Epistemic status: hot takes. I have thought quite a bit about EA community building strategy over the past two years, but the style of this post is not "well reasoned and explained". I have been planning to publish something like this for months, kept pushing it back, and now I just want to get it out.
Why I am writing this
For two years now I have been a full-time community builder at EA Switzerland. In those two years I have seen very little high level public discussion about EA Community Building (EA CB) strategy. With this post I hope to spark that discussion and to add my two cents.
With "EA CB" I mean what national or city groups like EA Switzerland do, or should do.
TL;DR
General takes
"EA community building" is a fuzzy term. It is not clear what it actually means: which programmes count as "general EA CB" and which are "something else".
EA CB is an unattractive career path. Low pay, yearly grants, low social acceptance outside the bubble.
It is hard to get funding for EA CB work.
If the third wave of philanthropy is coming and "the ecosystem" is actually excited about "meta EA organisations", there is very little signalling of that towards current EA CB orgs.
Within EA CB orgs we are constantly hustling. 80/20 is the rule, always. We feel like we don't have time to think about the bigger picture. We make very little progress in finding out what the best strategy is. We go with the flow of the funders: we do what gets funded.
Strategy takes
There is no agreed upon strategy for EA CB. We should talk about it more. At least the overall direction should be clearer.
I think community building as a meta-level EA project has a limited purpose of existence, if "community building" means building a community of likeminded people across different cause areas.
I think the main task of regional meta EA organisations (currently called EA CB orgs) is to incubate the ecosystem that should exist in their region. This means:
a) Tackle the problems that can globally be tackled "best" in your region, mostly by incubating new orgs and projects. Examples through a Swiss lens: AI safety in Zurich with its technical universities, policy in Geneva, entrepreneurship in Basel, where a lot of startup infrastructure already exists.
b) Provide a productive environment for the people who choose to live in your region, no matter whether their problem could counterfactually be tackled better somewhere else. In practice this means co-working spaces and basic community events, so that people who live in one place can meet and work.
c) Find and train the few people in your region who are actually interested in "EA as a whole philosophy"[1]and have the skills to either
(i) lead orgs that tackle the known biggest problems (captains)
employing less EA-minded but talented people (rowers), or
(ii) research new EA cause areas (map makers).
Everyone without those skills is better helped by us founding more orgs that actually tackle EA problems, where they can start working as employees.
Overview of the field and the current situation
EA CB as a paid job has existed since around 2016 (as far as I know).
Most groups have been funded with 1 to 2 FTE for most of that time.
In 2026 there are maybe 40 to 50 FTE of paid EA community builders around the world, in roughly 25 groups.
The main coordinator and driver of EA CB over the past years has been CEA, mostly through the Community Building Grants (CBG) programme and directly through EAIF grants.
In early 2026 the CBG programme changed to a "grant only" programme. Grant evaluation moved to EA Funds and most other support from the Groups team ended (see this post).
In summer 2026 some EA CB groups received funding directly from Coefficient Giving through the Effective Careers RFP. This funding is explicitly not meant to support "general EA CB". The Network for EA Support and Training (NEST) was also funded directly by Coefficient Giving.
Most EA CB groups run "same same, but different" programmes. E.g. everyone has something like a career advising programme, but so far there has been very little exchange about "how do you actually do it?". Things get invented from scratch again and again.
Typical programming of an EA CB group with 1 to 2 FTE looks like this:
An intro fellowship or reading group on EA principles, once or twice a year
Intro talks and tabling at universities, plus support for student groups
1:1 career advising, career workshops, sometimes a career programme like our 🇨🇭 Impact Cohort
A yearly retreat or community weekend, sometimes a national summit
Regular socials, meetups and discussion evenings, often in several cities
A newsletter, a Slack or WhatsApp community and a public events calendar
Being the welcoming face for newcomers and connecting them to the global community (EAG applications, referrals to 80,000 Hours and so on)
Promoting effective giving, for example as a Giving What We Can pledge partner
Support for volunteer-led local groups around the country
Fundraising and grant reporting for the group itself
Assumptions
These are the main assumptions that lead me to my take that "a big part of what regional meta EA orgs do should be incubation".
