A familiar pattern: EA organizations promote charities that help people in the developing world. A critic accuses EA of forcing people to be rationality robots. EA defends the use of rationality in altruistic decisions. But both sides miss the point: it demonstrates at best a lack of imagination and at worst coldheartedness to think that only a rationality robot would believe that African lives matter. I'm guilty of this too: it reveals my own prejudices when I think about helping people in the developing world (or livestock) as “giving from the head”, rather than “giving from the heart”. Promoting EA will require changing values, not just making people more rational.
People are not malfunctioning utilitarian robots
Frequently, EA outreach starts from the implicit assumption that, deep down, people value all lives equally. In this narrative, the reason that people don't give to GiveWell-recommended charities is Kahneman-style irrationality. For example, supposedly people have biases such as scope neglect that prevent them from implementing their consequentialist values.
A typical EA example is the comparison between paying for a guide dog to help a blind person in the developed world versus curing many people of blindness in the developing world. To a utilitarian, choosing the former could only result from irrationality. But it's plausible that most people aren't utilitarians and don't care very much about people in the developing world. Even in surveys of philosophers, who would be expected to be more utilitarian than the general population, only a quarter are purely consequentialist.
Rationality alone probably won't lead to EA
Some people might argue that non-utilitarians will become utilitarian if they become more rational. This argument relies implicitly on a belief in moral convergence, which is difficult to defend if one rejects moral realism, as many EAs do. These are very complex debates, which I'll discuss more in a followup post, but the idea that EAs can be created through rationality training alone should be viewed with skepticism. (This is another reason I'm skeptical of the ability of CFAR and similar organizations to have a positive effect outside of some very specific populations.)
A comes before E
In short, people can't optimize for values that they don't have. For the majority of ordinary people, who don't share the egalitarian, utilitarian-ish values of EA, “the most good you can do” is meaningless. This means that we need to start by spreading our values, before talking about implementation. Though rationality exercises won't be useful for this, countless social movements have proven that it is possible to change people's values, typically by combining various types of emotional appeals. Research into the causes of changes in values will be extremely important for the future of EA.
Expanding the circle of compassion
Instead of “the most good you can do”, a better message for some audiences may be “expanding the circle of compassion”. The idea that human culture has become more enlightened by being compassionate to those different from ourselves is catchy, emotionally appealing, and tends to approximate utilitarianism in practice. It may be particularly suited to some audiences, such as religious organizations.
During the holiday season, it's nice to return to the compassionate roots of effective altruism. As Julia Wise says in this excellent post, there's no shame in giving from the heart.
If anyone tests this, it'd be interesting if they reported back here, sharing how it goes.
Also, I think there are definitely some audiences for which expanding the circle of compassion is a better message to put forth than doing the most good. At this point, we're being vague. What are the different "audiences?" What determines who belongs to which one. I think tests about which messages are better for which audiences is better than testing for the assumption that one message is better in general.
Wouldn't this be difficult to test. I mean, the goal of effective altruism outreach is to reach out to thousands and convince them to join the movement. So, if someone posits it's better for us to reach out with a message of expanding the circle of compassion rather than doing the most good, it seems like we'll need to do prior reasoning about what we expect to work better. I mean, will anyone randomly assign activists to spread one or the other message, wait several years to see how convincing one message or the other was to see how it changed all sorts of hard-to-measure behaviors across hundreds of people, ensure those reached out to only receive one message and not the other, even as they may join effective altruism, and reach a conclusion?
When it reaches the scale of building a social movement, it might be beyond the scope of science. In the mean time, others might think it imperative to try building the movement or changing the values of others whether than waiting for the tests results to come back. I'm not saying it's impossible. It's just seems it would be so hard to test this, as hard as anything I can think of really trying, it might defeat the purpose. Vegan or human rights activists don't wait for tests. That hasn't stopped them from being successful. It's possible they could've been more successful, but knowing there will be some success might be better to them than expecting maybe none or maybe more.
Psychologists run experiments at that sort of scale, but they do so in controlled environments. We won't have that privilege. Maybe I'm thinking of something too grand. Maybe you're thinking of small and short-term experiments where people are exposed to one or other behavior and fill out a survey responding how much it changed their impression of helping others. I'd doubt something like that would tell us anything interesting about long-term behavior change though, which is the goal of effective altruism outreach.