The GiveWell Blog

Here’s to fear of failure

A favorite saying of foundation people is, “You can’t be afraid to fail. You have to be ready to take bold risks.”

I’ve thought about it, and I think they’re going too easy on themselves. I put it to you: we need both funders of innovative projects and funders who focus on what already works. But right now, the latter is the one we need more of.

I believe that foundations today almost exclusively focus on high-risk, high-reward, unproven, innovative projects. Exclusively, to the point where there is no mechanism for proven, effective, scalable projects to get as big as they should be. My evidence is not as concrete as I’d like it to be, because foundations are too busy cowering behind locked doors to tell anyone anything about what they do. But here it is anyway.

  • Exhibit A: from the horse’s mouth. Joel Orosz, Insider’s Guide to Grantmaking (pg 19): “Most foundations focus on encouraging innovation rather than on supporting the ongoing programs of nonprofit organizations.” Orosz reasons that individual donors, who give far more, will pick up the slack of funding what works. Problem is, they have no way of doing so. Take it from a guy who tried.

    As I’ll argue below, figuring out what works is actually far harder than figuring out what might work. That’s where foundations’ staffs and resources are most needed.

  • Exhibit B: from the charities. Check out our Round 1 instructions, especially these parts:
    • “We know that helping people can be difficult to measure, but we have a strong preference for methods that we see strong reason to believe have helped people in the past.” (Pg 2)
    • “we suggest that you pick an existing, established, well-documented program for which you have thorough information readily available.” (Pg 3)
    • Please focus on your existing programs and activities … [we] will likely not award programs for which little precedent exists. Rather, we are looking for organizations whose activities have worked well …” (Pg 4, emphasis in original)

    Now check out the Round 1 submissions (available under each cause on Note that even though we worked exclusively with large established charities and asked for projects that have worked before, we still received huge numbers of proposals for completely new projects. The only explanation I can think of is that charities are that accustomed to sending in new ideas, and that unaccustomed to talking about what they already do.

    Now check out the applications that didn’t misread our instructions, and did talk about existing programs. The evaluations are, frankly, a mess, with practically no exceptions. They’re in draft form, they’re disorganized, they neglect to mention clear methodological concerns … for the most part, not even our finalists were able to show us in a clear and readable way what their programs have accomplished in the past.

    I don’t blame charities, I blame foundations. We thought our questions would be easy for charities to answer, but as it turns out, “what do you already do?” is a question that none of them seem ready for. If foundations were asking this question, I have to think they would be.

  • Exhibit C: the logic of laziness. There is no pressure on foundation people to do anything – so I expect them to take the easy way out. Although many point to their “courage” in taking on high-risk projects, in reality a high-risk approach is the easy way out for a funder. When you’re looking at what might work, instead of what does work, you don’t need any analytical abilities; you don’t need to interpret data; and you don’t need to be ashamed when you fail, because after all, you were taking a risk. You can follow your gut and your personal relationships all day long, and shrug your shoulders when it doesn’t pan out. I’m not saying there’s no place for this type of grantmaking, but it sure seems less sweaty than grantmaking based in the here and now.

Since Straw Man boxing is a favorite sport in this sector, I expect many of those who read this to respond, “But what would we do if nobody funded new innovative ideas?” I’m not suggesting that nobody fund new innovative ideas. GiveWell exists because some people took a chance on an idea. So does every other charity. What I’m saying is that right now, foundations work almost exclusively in the high-risk sphere, and that we desperately need more people identifying and scaling up what already works.

We need to learn more about what already works, so we can get those low-risk dollars from individuals to fund it and help lots of people. We need funders who go beyond people they like with ideas that sound cool, and get into the messy business of learning how the world actually is. We need more people who hold themselves to a high burden of proof, and the hard work and analysis that comes with it. We need more people who are afraid to fail. Could you be one of them?


  • Carl Shulman on December 12, 2007 at 6:31 pm said:

    “When you’re looking at what might work, instead of what does work, you don’t need any analytical abilities; …You can follow your gut and your personal relationships all day long, and shrug your shoulders when it doesn’t pan out.”
    It seems that one solution is to evaluate the projects ex post, and fire program officers who don’t get enough hits.

  • That’s exactly what I’m saying doesn’t seem to be happening.

    (As an aside, I would disagree with you about firing program officers who don’t get enough hits. The sample size is too small and the issues are too complex to make things that results-oriented. The purpose of ex post evaluations should be, rather, to figure out whether the thing worked and what that means for future allocation of capital.)

  • Carl Shulman on December 12, 2007 at 6:59 pm said:

    The sample size of grants an officer evaluates varies, as does the confidence level attached to predictions of success (ex ante, do the officers claim that 1/3 or 1/10 of projects will work?)

    Of course ex post evaluations should guide future allocation of capital but there are limits to the ability to generalize to novel areas, and you need some way to create incentives for your ‘venture philanthropy’ staff. I would expect to be fired if I were to fail to deliver large measurable results in at least 20% of my major initiatives, despite limited sample size and extreme complexity, and I am glad to have those extra incentives.

    You could also just limit yourself to incentive pay if you don’t want to fire people for failing.

  • I don’t really agree on the specifics, though this doesn’t matter much because we do agree on the principle of holding people accountable using empirical data.

    I think that in a relatively small organization or for a relatively high-level employee, it’s just silly and unnecessary to have “hard requirements” of the sort you describe. As long as you collect the data and look at it, it’s easy to identify underperformers while factoring in softer evidence.

  • Carl Shulman on December 14, 2007 at 2:22 pm said:

    I think you’re reading more rigidity than I intended. Any good integrated process of employee evaluation (incorporating soft evidence appropriately) should output a general lower bound on acceptable objective performance for a large enough sample. The important thing is that employees can know with some confidence how different results will affect their careers and have good incentives.

    On the other hand, people do tend to overweight softer evidence (which is contaminated by political skills/excuses and corruption/personal relationships), particularly in the nonprofit sector, and tend to go wrong more often than not when they overrule data-driven formulae. See Philip Tetlock’s “Expert Political Judgment,” doctors vs expert systems on medical diagnosis, etc.

  • annakat on September 1, 2008 at 7:11 am said:

    Yes we need to focus on the tried and true ideas that have worked and still work and bring in the money first, then branch out with new ideas, but the first should be what works. I don’t understand why they don’t have a list of priorities on fund raising. IE. (1) List what has worked #1 for years. (2) List what has worked #2 for years, and on down. The should have some type of data to fall back on. Then they should have new ideas which would be added in to the mix.

  • Good Therapy on October 1, 2008 at 1:42 pm said:

    It takes different avenues to make a success of business today and a manager that can take all of the various ways and set a direction. Stopping the tried and true vs. the new and innovative ideas would be a disservice to fund raising, you need both for success.

Comments are closed.