is “Effective” in the “Effective Altruist” sense. An “effective” organization here is one where the world is better off with it in existance than if the counterfactual resources required were used for the “next” best utilitarian cause. A lower bound for this should be if all of the organizational employees earned to give in their highest paying jobs possible, and all of the resulting money, with that from organizational donors, would go to the AMF instead.

So for example, if a 5-person division receiving $100,000 per year were “effective”, it would mean that it is doing more good than if all 5 people would have instead gone to work and donated that money (say $20,000 each per person per year), plus that $100,000 all would have gone to save the AMF. This is a really tough requirement given the efficiency of the AMF, but is what some of these groups claim.

Creating a charity or finding any cause more efficient than the Against Malaria Foundation is just incredibly difficult. I’m quite sure that when the AMF was starting it couldn’t have expected to reach the efficiencies and successes that it had. One reason for this is that Givewell gets to cherry pick the most efficient outcomes rather than the best plans, and it seems like these would fall on the end of the gaussian curve of luck as well as skill.

This isn’t to say that it’s a poor idea to create an Effective Altruist organization, in the same way that it’s not a particularly poor idea to create a business knowing the poor success rates. It may be with the right idea you can expect a reasonable expected value, even given the chance of failure.

What it does mean is that I would expect that organizations attempting Effective Altruism should be failing. I’m not sure how many, but 90% wouldn’t intuitively seem unreasonable to me given that’s the bar for what I would assume an easier goal (business success as a Y-Combinator acceptee).

But looking back, have any failed? Have any groups in this area really admitted failure after looking at some difficult metrics?

Some possible groups in this area that may be “Effective Altruist” that come to my mind are:

  • Givewell
  • The Future of Humanity Institute
  • Effective Animal Altruism
  • Effective Fundraising
  • Aidgrade
  • 80,000 Hours
  • Giving What We Can
  • The Life You Can Save
  • The Machine Intelligence Research Institute
  • The Center for Applied Rationality
  • Leverage Research
  • Think! Network

One alternative may be to imagine that we are in a gold rush where there is so much unexplored area in this domain that almost each new venture succeeds. If this were true I would think that there should be a lot new EA organizations until this stops being the case.

Another alternative explanation is that many of these groups aren’t actually “Effective” in the “Effective Altruist” sense, and don’t care to be. Or that there are so many definitions of “Effective Altruist” that each organization gets to pick their own specific one.