And Other Recommendations for EAGs Based on The Art of The Gathering

The Motivation: I read The Art of the Gathering awhile ago, and thought it had a lot of interesting insights for greater community building efforts. I realized that EAGs^[1] might be a ripe place to try to apply some of these insights, and so I set out writing this piece. What was meant to be 10 similar-in-length suggestions naturally turned into a piece focused on the purpose of EAGs, but I hope you’ll still read over the other points too in the original spirit of the post, I think they have something to offer too.

Epistemic Status: All the principles themselves have strong grounding in the book and are things I feel quite confident Parker would endorse. I’ll make concrete suggestions below for each, and also venture a long commentary on the potential purpose of EAGs, which are both attempts at applications of these principles and are more speculative and should be taken as “Tristan’s best guess”.

A brief note: CEA and individual EAGx organizers are already doing many of these things to some degree, and as such I hope to not only introduce new ideas or frames, but also highlight and reinforce some of the great things they’re already doing.

1. EAGs should have a clear purpose.

I think this is the most important point, so I’ll spend a great bit of time on it. To put it another way, every gathering needs a greater why that justifies the what. I don’t think this is spelled out well for EAGs and has been the cause of some confusion, and I think it would be good to decide on, and then state, a clear why for EAGs. Below I’ll mention four different whys I think are at play, and what prioritizing them (more) would mean.

For Learning:

Talks are consistently an important part of EAGs, and some have argued that they’ve been underrated^[2] in the process of promoting 1-on-1s. Significant staff time is spent organizing the content, giving indication that even if 1-on-1s are considered to be more valuable, there’s got to be something worthwhile about them.

I think the ToC here is that the talks afford a particularly good opportunity for others to further learn about a topic they’re interested in but maybe haven’t fully had the chance to explore yet, something likely to be especially valuable for those newer to the EA space.

But the talks do actually afford more than this^[3]: they are also a sort of podium by which CEA can try to diffuse ideas throughout the community, whether that be promoting a new potential cause area, a better recognition of needs in the EA space, or desirable cultural norms. Focusing the content in 2017 on operations, for example, allowed the community to address a significant gap at the time. They’re also a way to expand impact outside the conference when recorded, which could be important as CEA begins to focus more on EA branding and outwards content.

Either way, I think many people only have a vague sense of the purpose Learning currently serves, and I think we’re even further behind on figuring out how to measure the effectiveness of Learning specifically vs the general impact.

Implications:

Focus more content on connection. If talks aren’t just informational, but also are used to mold the norms of the community both in and outside the conference, maybe more talks should be had in service of helping people make better/more connections. Yes, there’s already some of this in the first timers meetup, and motivation in the introductory talks, but I still think far more content should go towards this end given 1-on-1s seems to be the greatest source of value.
Be more explicit about the purpose of content at a given EAG. Perhaps there could be some short statement placed in the attendee guide or something like that that quickly explains “here’s what we’re focusing on this EAG and why”. Or perhaps the event descriptions could include more of the motivation behind having that specific talk.
More clearly emphasize EAG as a learning opportunity and a place where the presentations are also quite valuable.
Possibly restructuring and adding in other types of content, like those at academic conferences. If there’s an important new paper out I’d love to have the writer come and defend it, perhaps with a panel or if not by responding to questions from the crowd.
Do more educational talks for EAGx and less for EAG. This already seems to somewhat be the case, but traditional talks are likely most helpful for those who are new to orient them to the space, meaning EAGx is likely a better venue to be talk heavy.

Reasons for:

Amy: Talks can help model desirable norms, e.g. reasoning transparency, which those in the audience can then better internalize and carry into the rest of the weekend (and perhaps even outside the conference).
Amy: Talks can give a platform to those from underrepresented groups at the conference, helping to make sure others don’t feel alienated based on their status.
Amy: Talks can create a motivating atmosphere where seeing exciting new things being done can help motivate you to do something similar.
Miller: Talks create a common base of experience for others to build off of when making connections.
Miller: Talks create a shared emotional experience that is more likely to cause the ideas to sink in deeper.
Miller: Talks can reinforce your ethical and intellectual identity and prove more motivational than just consuming the content alone at your laptop would be.

