I think this point could be basically seen as “we don’t want to make the appearance of powerseeking” worse

It's not about appearances for me. A portion of EA is seeking power, by any normal definition of the words. In AI, we want to influence government policy on AI regulation, and to influence the direction of AI companies and AI company research. This has resulted in EA people having significant power in the AI space. I would guess that all the top AI companies have a significant EA presence in some form or another. 

Seeking power is not a bad thing. There will always be power in the world, and I would rather it be in the hands of good people with ethical goals. I personally want people who share my political ideals to seek and gain power so they can implement their beneficial policies. 

The problem is that if all the people actually gaining power are from a highly privileged, homogenous minority of the population, there is a high chance of blind spots, bias, and self-interest creeping in. This could easily lead to the power being deployed, even unintentionally, in service of the privileged minority at the expense of everyone else. This has happened so many times in history it's hard to count, I think it's naïve to pretend it couldn't happen to us. 

Great post - there are potential trade-offs here which are underdiscussed.

I’m definitely in the global EA camp, but not primarily for justice / equality reasons. I think the main benefits of global EA are:

1. Cause Prioritisation: In practice, I think cause prioritisation in EA is heavily influenced by culture, wider society etc, and not just by rational thought. I think cause prioritisation influenced by every living culture is more likely to be closer to some universal truth, than cause prioritisation primarily influenced by the cultures of the US, the UK, the West and high income countries.

2. Local Problems, Local Influence and Local Knowledge: I think many high impact interventions in biosecurity, global health and well-being and farmed animal welfare in EA require extensive local action in middle and low income countries. Narrow EA will both lack the local knowledge and local influence / power to deliver these interventions.

3. Capture by / Drift Towards Selfish Elite Interests: You acknowledge a related optics concern in “I have a sense that the way that this strategy is pursued often leads to an optic of secrecy and icky elite-control”, but I think cause prioritisation in EA is very uncertain and therefore very vulnerable to capture by the selfish interests of groups inside it. I think global diversity of EAs protects against this risk.

4. Cost-Effectiveness of Community Building: Money goes further in low income countries, which means the same investments in community building should lead to more EAs.

5. Outsized Influence of Small Countries in Multilateral Orgs: This isn’t about middle and low income countries specifically, but I think CEA and Open Phil should specifically invest in community building focused on careers in government in a country with a very small population, to help the country advocate for good ideas in multilateral organisations.

I continue to think that EA Community Building should prioritise places like Oxford and Harvard.

But I also think places like Delhi or Nairobi should be prioritised over less influential British universities like Durham, Edinburgh, Sussex, Bristol, etc.

This is pretty emotionally raw

FWIW it came across as pretty level-headed to me.

I’ve started feeling super guilty and sad about how much I, the EA community, have wasted on supporting my participation in various community building and research endeavors - I’m not really any more capable or competent at doing the things I’ve done than a local American graduate would have been.

I obviously know much less about you than you do. But speaking to my own experiences, the second part of this rings false:

• I wouldn't hire someone to do something for me unless I thought they were the best person who would accept an offer. This makes me think that the people who hired you thought you were more capable and competent than the other available people.
• It's easy to overestimate how many people are as capable and competent as you are — especially if you've spent most of your working life in a bubble where all the people around you were hired through an intense, selective process. I gave up on EA work for well over a year because I overestimated the world (or underestimated myself).
• Someone getting hired from a very different background, such that they had to struggle more and overcome greater barriers just to be competitive, seems likely to end up being more capable/competent than their average peer in the long run, even if they were roughly equal at the time they were hired. 
  • If one person starts a 1600m race 50 meters ahead of someone else, but they're both tied as they finish their first lap around the track, the person who started from behind is probably going to win. In this metaphor, you're the person who seemingly started behind (though I'm making an assumption from very little knowledge); if I were looking for talented people to bet on, you'd stand out on that account.
  • More broadly, I'd guess this is part of the reason that such a large fraction of successful U.S. startups have at least one immigrant founder; at whatever point these founders successfully entered the American business scene, they were presumably growing / developing a bit faster, on average, than their local rivals.

