EA Diversity: Unpacking Pandora’s Box


I hate talking about diversity. I hate watching other people talk about diversity more*. My first reaction to the (most) recent thread in the EA Facebook group about diversity was ‘oh God, not this again’**. So I have a lot of sympathy for the view, most commonly expressed by social conservatives, that as a society we should spend a whole lot less time thinking and talking about diversity and a whole lot more time deciding on merit. On a purely selfish level, I would enjoy life more that way.

The reasons why I hate this topic are varied, and some of them are too petty or personal to be worth expressing here, but a few of the more intellectual reasons are relevant:

  • This whole area is a communication minefield; words like ‘diversity’, ‘political correctness’, and ‘social justice’ mean dramatically different things to different people. Without active efforts to avoid said mines, participants in such a debate are likely to end up disputing definitions.

  • The Mind Projection fallacy is deadly on a topic where views tend to be both strongly entrenched and strongly based on (divergent) personal experiences. Large gaps in experience lead to large inferential distances, and large inferential distances make persuasion difficult. That is, it is difficult even when you know they exist, never mind when you don’t.

  • The above combine in a nasty way. Opaque or incomplete arguments can get uncritical support from those who agree with the conclusions, partly because their knowledge or experience allow them to unconsciously ‘fill in the blanks’ and partly because of the halo effect. And lots of arguments, in any debate, on any side, are opaque or incomplete. The danger is that you end up with two groups talking past each other, neither of them having the ability and willingness to bridge the gap, yet both increasingly confident in ‘their’ position as they are reinforced by ‘their’ group. Which in all probability they just shouldn’t be.

Reading the above, you could reasonably ask: why am I of all people writing a post about diversity? Because at some point I looked at the world and considered the evidence. Unfortunately for me, my best guess is that diversity is rather important. In particular, I think that it is too important to be left to the type of discussion I just described, and I want to make the case for why it deserves at least some of our attention.

Getting it clear

As noted, diversity can mean a variety of different things. The different things have different upsides. I’m going to non-exhaustively focus on four types:

  1. Diversity of talent.

  2. Diversity of experience.

  3. Diversity of opinion.

  4. Diversity of appearance.

I want to talk about:

  • What does this mean? Why does it matter?

  • How are we doing?

  • What could we do better?

I certainly do not think I have all the answers, especially to the last of those. My hope is that by approaching things in this way we will at least be asking the right (well-defined, relevant, interesting) questions.

Finally, I want to pre-empt the obvious response that these are interlinked; in particular your experiences are linked to your talents, opinions, and appearance. My point is that different types of diversity have different upsides, and that while you can get multiple types at once, you often will not. So I still think they’re worth considering separately.


If you want to get things done and done well, you need a variety of types of people to get there. You need the analytically-minded. You need organisers. You need leaders. You need communicators. And so on. People have widely differing skills and knowledge, and we achieve more by complementing each other, allowing each of us individually to focus on what we’re really good at. This idea appears in economics at the country level, in the form of comparative advantage. On the level of small groups, the concept of varying skills meaning more productivity has a long history, in business and out.

EA is heavy on mathematicians, programmers, economists and philosophers. Those groups can get a lot done, but they can’t get everything done. If we want to grow, I think we could do with more PR types. Because we’re largely web-based, people who understand how to make things visually appealing also seem valuable. My personal experience in London is that we would love more organisers, though I can imagine this varying by location.

With that said, this is not an area I’m too worried about in the medium term. We’re doing pretty well at elite universities. Elite universities score poorly on the other types of diversity I mentioned, but they do turn out people who go into a variety of fields with a variety of strengths; they are diverse in this sense. Just expanding our recruitment there would fix this problem. On a more personal level, I haven’t had any difficulty talking about EA to my non-quantitative, non-analytical friends...from elite universities. I think that’s because communication barriers are more often to do with differing experiences than differing talents (see next section).


Different people have different past experiences. Those experiences are often critical in deciding what you find intuitive, who you can relate to, and how you approach the world. Trying to persuade someone of a new idea is hard anyway, and if you have highly divergent past experiences it is doubly hard. The things you believe immediately might be a big leap for them, and vice-versa; once again inferential distances loom large. If we want to expand, we need varying experiences so that we can communicate effectively with a variety of people. Even once people with differing experiences have reached very similar conclusions, those different routes provide ample opportunities for learning and deeper understanding. On the flip side, homogeneous experiences can easily lead to poor evaluation of the choices and opportunities available to those with different experiences.