Reasons against:

The events team at CEA has tried to emphasize over the years that 1-on-1s are the most valuable piece of EAGs, and I’ve found this to hold true for myself. There’s also a difference here between types, where workshops afford a chance to talk and dig deeper with a group around you and better blend connection and learning, talks perhaps the most solitary thing you can do at an EAG.
CEA is trying to build a brand, and has a recipe for EAGs. One conversation I had indicated that there’s a pressure on organizers to keep to a similar sort of template as has been used before, as the risk of running a low-rated conference is greater than the expected gains of experimenting, given ratings for EAGs are already fairly high. If that’s the case, I wouldn’t be surprised if there were organizers who’d like to have had less talks in the past, but didn’t do so because of fear of straying from the successful EAG formula.
1. Against: Ollie notes that experimentation is encouraged at the margin, and that there’s plenty room to experiment while sticking to the known model.

For Connection:

A measure of success often referenced for EAGs is something like “new connections per person” where a new connection is meeting “someone you feel comfortable asking for a favor^[4]” at the event. On top of this, there’s a pervasive sense that 1-on-1s should be the focal point, one organizer saying 1-on-1s are “consistently reported to be the most valuable experience for attendees”. Connections seem to normally refer to two different types of connections, a more professional type that alludes to a sense of networking, and a more friendly type, the metric of being able to ask them a favor treading the line between the two.

Here, I’m splitting those into two different categories, and taking connection to mean something closer to networking, covering friendship below. Networking itself doesn’t even fully cover it, as only part of the impact likely comes from networking in the traditional sense of increasing one’s points of contact in a given field so they’re better able to complete the job search. For example, another aspect of connection seems to be giving a chance for those already doing direct work to come together and use their conversation as a sharpening stone for increasing their impact. But perhaps the most important aspect for me personally has been more general career guidance others have been able to give, indicating there’s maybe another mentor-mentee angle to connection that is most impactful here.

Each of the above views on what type of connection is most impactful will guide us in different directions when deciding what to focus on in the realm of connection. Imagine a meetup for AIS, for example, where you now wish to split people into groups for discussion. If we are focusing on traditional networking, we’ll likely try to pair people up based on those they haven’t met yet. If we’re focused on giving space for those doing direct work to meet and bounce ideas off one-another, we’ll likely pair people up by specific workstreams. If we instead want to focus on fostering the mentor-mentee relationship, we’d likely match people based on years of experience. There are likely other frames of focus too, so defining them and then figuring out which frames are most needed for connection at a given conference, and how they should be implemented, are likely the next best steps.

Implications:

Do more meetups. There are always more lectures than meetups, and meetups aren’t normally optimized for networking or sharpening ideas between people. If they were, we would see more things like “researcher meetup” or “fundraising meetup” and we would see these at an increased frequency (instead of just one opportunity that if missed means we totally miss the opportunity).
1. Against: Ollie mentions that meetups have higher variance in terms of value, where poorly executed ones can be particularly alienating.
  1. Response: It’s hard to know what exactly makes a meetup go poorly, but I don’t think that meetups should generally have higher variance than other forms of connection (and I at least haven’t experienced them that way). I do think EAs may be slightly more socially awkward on the margin, which can make running successful meetups harder, but this seems overcomeable with experimentation. Perhaps having smaller group discussion is best. Perhaps more prompting from the meetup organizer would be good, e.g. giving people further guidance on what to talk about (I had a speed friending session recently where each new match came with a new question shown on the projector, and I think this seemed to be a fairly good addition).
  2. I’ve also assumed thus far that the improvements should come from better helping or training those currently giving meetups, but perhaps the profile of who gives the meetups should be changed. Encouraging meaningful discussion between participants is an art of its own, so perhaps meetups should be led by those who have spent more time organizing, rather than someone with a particularly deep experience in the given subject. There are many community builders who come to these events, and they would arguably be better positioned to lead a “AI Governance meetup” than someone with deep experience in the field who hasn’t done much organizing. With those people, CEA’s role in helping them out would be more focused on helping them identify useful questions specific to the area, which seems perhaps easier to teach quickly than how to organize a good meetup.
Encourage networking with others in mind. Ask people to spend more time thinking of how they can connect those they are talking to to others in their network, rather than focusing narrowly on what they have to immediately offer one another in a 1-on-1.^[5]
Consider further attempts at encouraging mentorship. CEA used to run both a Stewardship program, where mentors were paired with specific mentees chosen by the CEA team, and a Guides program, where group organizers and newcomers would meet together more informally. These both sound great to me, and I’m excited that they’re planning to revisit these.
Change the way impact is measured. I don’t have anything much better to offer, but even “someone you feel comfortable asking for a career related favor” would be more accurate here (asking about “impactful connections” directly might also work better). CEA also does follow-up surveys, which provides an opportunity to ask some questions like “What actions have you taken as a result of connections you made at the conference that you wouldn’t have otherwise?” which I think might present the best data on the effectiveness of connections. As such, I think it might be worth monetarily incentivising those followup surveys.

