(Crosspost) A Response To "Making Discussions In EA Groups Inclusive"

Written by EA Forum user ozymandias.

Recently, an article was posted on the Effective Altruism Forum about making members of marginalized groups feel more included in effective altruism. It was written by more than twelve anonymous contributors. It argues that, because it is important that underrepresented groups feel welcome in effective altruism, we should be particularly cautious about certain conversations that risk making members of underrepresented groups feel uncomfortable. It suggests a long list of topics that effective altruists should think carefully before bringing up in effective altruist spaces, perhaps to the point of de facto banning them.

I agree that it is important to make marginalized people feel welcome in effective altruism, for several reasons. Some categories of people (women, people of color, people with certain disabilities, religious people, poor people) are underrepresented in effective altruism compared to the general population. Something about effective altruism may be driving those people away and causing us to lose out on talent and unique perspectives. Other categories of people (LGBTQA+ people, people with certain other disabilities) are over-represented in effective altruism compared to the general population. If the effective altruism community is homophobic, transphobic, or ableist, these people will experience stress and unhappiness, which is bad for its own sake, as well as perhaps making them less capable of doing good.

Unfortunately, I am afraid that the approach this article is taking sacrifices important effective altruist values while not necessarily succeeding in being welcoming to members of marginalized groups.

I am not against the idea that certain discussion topics should be unwelcome in effective altruist spaces. I myself have argued repeatedly that effective altruism should be secular, which is to say that effective altruist discussions should not touch the subject of religion in any way. Even if the most effective intervention is to convert everyone to your religion, that is not on topic in an effective altruist space, and you should talk about it elsewhere. (That is, of course, separate from supporting a religious organization which happens to implement a program that is highly cost-effective from secular premises.) Even if your true reason for believing something is that God commanded it, you must either come up with an argument that people of all religions can accept, or say “I believe this for religious reasons, so I won’t argue it.” Religious conversations are notoriously heated and hard to resolve, and a lot of progress can be made on animal advocacy, global poverty, and existential risk without resolving whether God exists or not.

Many of the items on the list seem similarly off-topic in effective altruist spaces. For example, I see no reason why effective altruist spaces should host discussions of whether having sex with men is a moral obligation, whether trans people are trying to trick people into sleeping with them, or indeed of sex in general. Similarly, I see no reason for effective altruist spaces to discuss corporal punishment, whether we should kill severely disabled people who are unable to consent to euthanasia, or whether gay people contribute to the survival of the species. These are simply not on topic, and they’re conversations that are likely to get heated and to alienate people.

However, certain of the topics discussed seem crucially related to effective altruist causes. For example, I do not think it is true that we should value people in the developing world less because they are less productive, or that people in the developing world are poor because of character flaws. But if these were true, they would be vitally important crucial considerations for how we direct effective altruist effort. And there are claims which reasonable people believe that an uncharitable person might round off to one of those claims. For example, poverty in the developing world might be related to corrupt and extractive institutions, as economist Daron Acemoglu argues; some people might strawman that position as “people in the developing world are poor because they have character flaws such as corruption.”

Of course, no one is suggesting that the causes of poverty in the developing world should be off-limits as an effective altruist discussion topic; that would be absurd. Instead, it appears that– at least on certain topics– the article is arguing for a one-sided silencing of certain positions on a topic. We can attribute poverty in the developing world to colonialism, inadequate institutions, or the simple absence of economic growth; we can’t attribute it to character flaws. We can argue enthusiastically that people in the developing world matter as much as people in the developed world; we can’t argue that they don’t matter as much.

So the article is not arguing that certain topics should be off-topic; it is arguing that certain topics should be on-topic, but that certain positions on certain topics should be forbidden.

There are several serious issues with one-sided silencing. Obviously, it does not let us self-correct if the forbidden position is right. But even if the forbidden position is wrong, there are serious costs. People who believe the forbidden position can’t bring up their arguments and have them debunked. People who believe correct positions can easily start to strawman their opponents, becoming less persuasive to those who believe incorrect things. The aura of forbidden knowledge may make those positions paradoxically enticing.