Confidence levels: Very High, High, Medium, Low, Very Low.
#
Assumption
Confidence
1
Positive impact: it does not matter whether the hours are worked by "an EA"
Very High
2
Scouting, not converting
High
3
Cause-area-first outreach produces career changes far more efficiently than principles-first
Medium-High
4
The community people need to boost their productivity is small
High
5
Once a field is mature, coordination across cause areas has limited value
Low
6
Way more people apply for jobs than for funding
High
7
If you don't look into what should exist in your region, nobody will
High
1. Positive impact: it does not matter whether the hours are worked by "an EA"
The captain sets the course, the rowers row. If the course targets one of the biggest problems and the rowers are skilled and well coordinated, what they personally believe is less relevant. They don't have to be EAs.
Why I believe this: I expect that most of the positive impact in the world is done by non-EAs, and has been in the past. Simply because non-EAs are 99.9% of the population. Very few people are interested in the core idea of effective altruism, but way more people have the skills to actually improve this world, even if that is not what they are actively aiming for.
Confidence: Very High
Comments: We still need EA-minded captains who make sure the course is set right, and map makers who discover new neglected problems. The core takeaway for me: within "EA organisations" there can, will and should work plenty of people who are not core EA-aligned.
2. Scouting, not converting
By the time you meet a 25-year-old, roughly 99% of nature and nurture has happened. Whether they care beyond their own circle, think scientifically, or reason about cost-effectiveness is already settled. None of the typical EA CB programmes changes core traits. Principles-first outreach and programmes should therefore be very selective, and the target group is very, very small.
Why I believe this: Anecdotal data. I think all of us know the feeling of talking about EA with talented non-EA friends for years, and it never happens that they jump on the ideas and start acting. If being close friends with an EA and hearing about the ideas of effective altruism over years does not change people's behaviour, I highly doubt that the typical programmes EA CB orgs run have the potential to do so.
Confidence: High
Comments:
This does not mean that changing people's behaviour over the long run is impossible. But I assume it would need way bigger programmes. For example a school that teaches kids from 12 to 18 the principles of effective altruism, rationality and so on. (I am not suggesting this, it just shows what size of programme it might need.)
Failure mode: I think trying to convince or convert people is one of the biggest failure modes in EA community building. So many hours, so much energy and money invested with no outcome. I still frequently find myself energetically pitching EA to someone who has never heard of it, feel like I am moving the needle on their career path, and am then disappointed when they never come back to it and never follow up on any of my suggested next steps.
It is also damaging to have half EA-aligned people in your discussion groups and programmes. At some point they lower the quality of the whole group and might put off the most talented and promising members. A metaphor I like to use is sports. Imagine playing beach volleyball. The quality of the whole game depends on the level of the weakest player. It is way less engaging and fun for the other three, who could play at a higher level.
3. Cause-area-first outreach produces career changes far more efficiently than principles-first
Only a small fraction of the people who are willing to work on the biggest problems are interested in EA as a whole philosophy.
Why I believe this:
It connects to "Scouting, not converting".
At EA Switzerland we have done both in the past years: principles-first and cause-area-first outreach. The cause-area-first outreach (first AI safety, now biosecurity) performed way better. A fuzzy data point, but a data point.
Imagine two people on the street. One asks you: "Hey, are you interested in finding out what the biggest problems of our time are and how we can solve them? Do you want to learn about scale, neglectedness and tractability?" The other one tells you: "Hey, I want you to help me build better masks so we are prepared for the next pandemic. I offer you a 100k salary. It is a challenging engineering task, but you are skilled and I want you on my team." To me it is very likely that the second person will get much more interest and get more skilled people to actually work on the problems.
Confidence: Medium-High
Comments:
For me this simply means: we should do field building in the specific cause areas and then do cause-area-specific outreach. I assume way more people will then actually change their careers to work on the biggest problems of our time.
Again, this does not mean we should not teach EA principles at all. It means the target group that is interested in the broader EA thinking is just way smaller (see "Scouting, not converting").
4. The community people need to boost their productivity is small
Diminishing returns kick in very fast. The first other person is the most important one. More than seven are hard to handle.
Failure mode for a local CB org: trying to make everyone somewhat friends with everyone.