Reasons for:

This is what participants reflectively value: This survey (mostly of EAGx attendees) showed that “new connections” are the primary source of value (44% mentioned), followed by “learning about new career options” (38% mentioned), and then “learning about EA cause areas” (30% mentioned)^[6].

Reasons against:

The above data for this is messy and could be mostly focused on friendship. When I answer the question as it’s currently phrased, favor doesn’t register as a work related favor but something more personal, a metric for some deeper connection that implies a dependability on that person that often leads me to give fairly low responses.

For Friendship:

Another sense is connections as friends, where the value comes from diffuse and harder to measure metrics that blur between a distant proxy value for direct impact and just generally improving the community as an intrinsic good that’s valued in-and-of itself. No one has explicitly argued for this as a reason for EAGs as far as I’m aware, likely because doing so would feel like an ill-use of funds, charitable dollars going towards something that enriched us, rather than the greater world.

I don’t have the space here to properly motivate this, but I do think that friendship is a justifiable end for these conferences, at least worthy of greater focus than it currently gets. In short, my main grounding for this is thinking that the fact the EA community exists is a strong force for good in the world, and that we likely will become only a hollow sense of community, one that isn’t even deserving of the name community, if we only focus on fostering professional connections. Communities are made up of friendships, and without those underlying friendships I doubt EA will be a lasting institution.

But if that doesn’t convince you, let me give you a brief pitch that’s a bit more direct. Robin Hanson has argued that the majority of the influence achieved by the 60s counterculture movement came not in the early moments of the movement, but later on when their careers progressed and put them into positions with more influence. What allowed this to happen was building a strong sense of identity around their community, forming friendships with the movement and its values at the core, changing people at a deeper level than is possible with a “community” built around mere professional affiliation.

Implications:

Running significantly more speed friending events (assuming they actually encourage friendship).
Focus more on those with few to no friends in the community. Maybe have a corner that is marked as “looking for someone to have a spontaneous 1-1 or conversation with”. There was something like this at EAG London, though I’m not sure how successful it was. Alternatively, maybe each person should be randomly assigned a group at the beginning, with an hour-long event at the beginning to get to know them. This way each person would have a friendly set of faces going into the weekend, and there could even be tables specifically marked out for them, so they have a sort of shared home base to return to throughout the conference.
Run more events that aren’t directly impact focused. Meditation is a staple of EAGs and a common example of how this can go well. “Anything but EA” hour^[7] at EAGxBerkeley was an even more straightforward example of this, and I think we can come up with many more similar events if we focus on this form of meetup.
Focus more on events outside of the conference itself and making them go well. EAGs “carry momentum” and are often an entire experience for people that is incomplete without outside social activities like exploring the city or going out to eat when dinner isn’t served.
Focus more on how to sustain friendships once the conference is over. Right now it’s sort of “connect while you’re here!” when it should be something like “connect while you’re here and then keep up with close connections!”. One recent meetup I went to involved finding an online co-worker or accountability partner, which seems like something heading in the right direction.
Focus less on breadth and more on depth. The amount of close friends we can actively keep up at any given time is debated, but it’s at least clear that those holding 20 1-on-1s are not going to convert all their interlocutors into close friendships. As a random anecdote, I made the mistake at EAG London of booking 1-on-1s around dinner time after the venue closed, which interrupted the flow of continuing with whoever you might be talking to as the venue closes to dinner plans. The conversations were still good, but I think it actually would have been better for me to continue with others into dinner, and don’t plan to book meetings after 19:00 in the future.

Reasons For:

EAG is one of the few places where EA feels like an actual community. I think this is especially true for those who don’t live near a hub or thriving city/university group. If EAGs stopped today, I expect that many connections across different areas of the community would likely falter, and that EA would likely only limp on as a community, with more and more people choosing to focus on their individual cause areas and building up connections within them year by year. To be clear, I don’t think friendship needs to (or should) be the singular why for EAGs, but I think there are ways it could be encouraged without detriment to other whys that seem high quality to me (e.g. finding ways to encourage continued connection after an EAG).

Reasons against

EAG is funded by charitable money, and impact is our focus. Without connecting back to impact, something that would be very hard to do with this focus, we’ve lost the connection to our original aim undertaken by this community. EAGs are for improving the world, not our own happiness.