I think the community-level questions you raise are worthy of much more discussion, and there's no way I can address them fairly in the scope of this comment — but apart from that, I do endorse raising your own self-estimation a bit.

Great post! 

On whether the decision has been made to do 'narrow EA' as opposed to 'global EA', I also have the impression that this is what leaders such as those on the longtermist CB team at OP have decided. I think this decision has been made on the basis of research such as this. I also agree that more transparency on this would have some benefits. 

The decision to go for 'narrow EA' seems to go against most of the social change research I'm familiar with, which usually stresses the importance of both 'inside track' and 'outside track' interventions. But I'm not sure how much weight to give this research.

My gut feeling is that, putting to one side the question of which is the most effective strategy for reducing x-risk etc., the 'narrow EA' strategy is a mistake because there's a good chance it is unethical to try to guide society without broader societal participation. 

In other words, if MacAskill argues here we should get our shit together first and then either a) collectively decide on a way forward or b) allow for everyone to make their own way forward, I think it's also important that 'the getting our shit together' has broad societal participation. 

The paragraphs below are partly responding to your framing of the issue here. If you frame it as "we can either have 4x attendees and not be inclusive by sponsering flights and visas, or 3x attendees and be inclusive" that's pursuasive, if you're saying "we can cut the costs of this conference by a large amount by not sponsering any flights or visas, which means more malaria nets or more ai grants and I think that's worth it" that's potentially pursuasive, but when you frame it as about about the project of inclusion in general, then I do feel like you're making a mistake of unevenly placed skepticism here. 

I think community builders and those funding/steering community building efforts should be more explicit and open about what their theory of change for global community building is

I do think meta orgs could be clearer about their theory of change, but to get there via the questioning of the value of diversity seems like an odd reasoning path, the lack of clarity is so much deeper than that! I feel like there is some selective scepticism going on here. If you apply this skepticism to the bigger picture then I don't see why one ought to zero in on diversity initiatives in particular as the problem. 

Firstly, I think it would be illustrative if you said what you think is the point of community building, in your view? Community building is inherently pretty vague and diffuse as a path to impact and why you do it changes what you do.

 For instance, suppose you think the point of community is to recruit new staff. Then I'd say maybe you ought to focus on targetted headhunting specifically rather than doing community-building? Or failing that, training people for roles? As far as non-technical roles, it doesn't seem like there's a huge shortage of 95th+ percentile generally-high-talent people who want an EA job but don't have one, but there's lots of work to be done in vetting them, or training them. As far as technical roles, you can try and figure out who the leaders of relevant technical fields are and recruit them directly. If I wanted to just maximize staff hires I wouldn't do community building, I'd do headhunting, training, vetting, recruitment, matchmaking etc in tight conjunction with the high impact orgs i was trying to serve.

 Or, if you think the point of the community building is to have meetings between key players, then why not just only invite existing staff members within your specific cause area in a small room? From a networking perspective community building is too diffuse, there's not much in the way of real professional reasons why the AI safety people and the animal rights people need to meet. You don't need a huge conference or local groups for that.
 
I think when someone focuses on community building, when someone thinks that's the best way to make change, then (assuming they are thinking from an impact maximizing perspective at all, i suspect at least some resources people direct towards meta has more in common with the psychology of donating to your university or volunteering with your church than with cold utilitarian calculus, and i think that's okay) they're probably thinking of stuff which is quite indirect and harder to quantify, like the value of having people from very different segments of the community who would ordinarily have no reason to meet encounter each other, or the value of provoding some local way to connect to EA for everyone who is part of it. For these purposes, being geographically inclusive makes sense. Questions like whether people could sponser their own flights depend on how valueable you think that type of community building is, I agree that there's a difference between thinking it's valueable and thinking it's valuable enough to fly everyone in even if they don't have a clear intent to work on something that requires flying in like you did.  If community building is intended to capture soft and not easily quantified effects that don't have an obvious reason behind them, then I don't see why those soft and not easily quantified effects shouldn't include global outreach. Fostering connections between community members even if they work in different areas or are across the globe from each other, or taking advantage of word of mouth spread in local contexts, or the benefits of having soft ties on each continent such as a friendly base for an EA travelling for work to crash or having a friend of a friend who works in the right government agency for your policy proposal, seem like a valid type of "soft and hard to quantify" effect. Like right now, you can throw a dart on the map, and you will probably be able to find one ea in that country to stay with, and if you throw a few more darts then you can probably find an ea in a government and so on, in a handwavey sense I most people would say that this is a generally beneficial effect of doing inclusive global outreach for any policy or ngo goal.