What to do instead: enable small, intense groups or pairs, and apart from that focus on a professional networking atmosphere.
Why I believe this:
When you are the only person in your whole social circle who is interested in existential risks, it is extremely hard to motivate yourself to start doing research on existential AI risk. My guess is that most people will not start as long as they have not found another person who gives them social proof for doing so. This one contact is the single most important one: knowing and feeling that you are not an alien for following this neglected interest.
I think the best work and ideas happen in in-depth discussions between two people or in very small groups (fewer than six). One cliché example for me is Daniel Kahneman and Amos Tversky. I think there is a causal relation between how much you think and talk about a topic and the amount of good ideas that come out of it. And the best way to increase the time you spend thinking and talking about a topic is to have another person who is as excited about it as you are, whom you get along with well and enjoy discussing with.
As soon as you have found your crowd, your core group of 3 to 6 people you can ping-pong ideas with, your interest in new contacts drops drastically. Your social needs are met. The time you spend thinking about the ideas is already high, and more contacts become more of a burden than a benefit. One data point for this: senior EAs rarely join beginner-level events. It is very hard to get senior EAs to mentor younger EAs or to participate in our EA CB programmes in any way.
The typical fellowships seem to be convinced of this as well. They bring together small groups of around six people to build close relationships, likely to increase the time people actually spend discussing the topics (and to make them stay).
When I personally compare the importance of my EA connections here in Switzerland with the feeling of being part of a global movement trying to do good, my local people are a hundred times more important to me, for my wellbeing and for my productivity. I actually still feel lonely from time to time at EA Global conferences, looking for familiar faces.
Confidence: High
Comments:
Beyond those core people it is professional networking, not "community".
"I came for the ideas and stayed for the people" is, I think, only true for the few core people. It is not about the feeling of "having a global community".
5. Once a field is mature, coordination across cause areas has limited value
Diminishing returns are hit fast. We don't need EA community building organisations that try to frequently bring together all the people from the different cause areas. (Professional networking within a cause area is important.)
Why I believe this:
See assumption 4. When your basic social needs are met and you have your genius colleague to discuss your ideas with in depth, there is value in discussing those ideas with other experts in your field. But there are fast diminishing returns in hearing about completely different cause areas and what they are doing.
When I go to an EA Global, I meet other community builders and meta EA org people. I don't see many animal welfare people chatting intensely with the mechanistic interpretability researchers from AI safety.
At our office PEAKS there is friendly chat between people working on AI safety, effective giving and EA meta. But in terms of content this brings very little forward. It is mostly social interaction.
Many of the people working on the problems are actually not EAs (see assumption 1). Those rowers are not interested in chatting about the "bigger philosophy". They care about their specific problem (e.g. "How can I build this awesome PPE?").
Confidence: Low
Comments:
Footballers and swimmers are both interested in sport. That does not make the conversation useful. Networking effects within a discipline beat networking effects across disciplines. "EA as a global community of likeminded people" is not a very useful target. It is of little use to found an organisation whose job is to build a community out of all the people doing sports. The swimmers will chat with the swimmers and the footballers with the footballers, and that is useful. If you want to learn how to organise a sports club or chat about operations, you are best helped by an organisation that offers exactly that support. (Look at Y Combinator to learn how to start a startup, don't ask the next best EA organisation.)
I think conferences like EA Global are effective because they bring together a bunch of different cause areas and enable the people within each cause area to find and talk to each other. I think the value is much less in the cause areas interacting with each other.
Sure, there is some value in interaction between cause areas, and a few people in each cause area should look into what and how other cause areas are doing. But this should not be a big goal of an EA CB org.
Failure mode: an EA CB org organises networking events and pushes people who work at EA-aligned organisations to attend and network with each other. Those people have very little interest in joining. Their social needs are met, they don't see the point in connecting with people from other cause areas, and many of them are rowers who are not interested in the overarching topic of effective altruism.
Sorry - that was repetitive... I think you got the point.
6. Way more people apply for jobs than for funding
That is why you, as an EA community builder, should apply for funding that incubates new projects, which others can then directly apply to as a job.