For Impact:

For the current team organizing EAGs, this is clearly what they’re optimizing for. As said by Eli Nathan: “EAG exists to make the world a better place, rather than serve the EA community or make EAs happy”. Most of what I’ve read to understand organizer’s views here has expressed a desire for this to be the output, where the impediment is just trouble measuring it, proxies like connections and what participants valued serving as imperfect temporary measurements that will hopefully soon be replaced by more concrete measures of impact as a result^[8]. Attempts to create such a metric seem to still be in development, but this thread from Eli Nathan indicates the best way to put it for EAGs might be “impact-adjusted plan changes per $”.

This also holds resonance with conversations I’ve had with more senior members of the community who see showing up as some sort of community service, as part of their impact portfolio done only so long as they see it as an effective way to spend their time.

Implications:

We should talk a lot more about the cost effectiveness of EAGs. I (and [many] [others]) find the cost of them to be crazy high (especially for EAGs, EAGxs are somewhat more reasonable) and think that, so long as this is our frame, we should have a much better metric by now to compare at least to other community building efforts, if not to the wider set of funding opportunities.
We should do fewer EAGs and prioritize funding more local groups. I think the limited data we have points in the direction that EAGs are significantly less cost effective than local groups. There was a survey of people doing longtermist work run by Open Phil in 2020 that asked respondents to rate various EA organizations/content/groups on a scale from “hindered my positive impact a lot” to “helped me a lot to have a positive impact”. They assigned point values and averaged responses, then calculating “net impact points” to create a standard metric to use to compare across and displaying a “% of net impact points” both weighted^[9] and not weighted, which should give a sense of what proportion of impact each item was responsible for. As you can see in the chart here, CEA makes up 6% of these, and elsewhere it’s said that EAGs are responsible for roughly half of CEA’s impact, so assume EAGs are roughly 3% of net impact points (NIP). This means that EAGs are less valuable than everything else in the chart, and that local groups are responsible for over 2x the % of NIP. We only have rough data right now, but one estimate puts total expenditure on EA groups at ~$1 million in 2019^[10]. In that same year, there were 3 EAGxs and 2 EAGs, which, taking estimates from the earlier post on costs,^[11] likely cost around $5.6 million, over 5x the cost.
1. Against: There is a lot of funding available for community building so the pool of resources available might be fully sufficient to do both.
2. Against: Even if funding were constrained, the bottleneck to further groups might not be more funding but rather something else, e.g. more quality community builders.
3. Against: The survey didn’t capture a number of other useful things being done in the community building space, whether from the people being surveyed being generally more experienced (and thus underweighting those efforts targeted at newcomers), or by just missing less concrete things like courses offered or offices for co-working.
4. Against: EAGs could be the second most impactful thing in the community building space, where other activities, e.g. co-working spaces, may be the things we should do less of.
5. Against: The above looked at local groups generally, and not just “local groups that CEA funds”, meaning that the comparison of how effective CEA funded local groups vs EAGs might look different, i.e. it could be even more effective, OR CEA could be unable to as effectively reach the most effective segments with its funding and thus should focus on EAGs.
  1. I’m somewhat skeptical here, but this is perhaps because I think there’s likely currently too high of a bar for vetting people to run groups, meaning I think that there are likely a number of people out there that could be running good-enough groups that we’re currently not enabling. I also think that there is significant value to funding multiple people to run groups, as I think group failures likely increase significantly when only a single person is running it. I’m not sure to what extent we should take this, but given we likely spent over $10 million on EAGs in 2023, I think it’s reasonable that more CEA funding could go towards groups.
We should consider whether having cause-specific conferences is better. One major assumption underlying EAGs is that these are best executed when we can bring together a number of different cause areas under the same roof, for cross-pollination of ideas. I actually do think this is good overall, but more for the friendship purpose mentioned above, to build an identity around EA, and am more skeptical if this is actually worthwhile purely in terms of impact.

Reasons for:

The obvious from before: these efforts are funded by charitable dollars, and we’re supposed to be doing the most good possible, which is defined directly by the impact we can have, thus indicating it as the metric we should be optimizing.
Linch thinks that aiming for community happiness is likely a worse “proxy goal” than aiming for impact directly and that “ultimately we’re here to reduce existential risk or end global poverty or stop factory farming or other important work. Not primarily to make each other happy, especially during work hours”
Nuno (in negative form) thinks that what we care about is not connections but whether they “lead somewhere” or not (and at least one organizer agrees)
Evie: EAGs are only useful in-so-far as they lead community members to better work in the real world.