Whereas if you don't have much faith in that soft and hard to quantify narrative, if you're pursuing hard, quantified impact maximizing, then why do community outreach at all? Why not instead work on something more direct like headhunting or fund some more direct work?

I'm sympathetic to "this theory of change isn't clear enough", it just seems weird to me that if you've accepted all the other unclear things about the community building theory of change, that you would worry about inclusion efforts specifically. If you were sending out malaria nets I would understand if you made the choice that gave out the most nets even if it was less inclusive, because in that scenario at least you would have some chance of accurately predicting when inclusion reduced your nets. But in community building, that doesn't make as much sense, if inclusion is hurting your bottom line how would you even know it? I feel like maybe you have to have a harder model of what your theory of change is before you can go around saying "regrettably, inclusion efforts funge against our bottom line in our theory of change", because it seems to me like on soft and fuzzy not very quantified models of impact, inclusion efforts and global reach mostly make as much sense as any other community building impact model, and when one is in that scenario why not do the common sensically positive thing of being inclusive at least when it's not very expensive to do so? 

Wow this is an impressively articulated post.

The one comment which you made which seems likely to be false, based even just on the clarity of thought and communication in this post alone is...

"there was no real justification for me to spend my own money moving to the US and staying here to work on these things whilst someone from Boston could have worked on them instead."

I might push it a bit further and suggest that perhaps your background and the strength of this post could be a small piece of evidence towards one of your "reasons I might be wrong", namely "might be significantly underestimating the value of diversity and inclusion "

I wonder if the focus on "narrow EA" is a reflection of short AI timelines and/or a belief that we need to make changes sooner rather than later.

It seems to me that "global EA" looks better the longer the future we have. Gains in people compound, and other countries may be much more influential in the future. If Nigeria is a global power in 50 years, then growing a community there now might be a good investment.

I don't think this has a clear answer though, since benefits from actually solving problems sooner can compound too.

I’ve started feeling super guilty and sad about how much I, and the EA community, have wasted on supporting my participation i

I think that in saying this, you're technically putting a rather low upper bound on the marginal value of a community building staff member as lower than your recruitment and moving costs, which has implications for what you ought to think about community building (vs earning to give an amount greater than your moving costs, or working in a different cause area).

To expand on this in more detail: I think there is something incoherent in saying that it wasn't justified for you in particular to move to the US with the 100% intention to work (which is a much stronger case than flying people out to conferences who might one day work), but it is justifed for you to work on the project now that you have moved.  Why not discount the value of working on the project at all even if you are local, if there's truly such a big supply of other locals ready to do it who could have done just as well, such that the marginal impact is ultimately less than a moving and recruit cost? You can probably find a more neglected project to work on instead, one in which no one equally talented would replace you, one which is important enough that the flight isn't a material consideration, right? 

In fact why do community building and work so hard to recruit anyone, if it's not even worth the cost of flying yourself and your suitcases on site to get one new recruit, then why is it worth spending so many much more expensive hours of expensive labor on recruitment in general, regardless of if they're local to an ea hub or not? I wouldn't be surprised if the cost of staff time spent to vet and recruit and hire one local is greater than your moving costs.  

Is the scale really so precisely balaned that the your flight is what tips it, probably not, probably either you're working on entirely the wrong thing, or you're working on the right thing and the flight and moving is a rounding error. If money is at such a premium that moving tips the balance, why not instead move to the highest earning city and earn to give? I bet boston/nyc jobs pay better than south african ones, you can make back the cost of the moving manyfold. 