Why I believe this: All the projects we have launched at EA Switzerland in the past years point in this direction. Nobody started those projects by applying for funding themselves. It needed us to apply for funding first and to open a hiring process. Once we did, more than 30 people applied every time.
Confidence: High
Comments: Failure mode: not applying for funding.
7. If you don't look into what should exist in your region, nobody will
It is very unlikely that someone else in your region will look in depth into what could counterfactually be worth founding there. If you as an EA CB org don't do it, very likely nobody will.
Why I believe this: All the projects that do not yet exist but could exist in your region.
Confidence: High
What EA CB orgs (regional meta EA orgs) should do at the moment
Three things:
Tackle the problems that can globally be tackled "best" in your region, mostly by incubating new orgs and projects.
Provide a productive environment for the people who choose to live in your region, no matter whether their problem could counterfactually be tackled better in another location.
Build a "captains school": find and train the few people in your region who are actually interested in EA as a whole philosophy and are capable of leading rowers or researching new cause areas.
Let's start with 2 and 3, as they are short.
2. Provide a productive environment, aka a co-working space
Providing a productive environment for the people who choose to live in your region (no matter whether their problem could be tackled better elsewhere) basically means building a co-working space where those people find the setup they need for their work. Not everyone needs such a space to be productive. But many people prefer not to work from home. They like having a place to go to and the social interaction with colleagues, even if those colleagues work for other orgs.
How to set up and run a co-working space is out of scope for this post. My strategy take here is: Be a hub, become a hub - do not spend much on keeping local groups alive by force.
I think it is a very common story: EA local groups thrive when there are great organisers, and when those organisers leave, the group goes inactive again. There are a few cities with ongoing thriving EA communities. Those are hubs like London, Oxford or the Bay Area. I think this will not easily change. The most engaged members of a local group will, at some point, when they want to do high-impact full-time work, move to a hub. So if you want to do successful community building within your country in the most cost-effective way, I think you should look at which cities in your country could become a hub, meaning a city where people want to move to in order to do high-impact work. In Switzerland this is Zurich, and at the moment we are focusing on making Zurich a hub.
Becoming a hub usually means having an in-person co-working space (PEAKS) where everyone can meet at one central point. It usually means a certain density of projects and people who already work on cool stuff. So we try to launch new initiatives and to bring the people who are already around into the co-working space.
On the other hand, we are not investing a lot of resources into other local groups, for example Basel, Luzern or Bern. We support exciting organisers when they appear. But since we expect those people to leave at some point, and since it takes a lot of effort to keep a city alive that has no hub potential, we save those resources. This might change once you have more FTE and don't have to focus only on the most promising city in your country. But as long as there is no funding for it, I think it is just too resource-intensive to maintain volunteer-led EA groups in cities that don't have, for example, a high concentration of talent.
So the failure mode is: investing too many resources into keeping barely active local groups alive, and thereby neglecting to build a productive working environment for the people who are there anyway, aka an "EA hub".
3. Build a "captains school"
I have not yet thought a lot about what this captains school might look like. But I assume it will be a way smaller programme than most of the typical, broad outreach EA CB programmes where "everyone is welcome". It will include paying attention to the fact that only people of a similarly high level come to your metaphorical beach volleyball training.
A failure mode I see at the moment: with 1 to 3 FTE, groups do too much principles-first outreach for a broad audience. First teach them the EA principles, then try to encourage them to change careers into one of the high-impact cause areas. And invite or accept too many "not so good beach volleyball players" to the training. This means neglecting the incubation of new fields, and with it the cause-area-first approach.
1. What incubating new projects looks like
(These are rough thoughts, developed back and forth with Claude. I hope my core points come through, but as I said: better look at Y Combinator and others for better advice.)
The model I use for this has six levels. A project is anything you try to get off the ground: a field, an organisation, a policy change, a research group. Your organisation does not sit on the ladder itself. Its job is to start projects and carry them up the levels until they no longer need you. That job is different at every level.
Level
The question
How you know you are there
Your job as the meta org
Rough cost (USD)
L0
Is it worth trying here?
A short written assessment of whether your region has an advantage for this problem: relevant talent, universities, industry, access to government. The output is a ranked list of options and the cheapest test for the top one.