Reasons against:

Tying into the positive I mentioned for friendship as a why, I think EAGs currently sustain the community at large in an important way, and I think further optimizing of the conference to become e.g. more networky would be a loss for the community and likely contribute to it becoming more hollow over time. For those not in hubs, not in EA jobs, where will they go to experience EA as a community?
We don’t have a good metric for measuring this currently not for want of trying, but because it’s hard to develop, and any attempt will likely neglect a current intuitive sense I think many have that EAGs are impactful and important, despite being easily quantifiable.

2. Don’t be a chill host

Priya talks at length about our perception that the “chill host”, the one who’s laidback and more often to say yes than no, is only gracious on the surface, and actually doing a disservice to the event by shying away from their nay-saying duties. This begins with the invitation, or the admission process, where we have a strong prior to not reject anyone because we don’t want to be rude. But if we’re planning an event with a defined purpose, some people will fall out of that purpose, or will impede others trying to focus on that purpose, and by saying yes and letting those people in, we’ve betrayed that initial purpose and those who we’ve invited expecting it.

Different purposes above will lead to different decisions as to whether or not one should “open EAG”. I think part of the disagreement over the admissions process and who to admit has come from disguised disagreements on the why, and I think a fair bit of clarity on who we should admit will flow from deciding that. But even then, most of Scott’s post focuses on arguing for the Impact Why, so if this is the route we choose, we will likely have much discussion to go through to determine what sort of admissions process leads to the highest degree of impact.

This practice doesn’t stop with the admission process though, Priya thinks organizers should carry this perspective into the event too. Organizers have the opportunity to set rules for the event, to dictate what their attendees can and can’t do, and similarly to the above, when done in service of the why of the event, these rules can be gracious in the end despite seeming draconian on the surface^[12]. The important bit here is making sure that the restrictions connect to the why, but I think that the CEA Events team is often putting much thought into the why and how the content connects to that, and I’d trust them to make more rules during the event that enhance achieving the why.

Concrete Suggestions

Require people to spend at least five hours preparing for attendance. In nearly all the potential purposes at play here, conferences would be better if people came more prepared. Every moment is subsidized by charitable money, and I don’t think all attendees have internalized that, especially when I see non-filled out Swapcard profiles. Five hours is somewhat random, but I think every (non-Senior) attendee should be spending at least a couple of hours preparing their swapcard and looking for others to connect with, as well as a further couple of hours thinking about what they want to give and get out of the conference.
Set norms for conversations with senior EAs. I don’t know what the right norms are here, and it’s certainly a tricky balance when some people are overly timid in reaching out because you don’t want to add any further burden to them. But at least some people seem to be reaching out too much, as multiple conversations I’ve had with more senior EAs have mentioned repetitive conversations about basics or unprepared conversations where the new EA comes without a clear sense of what they want to talk about. Some senior EAs have mentioned feeling burnt out as a result, which I think is a real shame, and something we should try to avoid if possible. One way to do that may be requiring that those booking a meeting with senior EAs have a theory of change for the meeting, or at least some clear focus that poses questions that couldn’t be answered by five minute google searches.

3. Be clear about the admission bar:

Despite any changes or thoughts on the above, it’s important that organizers communicate their purpose so people understand why they’re not admitted.

Right now, this is a mess. After spending ~5 hours reading forum posts and CEA material on EAGs, I still don’t have a clear idea of what the admission criteria, their “bar”, is and what they’re aiming for. The FAQ articulates things a bit further, but only to say: “EA Global is mostly aimed at people who have a solid understanding of the core ideas of EA and who are taking significant actions based on those ideas, [many]… professionally working on effective-altruism-inspired projects or figuring out how best to work on such projects”.

But what counts as a project? “Professionally” implies that I should be doing EA work as my day job to count, but then the addendum of “figuring out” broadens this to just about any single person who would be interested in attending an EA conference. What really is the line between “figuring out how to best work on such projects” and “interest in learning more about what to do with [EA] ideas” from the EAGx admissions criteria? Sure, EAG is aimed at people with a “solid understanding”, but that’s vague too, and even that vague sentiment is qualified with a “mostly aimed” before it, indicating that this is just one focus of EAGs, amongst others. AND even the entire sentiment is contradicted elsewhere in the FAQ. In response to the question “What if I’m early in my career or still figuring out what I want to do?” CEA says: “You don’t need to have a long impactful career behind you or a polished plan for impact to be admitted to an EA Global or EAGx conference. Some of the strongest applications are from people who are too early in their careers to have taken many ‘significant actions.’” And there are tidbits of other factors hidden elsewhere, e.g. a record of overconfidence or excessive self-promotion can be motivation for rejection.

