Summary: The Open Philanthropy Project (Open Phil) aims to follow what it calls hits-based giving, which means it makes risky bets and many of its grants may end up failing. I agree with this idea and I believe that donors should generally be less risk averse. Good Ventures, the foundation that financially backs Open Phil, behaves more conservatively than a “hits-based” approach would predict, and it probably ought to take greater risks in the interest of doing more good.
What do risk-averse and risk-neutral approaches look like?
Let’s begin by talking about investing. Suppose you’re an investor with a billion dollars. You’re risk-averse: if you could take a bet with a 75% chance of doubling your money and a 25% chance of losing it all, you wouldn’t take it, even though this bet has a high expected value–you value your first billion dollars a lot more than you value your second billion. So how do you invest?
- Put some of your money into bonds, which provide a safe and stable (but small) return on investment.
- Put most of your money into riskier ventures like stocks, real estate, and private equity, but buy many different assets to ensure good diversification.
In contrast, a risk-neutral investor is someone who doesn’t care about risk–they don’t care whether they have a billion dollars, or a 50% shot at two billion with a 50% chance of nothing at all. Not everyone agrees on what risk-neutral investors should do, especially because essentially no investors don’t care about risk1, but it involves atypical behaviors like buying stocks with leverage.
Similarly, suppose you’re a risk-averse donor with a billion dollars: all else equal, you prefer to do more good, but you also want to make sure that you have a guaranteed positive impact. What do you do with your money?
- Give some of your money to safe bets that have a high probability of doing good–things like GiveDirectly and the Against Malaria Foundation.
- Give most of your money to many riskier ventures with high expected value. Many of them will fail, but you can support lots of projects, so you have a high chance of doing a lot of good–more than you would just by giving all your money to GiveDirectly.
Truly altruistic donors should not be (very) risk-averse. Self-interested donors have reason to be risk averse: if they do good, they get social status and can feel good about themselves. Doing twice as much good does not earn them twice as much respect. But altruistic donors should maximize expected value. Marginal donated dollars generally do less good than prior dollars, which means altruists should be weakly risk averse with respect to money; but they should not be risk averse with respect to good done, because there is no diminishing marginal utility of utility. Brian Tomasik lists some reasons why altruists are not truly risk-neutral, but generally speaking, altruists should be much more comfortable with risk of not doing good than almost any investor is with risk of losing money2.
If a rich donor wanted to maximize expected value rather than maximizing their probability of doing a lot of good, what would they do?
- Give no money to safe bets unless they ran out of options that had higher expected value.
- Give each marginal dollar to the thing with highest expected value until that thing no longer has the highest expected value; then give to the next-best thing until that no longer looks best; etc. (perhaps giving to a few things in parallel if it’s more efficient).
A risk-neutral donor with a billion dollars would probably not give all billion dollars to one cause, because the billionth marginal dollar would probably do much less good than the first marginal dollar. But they would prioritize filling high-expected-value funding gaps.
Now, how does Good Ventures behave?3
- It gives about half its money to safe bets like GiveDirectly and AMF.
- Through Open Phil, it gives a similar amount of money to a wide variety of projects with varying expected values and varying levels of risk.
This sounds a lot like the risk-averse donor I described, and not as much like the risk-neutral one.
How should Good Ventures behave differently?
If Good Ventures behaved risk-neutrally, we would expect to see some behaviors like these:
- It makes grants that lots of people disagree with because they believe the grants have basically no chance of succeeding.
- It researches a cause or grantee, does not think their approach is promising, but makes a grant anyway because the payoff would be extraordinary if it did work.
- It pretty much fills the room for more funding of the most promising small cause areas.
Good Ventures is doing something like #3–it currently provides a large chunk of the total support for the anti-factory farming and AI safety causes4. For #2, I only know of one instance of Good Ventures making a grant based on this sort of reasoning–its grant to MIRI. I would like to see more grants like this. For example, I believe Open Phil’s decision not to support clean meat development did not make sense for a risk-neutral organization: clean meat has a non-negligible probability of succeeding, and provides a huge benefit if it does succeed. Even if Open Phil’s research led it to believe that clean meat was not technologically feasible, it would still be crazy to say that the technology had less than, say, a 1% chance of working; and 1% is probably enough to justify supporting research in the space5.
I have seen a few Good Ventures grants that I believed should not have been made. But I have always disagreed with a grant because it looked like it would have too small an effect size. I have never thought that a grant had a high payoff but too low a probability of succeeding; I should expect to see a few such grants from a risk-neutral organization6.
In short, we do not see as many risky behaviors as we should expect, and we see Good Ventures investing more money than we’d expect into safe bets with relatively low expected value.
Here are a few concrete actions I believe Good Ventures should take if it wants to behave risk-neutrally. These probably deserve some justification, but I will neglect that for now in the interest of keeping our scope narrow.
- Do not make any grants to GiveWell top charities except insofar as those grants provide them leverage to bring in a lot more money elsewhere.
- Make grants to organizations that it does not expect to succeed, but that have high expected value. The two first examples that come to mind are funding clean meat and making more substantive grants to MIRI. This could also include grants in more esoteric cause areas.
Some investors, such as Tex Thornton and Ronald Brierley, have behaved close to risk-neutrally by staking nearly everything on risky bets with huge payoffs, and as a result some of them have become tremendously wealthy. (Of course, many other investors whose names have been forgotten made similar bets that did not pay off.) ↩
Does Good Ventures have any special reason to be risk-averse that Tomasik does not discuss? Reputational considerations come to mind, although these seem far less important than other factors so I don’t believe they are sufficient to make Good Ventures much more risk averse. ↩
I am not sure how to separate Good Ventures and Open Phil here. Open Phil is not the same entity as Good Ventures, but it gets all its money from Good Ventures, and money that Good Ventures spends elsewhere is money that it’s not giving to Open Phil. Open Phil has claimed that it wants to behave risk-neutrally and I have not seen Good Ventures make such a claim, so Good Ventures could plausibly want to make grants to safe bets like GiveDirectly while funding big risky bets through Open Phil. Regardless, I believe both should want to maximize expected value and should not be risk-averse. ↩
It doesn’t max out their room for more funding, but there exist some non-risk-aversion-related reasons for wanting to do this–it can create bad situations if lots of organizations depend exclusively on you for their funding. I’m not sure how much this matters but Open Phil believes it does and this does not appear obviously unreasonable. ↩
It’s arguable exactly how beneficial clean meat would be, which determines what probability of success we should be willing to accept. Perhaps Open Phil analyzed this internally, but the writeup did not include any such analysis. The writeup primarily addressed why Open Phil does not expect clean meat to work in the near future, without discussing explicit probabilities or whether the upside makes it worth it. ↩
They and I will naturally have some disagreements about the effect sizes and probabilities of success for various grants. Even if we’re both risk-neutral, we will still sometimes disagree because we have different background beliefs and access to different evidence. Thus I should expect to sometimes believe that a grant has too low an expected value, and sometimes believe that a grant has too low a probability of success. If I only ever see the former and not the latter, that suggests that Good Ventures is more risk-averse than I am (and therefore too risk-averse, because we should both be close to risk-neutral). ↩