Write the assessment or commission it. Nobody else is looking across cause areas for your region. If you don't do it, it does not happen.
under 5k
L1
Is there real potential?
Someone outside your organisation invests their own time or money without you paying for it. A professor spends a weekend on it. A ministry agrees to a second meeting. The same people come back to a third event without being reminded.
Provide the room, the mailing list, the introductions and a small budget. Run the test. Do not build structure yet.
5k to 30k
L2
Does it work with a paid person?
Someone is paid to work on it. One programme has run from start to finish, or a prototype exists and people use it. This is the first level where the project has somebody's full attention.
Secure funding, find the right person and hire them. Provide fiscal sponsorship, payroll and an office, and take on the parts of the job the new hire should not be doing.
100k to 250k
L3
Is it a functioning organisation?
Programmes repeat with similar results. Two to five paid staff. A funder has renewed rather than started. Someone other than the founder could run the next round.
Step back from delivery. Introduce the project to funders directly instead of passing money through. Help it build a board and push it to take over its own operations.
300k to 1M
L4
Does it work without the organisation that started it?
Own funding, own board, no longer fiscally sponsored by you. If your organisation stopped tomorrow, this one would continue. The money can still come from EA funders. What matters is that the project holds those relationships itself.
Let go, with a date and a written handover that says what you will and will not keep doing.
own funding
L5
Is the problem smaller?
The project has achieved what it realistically can, and more money would buy little extra.
Nothing. This happens out in the world, not in your portfolio.
not a budget
Mistakes I have seen (and made) at each level:
L0: Skipping it. Most projects I know of started because another country did it, or because a funder had a budget for it. Not because anyone checked whether this region has an advantage.
L1: Reading my own enthusiasm, or that of the people already around me, as a signal. Twenty people at a free event with catering tell me very little.
L2: Never applying for funding. The cheap tests showed interest and potential, and the project just carries on as a volunteer activity for years. Nobody ever decides against professionalising it, it simply never gets put on that track. Also common: hiring the most enthusiastic volunteer, who might not be the right fit.
L3: Mistaking founder effort for organisational capacity. Simple test: if the founder takes two months off and the next programme cycle slips, the project is still at L2. Running a small organisation reliably takes at least three different skill sets (operations, strategy, delivering the programmes). It's not easy!
L4: Calling it independent because the legal entity is separate, while we still run its payroll, its fundraising and its operations. Or: never really letting go, so the project keeps us from incubating the next one.
Two Swiss examples of what L4 looks like: The Simon Institute for Longterm Governance spent a few years as a working group at EA Geneva before it spun out as an independent Geneva think tank. Pivotal Research began as the Swiss Existential Risk Initiative (CHERI), was hosted and fiscally sponsored by EA Switzerland, and now runs as its own organisation with a fellowship based in London. Most national EA groups running on a community building grant, EA Switzerland among them, are themselves at L3.
The most useful thing this model did for us at EA Switzerland: we sorted our own projects into two columns, meant to leave and meant to stay. My take is that every regional meta org should do this once, ideally before a project gets its first budget, because it changes how the project gets set up. If a project is meant to leave, it gets its own name rather than running as "our X programme", we introduce it to funders directly, its budget is a separate line that could be lifted out, we think early about who could sit on its board, and everyone involved knows that our involvement has an end. If it is meant to stay, it runs under our name and our grant, there is no separate legal or financial setup planned, and the capacity it needs belongs in our baseline budget.
My expectation for a regional meta org in a region where the fields do not exist yet: most of the portfolio sits at L0 and L1, and the real output is a small number of L2 attempts per year. Once the fields are established, I think the portfolio should be largely empty of projects. What remains is the standing work: the research on what else should exist, the captains school and the hub.
Doubts
The biggest open question in my own model: where do the captains come from?
The incubation model depends on captains. Every new project needs an EA-minded person who takes the first paid role, and we can only get more rowers on board as long as good enough candidates apply for the director roles. At the same time I argue for less principles-first outreach and fewer "everyone is welcome" programmes (while having 1-3 FTE). So there is a tension between "less broad outreach" and "we need more captains". Without having had a deeper look into this, I expect captains to hang around in specific places. You need an entrepreneurial skill set combined with an EA mindset. Scouting incubator programmes and university startup ecosystems seems to be a solid first target group (compared to broad university outreach).