I do think that this bar has likely changed significantly over time, and where it currently stands is likely much more in line with what people expect than it once was, given that it seems restrictions used to be greater in response to capacity limits that are not still the case today. I think CEA didn’t do the best job communicating this in the past^[13] and probably caused some unnecessary alienation from people surprised to be rejected. But my hope is that, though there are still certainly still difficult situations, there are less over time, and we’re hopefully moving towards a world with a better understanding of what CEA’s bar is^[14], hopefully a bar that aligns with what we expect as a community for such events, with its purpose.

For now, I think CEA would say that the line for the event is their assessment of a good fit. But what is a good fit? Someone who can help others achieve their EA related goals. What does that person look like? I’m not sure, and answering that question further in depth and making it clear is, I think, an important next step for CEA’s events team.

Concrete Suggestions:

Make admissions policies clear. If this EAG is only for people in Eastern Europe, say that. If this EAG is only for people that have worked directly in EA, say that. The goal should be as few rejections as possible, because 99% of people should be able to know whether they fit the mold or not. Alternatively, you can address this on the other end by providing individualized, clear rejections. But this seems like a longer process that there likely isn’t capacity for, something that could be better achieved by proactively doing the above.
Make sure the application process reflects the underlying why. If the why changes from event to event, e.g. if learning is more important for EAGx, make sure that’s reflected in the application process e.g. maybe those with less knowledge should be given priority.

4. Make the introduction and conclusion priorities:

Priya stresses that the introductions and closings of a gathering are incredibly important, and that they are likely two of the most consequential moments of the event. The goal of the opening should be a form of “pleasant shock therapy” where your goal is to simultaneously instill in guests a sense of welcome and a sense of being grateful to be there (178)^[15]. A good closure involves first looking inward, contemplating what you learned or experienced there today, and then looking outward, thinking about how you will incorporate that into your outside life (259). I think both of these principles are ripe for application at EAGs, and have found myself generally disappointed with introductions, closings as more of a mixed bag.

Concrete suggestions:

Do more sessions which connect to the emotive aspect of what it means to be an EA. I have to brag on the EAG Global 2024 team here, they really blew me away. They invited two living survivors of the bombing of Japan to speak on their experiences and say a bit about what they hope the world will learn from them. I had chill bumps for most of their talk, and honestly could not imagine a better closer to an EAG. All closers should aspire to be as inspiring.
Never, I mean never, start or close with logistics. This is one of Priya’s axioms, and I think EA would do well to heed it, if we want to have a more meaningful way to start and cap our conference experiences.
Don’t cross scheduling events over the introduction and conclusion, and restrict the ability to schedule 1-on-1s at the same time. This is a big part of the problem and signals that these aren’t events everyone should come to.
Frame the introduction and conclusion as “highly encouraged” or do something to indicate that they are quite important. Right now they are just another event on the schedule, and reminders to attend are great but don’t seem to be generally working, as I’ve yet to see an introduction get above half the conference to attend. To build the energy, and to really induct people into the weekend, we need people to show, and the best way to do that is to change how the intro and conclusion are framed and make them something people want to show up to.
In the intro, have people greet each other momentarily, like when a preacher asks the congregation to “turn and say hello” to your neighbor (183). Plenty of people show up knowing no one, and though 1-on-1s can rectify that, it would be nice if everyone had a friendly face going in, so that it’s not a room, or conference, full of strangers.
Ask people to wear their EA shirts to the concluding talk and message this as emblematic of them carrying who they were and what they learned “here, this weekend” back out into the world, into who they usually are.

5. Set norms and be explicit about them:

By setting a norm for something, you give people the excuse to do so, and help guide how people are acting that weekend. I’d consider setting out some norms in a document and making them clear in the intro talk.

Concrete suggestions (for norms):

Talking to strangers is okay: I think EAGs are already a lot better about this than average spaces, but I think it can still be pretty intimidating for those that are new and that this is a pretty robustly good rule to put out (with maybe some small caveat like ‘but don’t interrupt’).
Be present in meetings: I use my phone to take notes sometimes, so sure there are exceptions, but the point of the rules is not to be rigid but to try to guide people towards behaviors that seem good. I haven’t experienced much distraction, but I think the few times I have could likely have been prevented by there being a clear norm of presence.