I don’t mean to imply that one-sided silencing is always wrong. For example, it would be derailing to bring up in the comments of a post about strategies for measuring diet change that you don’t think animals matter morally. Some conversations have to work from certain shared premises. An Effective Animal Altruists Facebook group could, quite reasonably, ban people for saying that animals don’t matter, because this is in fact not on topic for any discussion they have. But it is very important to sometimes talk about whether animals matter morally. An effective altruism movement in which that was never discussed would be critically impoverished.

The authors bring up that it would be “exhausting and counterproductive” for EAs to always have to discuss whether EA is a good idea in EA groups. But this is an argument against their thesis. It is vitally important that effective altruists consider whether EA is a good idea and engage with the best arguments of their critics. Certainly, it would be derailing to post “I think we have a special obligation to those close to us” in the comments of a post about educational interventions in sub-Saharan Africa. But if effective altruists were very hesitant to discuss the idea that effective altruism is fundamentally misguided, we would be perilously close to being a cult.

I am concerned that, not only will this effort make it more difficult for effective altruists to seek truth, but it will also fail to make effective altruism more welcoming to members of marginalized groups.

I think it is easy to confuse the views of marginalized groups with the views of people who consider themselves to be advocates for those groups. For example, many people believe abortion is an issue where men are generally pro-life and women are generally pro-choice, but in reality women are only slightly more likely than men to believe that abortion should be legal in all circumstances, and are equally likely to believe it should be illegal in all circumstances. While I don’t have polling data about many of the topics the authors of the article suggested should be generally off-limits, I personally know several disabled people who feel very strongly that they should not have children because of the risk of passing on their disabilities. They would feel very unwelcome if forbidden from expressing this opinion in a relevant discussion– especially if the policy were created with the specific intention of making them feel welcome!

The authors failed to address that, while forbidding one topic might be welcoming to one group, it might be unwelcoming to members of other groups. For example, most black people spank their children, at least sometimes; would a black woman who spanks her children feel unwelcome if an overwhelmingly white community is telling her that she can’t defend the parenting practices she chose because she believes they are best for her children? Similarly, many Muslims are socially conservative; is forbidding the expression of socially conservative beliefs unwelcoming to those Muslims? As far as I can tell, this question was not considered.

According to the article, the list of potentially upsetting topics was formed through asking marginalized effective altruists what topics make them feel unwelcome; none of the names of the effective altruists consulted are public. Of course, I understand why people might prefer not to speak under their own names. But I am concerned that this group may not be representative of the groups they’re from. Effective altruists tend to be different from people who aren’t effective altruists in many ways. Unless effort was made to combat this, the group that wrote the questions was probably richer, whiter, more educated, less religious, and more liberal than the groups they’re speaking for. All of these will affect what beliefs they tend to find offensive.

Further, I don’t actually think limiting discussion in this way is the lowest-hanging fruit for making effective altruism welcoming to marginalized groups and groups that are underrepresented in effective altruism.

I believe a more promising approach is the approach I took in my article about how effective altruists can be welcoming to conservatives. I do not suggest that we shouldn’t criticize Donald Trump. I do suggest avoiding jokes that have “conservatives are stupid” as a punchline, highlighting the effective altruist achievements of conservative politicians, using examples from both sides of the political aisle, and remembering that conservatives are in your audience and are listening.

Obviously, not every marginalized or underrepresented group will have similar low-hanging fruit. But I think this sort of approach is very promising, and a lot of it can generalize. Avoid jokes that hinge on a particular underrepresented group having negative traits. Highlight the effective altruist achievements of poor people, less educated people, women, LGBT people, and people of color. Remember that members of marginalized groups are in the audience.

There are other potential sources of low-hanging fruit. I’ve talked to local group organizers who said that they found the best way to attract new female effective altruists is to have two women commit to show up to every meeting, so that new women didn’t feel alienated by being the only woman in the room. I suspect a similar approach could be useful for other visible marginalizations, such as race and some physical disabilities.

Some advice is specific to the marginalized group in question. For example, disabled people often have a difficult time participating without appropriate accommodations. Local effective altruism groups might want to work towards being wheelchair-accessible and fragrance-free. Effective altruist websites might be built using accessibility best practices, such as image descriptions.

I believe that taking this sort of common-sense steps will make effective altruism more inclusive without compromising the ability of effective altruists to seek truth.