Another doubt might be that EA community builders don't have the skill set that is needed for incubation. That they are not entrepreneurial themselves and are much stronger in social interaction (for example, one-on-one career advising). I personally know many of the paid EA community builders around the world and I think most of us are generalists with a strong 80/20 mindset. We are starting, running and evaluating new programmes all the time. Whether it is a career advising programme or a new biosecurity field-building organisation is, from my perspective, not the biggest difference.
Summary: what I want to see less of, and what I want to see more of
Assuming the funding level for regional meta EA orgs stays around 1 to 3 FTE.
Less:
Less support for barely active local groups.
Less converting, more scouting. Find the people with whom EA ideas resonate immediately.
Fewer "EA is for everyone and everyone is welcome" programmes. It is totally okay that most people don't want to play beach volleyball and don't show up in the first place. It is also okay that there are different levels, and that you only join the training with people at your level.
This might be a straw man point - as this might not really be the case. But I still feel it in myself. It is socially awkward to "reject" and "exclude" people. I assume many EA CBs have the same characteristics as I do and don't like to reject people.
More:
Research the things that should exist in your region. What does the ideal state look like?
Make this ideal state happen. This mostly means incubate, which again means: do cheap tests (is there interest?), get the first FTE of funding for the project (nobody else will, so apply for more grants!), find the first talent (run the hiring; the person leading the new project should be a "captain"), be the operational backbone (fiscal sponsorship), and secure value alignment for the new project (become a board member). Let the project spin out if it is successful, or end it again if the theory of change does not work out.
Get more rowers on board, as long as you have enough captains. That is: as long as good enough candidates apply for the director roles of your projects.
Share your best versions of "captains schools", so others can copy them. (I have not thought about this yet.)
Create an EA hub in your region, very likely in the place where the most EA-aligned people hang around anyway. Build a co-working space so those people can be productive.
I think regional meta EA organisations should get more funding, with the specific mandate to incubate the ecosystem in their region (meaning: incubate the field building of EA cause areas).
Thanks
Thanks to everyone who commented on my earlier "hot takes" documents: Alix Pham, Bridget Loughhead, Capucine Griot, Carole Bibas-Barkan, Gergő Gáspár, Grace Adams, Guillaume Vorreux, James Herbert, Loïc Watine, Marcel Steimke, Marius Wenk, Maximilian Onkenhout, Milena Canzler, Niels Doehring, Sarah Bluhm, Tom Hutton and Ysaline Bourgine de Meder.
These are completely my personal (hot) takes. Even though others commented on earlier versions, this does not necessarily reflect their positions.
I am looking forward to more discussion about the strategy and future of EA CB.
By "EA as a whole philosophy" I mean being interested in the core question itself: how can we do the most good with the limited resources we have? People like this want to start from a blank page rather than from a conclusion. They enjoy the toolbox that comes with the question: scout mindset, scope sensitivity, cause neutrality, thinking in cost-effectiveness. And they are willing to change what they work on if the answer changes. Most people who are willing to work hard on one specific big problem are not interested in this question. That is fine, and it is not what I mean here.↩︎
A preliminary estimate, and a request for better ones.
Summary
I believe the standard literature estimates for the number of DALYs attributable to a case of stunting are too low, largely because they don’t account for the long term effects. This means that childhood nutritional interventions that reduce the prevalence of stunting may be substantially more cost-effective than previously believed.
Epistemic status
Exploratory and back-o...
I’ve been feeling pretty shaken since the METR report about the Hugging Face incident came out last week. Over the weekend, I wrote up some thoughts on how lonely the AI situation sometimes feels to me. It’s more personal than what I’d usually share publicly, but I thought I’d post it here in case it resonates with anyone else.
I’m very grateful to the man...
TL;DR: Kairos has raised $50 million from Coefficient Giving for two years of funding, one of the largest commitments they’ve made towards AI safety fieldbuilding to date. We’re using this to make an ambitious push for growing Kairos, broadening our portfolio of talent infrastructure projects and incubating new organizations. We’ve doubled in size in the last six mon...