6. Encourage longer (multi-hour) events:

I’ve never seen a multi-hour session at an EAG, and I think this is missing out on a fairly large potential. This would be helpful for the “making friends” why, but I think this is also defensible on account of impact as longer programming also increases the depth you’re able to go into. As Parker says, “spending 12 hours together as a group is fundamentally different from spending four hours together on three separate occasions. The longer you’re together, the more reality sets in...walls start to come down” (137).

Concrete suggestions:

Have longer events, especially meetups: When I went to EAGxNYC, I basically talked to anyone who was in the AI governance space. I would have eaten up the opportunity to gather together with AI Governance people for multiple hours. With time, we could have gone more in depth into various issues, and could have really taken apart issues and different perspectives. With time, we could have introduced our current work and gotten feedback from a larger group on what we could do better, avoiding redundancy as people could update on feedback already given. We barely had time to cover even a fraction of what we wanted to at the AI meetup at EAGxNYC, and I think you could likely gather a strong group to participate in this sort of discussion at any given EAG. “But we don’t have the space”…encourage events at restaurants near the venue. “But people would need to come and go”…fine, let them. What’s important is that there is a core group carrying things through that serves as a continual grounding point as faces change (or close doors too, there are reasons to support that as well).

7. Use the time before to build excitement:

The time between acceptance to attend an EAG and the first session starting is full of potential to build excitement for the event. The event begins as soon as someone has been accepted (possibly as soon as someone has applied) and more could be done to set positive expectations and excitement for the event to come.

Concrete Suggestions:

Open Swapcard sooner. The CEA Events team has done a great job of opening a Slack for the event sooner so that people can better coordinate on things like accommodation, which I think is great. Though I’m sure there are a number of logistical considerations that make this hard, I think setting Swapcard up sooner would also be quite beneficial, as releasing it two weeks ahead of the event doesn’t give sufficient time to go through and begin building a vision for your EAG.
[Less sure] Pick a time to have a “pre-game” for the conference by, say a week ahead of time, bringing a bunch of people together virtually to meet one another at random (maybe doing breakout rooms) so that people can have a taste before. CEA has already been running virtual “first-timer” sessions before EAGs, so another option would be to experiment more with smaller, more specific meetups before the conference.

8. Consider changing the space in ways more conducive to connection:

Spaces that are too large tend to reduce intimacy and conversation between attendees^[16]. This is one example of a number of considerations about how space can affect what’s happening in them, and it’s good to keep in mind various ways we can tailor the space to better facilitate connection.

Concrete suggestions:

Try to create enclosed spaces, or spaces that naturally encourage people coming closer together. This is as simple as bringing a picnic blanket when dining outside, as even though the grass is generally suitable for sitting, this blanket comes to artificially define a part of space that creates the pretense for coming closer. For EAGs, this could be requesting smaller tables rather than larger ones when given the option, or requesting more large rooms be split rather than kept open.

9. Create a temporary alternative world:

Being at EAG has literally been compared to being on ecstasy, and I felt an intense joy myself at my first. This feeling of entering into a new world, of coming into this really special environment that you’re so deeply grateful and excited to be a part of, can be encouraged by demarking differences between being “here” and “out there”. Priya mostly talks about temporary rule setting to achieve this.

Concrete suggestions:

Be okay with leaning into weird, EA specific things. A lot of EAG programming seems to be aiming to be professional so others in the space would take us seriously if they came, but I think there’s also space to let things be a little stranger.
Have everyone wear some random color the first day. If all EAs show up in black, it increases a sense of cohesion within the conference and creates a sense of convergence that also can play interactively with excitement on the commute into the conference the first day. Every person on the train wearing black makes you wonder “is that person an EA?”.

10. Model the behavior you want to see:

If you want your session, or the conference, to be a place where people feel welcome to open up and share personal things, model that. As an organizer, or speaker, you set the bar for what people are going to feel comfortable with, so if you want the conference to be a place where people open up more, share a personal story yourself at the introduction. This doesn’t just fit into the friendship why, but also could fit into impact, as e.g. greater openness about career failures could help others realize that they’re not set aside from successful people around them but are rather undertaking a similar journey and are just at a different stage.

