Thanks for writing this up, great post!

The example of the student is if fair, but my claim is that you can get those effects in cheaper ways than a retreat. For any given community retreat, only a couple people will attend with profiles for which it's realistic that a retreat can help them get a role (see my point about retreats having too low of a bar). I would want to support these in other ways.

I'm worried though that the scale is too small to power an RCT. But I do like your idea on that we should try to measure the effectiveness of these interventions more rigorously!

Maybe I misunderstood, but if retreats and conferences/summits are complements, this argument should not apply?

They might be complementary, but most communities don't have the resources to do both. (Unless they are a year apart, at which point any momentum you get fades).

Thanks for pointing out that we actually have some more data on costs from GSF retreats.

On the other hand, I think the argument of my post is being left unaddressed: that we should prioritise summits over community retreats. As Ollie mentioned above, larger events provide the same benefit per participant. EAGx Conferences may be at a similar cost per participant compared to retreats, but I would be really surprised if the same was true for summits, which are only one day. I would love to get a better sense of how expensive summits are per person thoguh!

I have had the chance to visit Mox recently (not as part of this program), and I strongly recommend that anyone to take advantage of this opportunity. It's an amazing space with value-aligned leadership as well as great members. It was long overdue for SF to have such a co-working space! :)

You’re right that these aren’t the same kinds of events. My claim is about prioritisation under limited time and resources: communities should focus on the events that are more cost-effective.

If a community only has the capacity to run one event per year, I would prefer it to be a summit rather than a retreat. If the community is large enough to run an EAGx, I would prioritise that, and then add a summit around six months later if capacity allows.

I'm not sure if CEA gives out travel funding for EA summits, but my take is that they should do so only in very rare cases. One-day conferences, at least in most places, should focus on people who live in the city they are held.

Thanks for these comments!

I agree that retreat costs can vary a lot, and it’s quite possible that many retreats are cheaper today than the ones Ollie looked at. That said, the best data we currently have still suggests that retreats-as-they’ve-typically-been-run are relatively expensive per person per connection compared to larger events.

I’m very open to updating on newer or more representative data. My prior, though, is that while retreats may have become more frugal, conferences and summits have as well — EA events overall seem meaningfully leaner than they were a few years ago.

On outcomes: I don’t think I claimed that larger events produce better outcomes per person. As you note, Ollie finds those are roughly similar. My claim is instead about scale: one-day summits produce many more total positive outcomes for a similar or lower cost, largely because they involve more people and avoid subsidised travel and accommodation. (Or they should)

Given that, I would be surprised if a typical 2–3 day, fully subsidised retreat (which is the default in EA) ends up more cost-effective per person per connection than a one-day, local summit — even if some individual retreats are run quite cheaply.

I see the extent of badness of social media platforms (for mental health or society at large) as orthogonal to the question of whether they can be used to cost-effectively attract talent to EA.

Short-form content doesn't have to sacrifice fidelity! It's going to contain less information, but the way to use it is to attract people to go onto engaging with long-form content.