In my opinion, this is where EA is weakest. We’re overwhelmingly young, overwhelmingly well-educated, and very well-off even by the standards of our WEIRD base. We lack parents, we lack people with years of experience in business, public service or charity, and for the most part we lack people for whom money or education or both has been a real struggle. At the risk of stating the obvious, it would be very bad if we ended up unintentionally excluding all of these groups.

Some of these issues (parents, age, experience) will plausibly fix themselves over time, because current EAs will enter those groups. Sometimes we might need to do a little work to ensure that people don’t drop out of EA as they drop in to one of these categories. Other issues, like different backgrounds, require more thought, and I’d welcome ideas others have on what can be done here. It might be as simple as re-framing our arguments depending on the audience, or it might be more fundamental.


Different people believe different things. To an extent, we expect homogeneous views within a group of people who were brought together by their shared goals; this is true in spite of the fact that Effective Altruism is a question. But it should still be possible to have a lot of thought diversity; most EAs are irreligious, but our popular 10% idea has its roots in tithes. Going further down that line, religious people are more likely to give to charity (though the relevance of this is disputed) and could be considered a natural target for Singer’s arguments. In a similar vein, most EAs are utilitarian, but there’s nothing uniquely utilitarian about the idea of EA; you can get to much the same conclusion via the Golden Rule. In so far as it is possible, we should want thought diversity, because the evidence suggests that it allows groups to find and correct errors more quickly. Another way of framing the same thing is that high thought diversity makes it easier to distinguish between facts and ideology, because ideologies will have their counter-ideologies expressed within the group.

I’m more optimistic about this area. We have a good variety views on economics, a noticeable but not overwhelming bias against religion, and a lot of healthy debate within the community generated by differing opinions on many other questions. Perhaps more importantly, I think the broad community already recognises our ability to ‘agree-to-disagree’ on these areas as a Good Thing. This should allow us to continue in this way and not fall victim to evaporative cooling.

Our biggest blind spots are probably that we’re short on social conservatives and overwhelmingly cosmopolitan. Perhaps these are just too core to allow compromise on (in which case we might want to reconsider that ‘Effective Altruism is not an ideology’ claim). But I think that people who are not on board with these views could easily still be on board with EA. For instance, there’s no necessary conflict between feeling more attached to those in one’s own country, as a non-cosmopolitan might, and donating to charities in other countries, especially when the effectiveness gap appears to be at least an order of magnitude. Certainly my greater attachment to my near and dear doesn’t stop me donating to effective charities. I sometimes think we give up a little too easily and say ‘well, people with That Particular View are never going to be on board with EA’. Personally, I have more confidence in our ideas than that.


This is where most debates about diversity start, and sadly where many of them finish, which is why I’ve left it to last. I am mostly talking about race and gender here, but the same argument could apply to any quickly identifiable characteristic. For example. it probably also applies to groups of highly-educated people, who are often identifiable by the way they speak rather than look. What I’m not going to talk about is discrimination on an individual level. An understated point in this debate is that you don’t particularly need racists or sexists to have virtual segregation on race or gender. This is in spite of the fact that Schelling’s work on this is almost 5 decades old. I’ll let Tim Harford (see link) talk for me here:

Now these brown eggs aren’t extreme racists; they’re happy to live in a mixed neighbourhood. But they don’t want their white neighbours to outnumber their brown neighbours by more than two-to-one...even a mild preference for the colour of your neighbour can lead to extreme segregation...although we as individuals may be rational and we may be tolerant, the society that we produce together may be neither rational nor tolerant.

How we are doing here seems incredibly variable by location, so I don’t want to generalize too much. One sense in which I think we are arguably doing badly is that many people aren’t aware of the Schelling’s work and also don’t intuitively think in those terms. If someone is a visible minority-of-one in a group, a Good Bayesian should expect that to be contributing to them feeling uncomfortable, even if nobody is actually doing anything to make them uncomfortable. That person also becomes more valuable in the group, because their very presence makes the next person ‘like them’ feel more comfortable, and so on. Note how superficial this arguably is compared to other things I’ve described; merely sharing a race should give you less in common with someone than sharing experiences. But the speed of being able to identify your unusualness matters; first impressions matter.

Further discussion

Are there other types of diversity I haven’t mentioned, that you think are worthy of note?

How would you tackle these issues?

Are they even worth tackling?

*Denise Melchin can surely vouch for this.

**The thread isn’t actually that bad. As I said, this was my first reaction, and it was before there were many comments.