Concrete suggestions:

Host more debates: I think these are a great format anyways, but especially for showing that negativity is okay and that extended, passionate disagreement can highlight things better than positivity or a carefully laid out lecture can sometimes. This could be facilitated in a number of formats, I like the idea of proper debates but fireside chats are another popular option that may work.

Finally, thanks to Ollie Base, Amy Labenz, Isobel Phillips, and Gergő Gáspár for comments on a draft of this.

^
I use “EAG” throughout here, but I’m meaning this in the “inclusive of EAG and EAGx” sense. I do think there are in fact some limited differences between them, e.g. Learning as a why is likely stronger with EAGxs than EAGs because the crowd is relatively newer (but honestly even this is contestable as you can change the type of focus of the learning depending on the crowd), but that they can mostly be referred to with the same term.
^
This post calls the idea of 1-on-1s as “cleary more valuable” a meme, but in making the argument that talks are underrated ends up arguing for considering “going to >1” talk, highlighting that even those in support of talks place significant value on 1-on-1s
^
This point wasn’t clear to me from the outside, so thanks to Amy for further clarifying.
^
The full definition from a recent EAG: “We define a connection as a person you feel comfortable asking for a favor. This could be someone you met at the event for the first time or someone you’ve known for a while but didn’t feel comfortable reaching out to until now. A reasonable estimate is fine.”
^
A lot of group organizers already do something like this, and one way I did this recently (thanks to Joseph Lemien for organizing) was meeting with a number of people searching for work early on in the conference, briefly doing introductions and then talking about our experience and what sort of work we’re looking for, and then trying to keep an eye out for opportunities for one another as we came across them.
^
For context, all other items were mentioned by less than 18% of participants
^
This was an hour where scheduling 1-1s on Swapcard were blocked and there was no educational content scheduled.
^
To put it another way, while learning and connection look at benefits accrued directly by participants, this metric is purely focused on the positive impact on others.
^
This process seems somewhat opaque, but from what I can tell at a basic level they tried to roughly determine how impactful the work of each applicant was and give more weight to those they deemed to be doing more impactful work. I’m going to go with weighted numbers because the process seems reasonable to me, even though I would ideally like to know more about how it was done.
^
I’m using 2019 because the best data we have is from 2022 and prior, where 2022’s landscape was quite strange due to FTX, and 2021/2020’s landscape was changed due to COVID.
^
I’m taking an average EAG cost of $2.8 million and EAGx cost of $325,000
^
The easiest way to see how problems can arise is to imagine the problem of many individual friend groups showing up to your gathering. As they begin to interact with one another, it could seem that connection is happening and that, despite them deviating from your plan, you’d be wrong to break it up. But Priya asks us to imagine the friendless invitee sitting in the corner of the table, without any such person to connect to. If your purpose is to create connections between all of your attendees, then it very well might make sense to cut their conversations off, and early on.
^
A post from 2019 mentions: “Now the typical person we regretfully don’t accept is probably pretty knowledgeable about EA and has been involved for at least a few years” which I don’t think was widely known. Other factors that used to lead to rejection: “A lot of people from your local group applied and we took some but not everyone; You’re an undergraduate or fairly new in your EA involvement; You’re in a field with lots of other EAs (e.g. you’re in the early stages of getting into cultured meat or machine learning); You’ve been several times before and we want to give a spot to a first-timer; You’re working in a field where we don’t expect to have people at the conference who can be very helpful to you.
^
One objection here is that being clear about the bar would allow for people to game the system, but I think applications could be thought of in terms of job applications. A good job description lets me know what the company is looking for without revealing the exact idiosyncrasies of their hiring process that would allow me to game it, and I think EAG applications should function the same, giving people a clearer sense of what CEA is looking for so people can avoid the whole shameful process if it’s clear from the start they wouldn’t have gotten in.
^
I will be throwing in some page numbers, so you can reference some of these more directly, but will spend most of the time talking more generally.
^
An interesting example given for this that resonates with me: One of the reasons people end up gravitating to the kitchen at a party is because people instinctively seek out smaller spaces as the group dwindles (69)

Thoughts on the Purpose of EAGs

1. EAGs should have a clear purpose.

For Learning:

For Connection:

For Friendship:

For Impact:

2. Don’t be a chill host

3. Be clear about the admission bar:

4. Make the introduction and conclusion priorities:

5. Set norms and be explicit about them:

6. Encourage longer (multi-hour) events:

7. Use the time before to build excitement:

8. Consider changing the space in ways more conducive to connection:

9. Create a temporary alternative world:

10. Model the behavior you want to see: