The Center for Effective Altruism commissioned me to write an essay on Giving What We Can’s pledge to give 10% of your income to the most effective charities. They did not ask that the essay have any particular contents, but I expect that I would not have agreed to write it at all if I’d thought that the pledge had a pretty negligible and uninteresting effect. I am a member of Giving What We Can.

There are two interesting questions about the GWWC pledge, “should I take it?” and “is it a good thing when more people take it?” I’m mostly interested in the second one, because people have written some thoughtful perspectives on the first (which is in any event highly individual: if the pledge will move you to more consistently save and budget in a way consistent with your values, you should take it; otherwise you shouldn’t.) The second question has received less attention, and yet more people are taking the pledge; Giving What We Can has recorded an average of 80% year-over-year growth. More people took the pledge last month than in the first three years of Giving What We Can’s existence.

Here are some reasons to think that’s a good thing:

There is something of a consensus in the effective altruist community right now that one should expect to do a great deal of good with their money. OpenPhil expects to spend their last dollar on better giving opportunities than GiveWell’s top charities, which do as much good as saving a life for about $3200. Effective small donors have reason to think they can do as well as OpenPhil. And many EA-affiliated charities and charities which many EAs want to see fully funded had a difficult time this year raising enough money to support their operations; most fell short of targets for expanding operations. An EA movement that directs more money likely does a lot more good.

It would be unhealthy for the effective altruism community if it was made up mostly of people who like talking on the internet about GiveWell and MIRI and not of people who tried to effectively direct resources where they’ll do the most. A few years ago there was an EA community survey that revealed that only a quarter of self-identified EA respondents donated at least 1% of their income, and the median donation was $450. That prompted some wondering if we were headed towards becoming such a community. I thought at the time that those fears were unfounded; it was a convenience sample, the average age was 25 with many participants students and many others doing direct work. I’d be curious to see the results of a similar survey today, but as far as I know one hasn’t been conducted. Until it is, I’ll be pretty uncertain about the benefits of movement-building: creating more people who identify with EA does nothing unless it creates more people who effectively do good. The pledge, though, is a sign of community growth that is very closely tied to ‘additional resources directed to important causes’. I think if the fastest-growing parts of EA are the parts with people pledging to donate at least 10% of their salary, that’ll be evidence that growth does not risk creating only uninvolved and vaguely affiliated people. (It is not particular reassurance that growth does not risk dilution of the ‘effectiveness’ part of the message, a concern I’ll discuss below).

And finally, it means a lot of very valuable data. If Giving What We Can is tracking their pledge members, we should have data soon on whether pledge-takers become more or less involved in effective altruism over the years after taking the pledge, where they donate and how much time and thought they put into the decision where to donate, and of course at what rate they keep their pledges. I think the value of this information is very high, and so I am excited at the high pledge rate.

There are also some reasons to think the pledges could be a bad thing. Firstly, if pledging is a bad idea for each individual effective altruist, or for many of them, such that someone taking the pledge would in expectation have been better off not taking it. People who are likely to be deterred from doing direct work by having taken the pledge probably should not take it; people who are likely to be unwilling to make creative donation decisions, like donating every other year to maximize deductions, as a result of their pledge probably should not take it. If people who take the pledge experience some kind of moral licensing effect and are actually less likely to donate, then no one should take the pledge. I take the first two concerns seriously but expect they apply to relatively few people, and that other people are more likely to do direct work as a result of becoming more firmly a part of the EA community through the pledge. I would be very surprised if the moral licensing effect were present.

Another potential negative effect of the pledge is that people might be less motivated to increase their income (since the pledge is for a fixed percentage, while aggressively pursuing a higher salary might be a better way to move more money). This seems like it could be more than contravened by following up with pledge-takers and offering consultations on how to increase their salary (or pursue direct work), but a better-formulated pledge might avoid the problem entirely. On the other hand, The Life You Can Save has a pledge adjusted for income with a nuanced formula, and it hasn’t caught on like the GWWC pledge; maybe people like benchmarks like 10% a lot more, enough to make that the best target even if it creates slightly less than optimal incentives.

Secondly, the pledge growth could reflect lots of people joining who don’t choose highly effective charities, even if they fulfill their pledges. I expect the cause priorities and favored charities of people recruited through Giving What We Can to systematically differ from those of people who joined through other sources; if they donate less effectively, then they could still drag the focus and energy of the EA movement away from the best opportunities. I expect this to be less of a problem with people newly pledged through GWWC than with people newly joining the EA community from other sources, though, since the pledge involves more thought and more investment.

Finally, the pledges could mostly fail to actually donate or stick to their donations, and the pledge drive has significant opportunity costs in money and staff time. If Giving What We Can is wrong about how much money to anticipate in expectation from new pledges, then they’ll have misallocated resources. My guess is that they are too optimistic about how much money to anticipate in expectation, and I’d more wholeheartedly endorse their pledge drive if they were reasoning from lower expected future donations, but even by conservative assumptions the drive looks like a good use of the resources that go into it. On the whole, seeing the year-over-year growth of the Giving What We Can pledge has made me a bit more optimistic that EA movement growth can be sustainable, impactful, and not damaging to our ability to think about hard problems.