So I think, logically speaking, either you're placing  the value of community building as net positive but inferior to earning to give an amount more than your move and recruit cost, or you should think the move was a net-positive relative to the next best person being hired, there isn't really a coherent narrative where the move wasn't worth it and locals who don't need to move should on the margin continue to work on community building rather than ETG (or a better direct work alternative).
 
(Edit: moving this to its own comment as it is a separate point)

I’m not really any more capable or competent at doing the things I’ve done than a local American graduate would have been - there was no real justification for me to spend my own money moving to the US and staying here to work on these things whilst someone from Boston could have worked on them instead. 

I dunno, moving/immigrating isn't that expensive in relative terms, when compared to the other costs of hiring people and doing great work. Like we're talking about a one-time cost of maybe 5-100k for a potential lifetime of valuable EA work (whether direct work or earning-to-give^[1]). 

Also, standard economics models would suggest that people move too rarely, rather than too frequently. 

1. ^{^}

   Importantly, even if someone stops being interested in direct EA work, or otherwise can't find a good fit, US salaries are just so much higher than salaries elsewhere (especially compared to developing countries), particularly on the high end. So the earnings potential alone is a strong argument for working in or immigrating to the US, on consequentialist moral grounds.

I lean more towards the narrow EA path of impact, but not too narrow.

I guess I think it’s important to fund at least some non-Western groups in order to provide a check against any potential cultural biases that we have.

But it goes beyond that, people from another culture might come up with different ideas or approaches that we might consider, even if there isn’t an obvious direct line between their culture and idea.

Now this has to be balanced against the counterfactual use of resources, but at a certain point, there’s decreasing marginal returns from recruiting yet another Ivy League/Oxbridge grad. And our most likely cause of failure stops being that we don’t have enough of those.

Sometimes I try to channel my anger at birth lotteries into motivation, along the lines of "someone who'd do a way better job than you in this situation is muscled out cuz they didn't have the foresight to be born correctly, so you really owe it to them not to eff it up", and the results are just as mixed as any "obligation" framing. Yet it seems more productive than convincing myself that that anger is actually the object level cause area, cuz if I did that I would just read the news all day about why outgroup is keeping us from making immigration policy less wrong and get zero done. 

Supporting a global EA community is expensive - e.g flying people to conferences in the US and UK from places like South Africa and India is often ~4X the price of local attendees travel costs; we have to sponsor travel and work visas.

Well, it is, but only as long as you assume that all conferences should be held in the US and the UK in the first place (for discussions on this, see this and this). 

Footnote 3 cracked me up.

I have mixed-feelings concerning your post. I commend it for being so clear and easily understandable. There are trends in EA, and yours go into the 'Let's narrow EA because diversity is overrated'. As you say, you have no empirical evidence that diversity is bad ; and I will just give a few qualitative examples to demonstrate the opposite, so my level of confidence is around 55%, no more. Sorry in advance if the arguments I'll lay will be hard to take in. Again, this is only my hunch.

1. It is well-known mechanism for people previously on the margins who have succeeded into prestigious, privileged places to be against diversity. I think about Priti Patel or Rishi Sunak ; take me in, but don't let anyone enter after me. I believe this comes from a desire of being liked by those who you admire and want to feel as an equal--turning back on the very people you could have belonged to if you had been born ten years later is a way to legitimize your ozn belonging to the prestigious, powerful group. To push further, because it's not you being 'mean', it stems from a place of self-doubt. You are not sure you deserve to be here (cf you referring not being better than your Boston pals) so its easier to think others like you don't. There is a sociological field that explains such attitude, but I can't find it now sadly. Some kind of survivor bias. 
2. I do think that fellow South Africans or people coming from low-income countries can bring insights EAs in wealthy countries can't. Thats the reason why the UN recruits people from these countries instead of giving the job to a white, wealthy candidate from a first world country who just got their masters in development. Its because people from these countries have experience living in these countries, and this a major arguments for global aid projects , which remain very important as McAskill said in a post here not a long time ago. AI devouring the rest of EA fields is a great concern for many, and I belong to those who think that AI is a big thing on its own--it requires a set of skills and political connections that must be separated from animal welfare, global aid, etc. I can develop if needed. AI is vastly important. AI is not the core of EA (yet, I hope). 
3. 'Something which is largely just a nod to political correctness or a lost sense of global justice' and there lies a big worry of mine. As an historian, I know that progress is not linear, and that progress is won by hard work (listen to the 80k podcast about slavery not being inevitable at all). This kind of argument is a pushback from a group of people who do not like to share resources, power and influence. And we see this pushback happening every single time progress is made, every time a group gets rights they were denied previously under false pretenses (racism or misogyny is a false pretense in my opinion). The problem here is that there is a hardcore group of people inside EA who do not relinquish competition from other people. Yeah, I know, EA is not supposed to be political. And I shall not be: my view is very pragmatic and I want the best people to do good. 

I know that such discourse isn't well-viewed in the community. And yet every day I see excellent people from the 'diversity' pool (understand women/non-white/non-STEM) being great at their new job, that they would not have gotten ten years ago before all the 'political correctness' talks. I intend to write a post giving figures as to why diversity is, in fact, necessary to keep producing the best people. And to give actionable answers. I expect pushback--that is the course of history :)

So, people. Think, I beg you. It is scary to have the feeling that while it is already hard to get an EA job, you have to compete with an always bigger pool of applicants because we give a chance to everyone. But if you care about impact, it is a reality everyone should accept. 

I see that I am regularly downvoted, just like I was expecting, but I was expecting people to least answer in response, explaining their reasoning. Because it's now easy to qssume that I am being downvoted by people who fear competition and refuse diversity for selfish reasons. Ockham razor here.

Edit: I see that either I'm not conveying my point well, or that people aren't ready to hear this. I still think that this 'pulling up the ladder behind' thinking applies to OP, and if it's offensive to say that, I'm sorry, and I mean it. I don't see how it is offensive--they don't know me, they don't care about what I think of them, so if it's not on a personal level, I don't see how this is hurtful. If someone accuses me of being something I don't think I am, I'll brush it off--I know myself, and if it bothers me I'll try to see if it's because there is a parcel of truth or not at all. I don't want to be mean or personal, I just think it's a valuable thinking process to shed light on.

What doesn't help is tthis feeling of everyone walking on eggshells when it comes to race or women, because one is afraid to say the wrong things/think that approaching these topics is useless and overrated. Even in our EA group which is probably one of the most progressive on earth about these issues, I know people feel these things I just talked about (they told me).

 I might also be wrong and misunderstand completely what the author and his supporters think, and maybe all these people are very inclusive and care abot diversity. The focus on impact is absolutely understandable but it feels like an excuse-- it doesn't justify narrowing down and being elitist for me. Both aren't incompatible, quite the opposite. But we're not there yet. I don't know what I should do--I still stand by my comments, but I don't see the point of discussing this further, and I would not want people to think that I have stopped responding because I realize that I'm wrong but I don't have the guts to acknowledge it. I just feel very, very tired of meeting hostility when I talk about these topics in this community. 

“ As more attention and resources are allocated to long termist causes, especially AI and biosecurity, as a feature of historical disparity in development and opportunity, it becomes less likely that aspiring EAs from around the world can contribute cost effectively to the EA community’s work either directly or in a field-building capacity.”

I don’t think this is true for biosecurity. EA Biosecurity people talk about Aum Shinrikyo’s attacks in Japan all the time - Japanese EAs are much better placed to improve biosecurity and biosafety in Japan than EAs in the USA.

Also, with regards to “more attention and resources going towards AI and biosecurity”, there are 3 important ways this could change, which we should keep in mind:

1. A lot of these resources are focused on policy work. Successful policy work could divert a lot of government attention and resources to these areas, making them suddenly a lot less pressing.

2. This flow of EA resources is primarily the result of the views of a few people at Open Phil. Changes in staff at Open Phil could have a big impact on this resource allocation.

3. EA could attract a large funder or produce another very rich earning-to-giver who is less longtermist.