Epistemic status: The takes below have arisen from (i) conversations I have had with participants who attended or organized 15 different retreats, (ii) feedback surveys and budgets from the retreats, (iii) conversations with the CEA Events Team and the Global Challenges Project and (iv) my own experiences with organizing university group retreats and attending several national and university group retreats.  

The claims made are mine as part of my internship project and should not be taken as representing the opinions of the CEA University Groups Team. I encourage readers to engage with my thoughts critically — it might not be the case that what I endorse applies to your situation. I think there is a reasonable chance that the value distribution might be more heavy-tailed than I have encountered, meaning that some of the value might not be captured here.

TL;DR

• EA groups can apply for Group Support Funding (GSF) to run retreats. In my internship project, I explored the value of these retreats so that we know which levers to pull to increase the value per dollar spent.
• The majority of the value from GSF retreats comes from social outcomes, especially from deeper connections people make.
• Having clear objectives for a retreat increases the chance of more valuable outcomes.
• Optimize for “coziness” over costly venues. Consider simple touches to make a cheaper venue feel homely, if possible.

Introduction

The Group Support Fund (GSF) is one of CEA’s main ways of supporting EA groups financially. This funding is meant to cover principles-first EA groups’ running expenses. One-third of the GSF budget is allocated to EA group retreats^[1]. Recipients receiving this type of funding must meet a higher bar, since retreats are more costly and difficult to organize.  

In my internship project, I wanted to learn how the GSF retreats currently provide value to participants. I was interested in finding examples of especially good EA group retreats, as well as an understanding of what a typical retreat looks like. CEA can then use this  information to advance our understanding, determine which levers to pull to “get more bang for our buck”, and pass along insights and advice to current and future retreat organizers.

My motivation here is to share my thoughts on how to get more total value from  GSF retreats. This is not meant as a call for running more retreats, but rather advice for improving the retreats that are already being organized.

0. My method

To gather insights on both typical retreats and standout examples, I used the following methods:  
i) 15 interviews with 11 GSF retreat organizers and 4 attendees, covering 11 different retreats globally^[2]
ii) Analysis of feedback surveys collected from all 15 GSF retreats^[3]
iii) Analysis of final budgets for the GSF retreats^[4]
iv) One meeting with the CEA Events Team & and one with the Global Challenges Project 

User interviews with attendees provided a more accurate representation of the value of retreats, while interviews with organisers helped identify typical retreat structures and best practices. I also used feedback surveys to further evaluate the value reported at these retreats.

1. What is a typical GSF retreat?

A GSF retreat is a multi-day event that brings community members together, usually outside the city, to engage in EA activities. There are two main categories of GSF retreats: 

• University EA group retreats (uni retreats): Smaller (median 15 attendees: from 7 to 45), generally open to everyone actively engaged in the university group, but sometimes with more selective applications.
• City & national EA group retreats: Larger (median 29 attendees: from 24 to 44), with formal applications and often more experienced participants.

The type and cost of the venue vary significantly, from bungalows on campgrounds to high-end resorts^[5]. For most retreats (around 60%) the food is home-cooked which also can help attendees connect more deeply with one another^[6]. 

There are few funding options in the EA ecosystem for groups hosting retreats, with the main source being GSF; one can also apply for Open Philanthropy’s University Group Fellowship and the EA Infrastructure Fund for retreats and/or national conferences. Groups can also seek funding outside of EA. This is often more specific to the local context (e.g., cross-country European groups can apply for Erasmus+ funding). Several groups partially self-fund retreats through a participant fee. Of the 15 GSF-funded retreats I sampled, 40% had a participant fee.  

2. Which niche do GSF retreats fill?

GSF retreats foster local and regional connections in EA groups by offering a space for shared experiences and reflections covering several days. Unlike EAG(x)s, these retreats focus on smaller communities with more interactive and participant-led activities. The national retreats can also serve as key touchpoints for geographically dispersed EA groups^[7]. 

3. How do GSF retreats create impact? 

3.1 Deeper connections 

The primary value reported by attendees involved social outcomes like new and deeper connections to other group members, feeling socially comfortable in the group, and enjoying the retreat. For organizers, the most reported value was having learned new skills and feeling more confident organizing their EA group.

Of these social outcomes, “deeper connections” were the most frequently reported values for attendees overall, and may also be the most important way the retreats create impact. The theory of change here can be explained as follows: 

• Most retreats are hosted by groups where many or all attendees have met before but don’t necessarily know each other well. Deeper connections take someone from being an acquaintance to being a friend, likely causing the attendee to feel at home in the group. 
• This sense of belonging in the group makes it more likely that these group members become part of the EA community over the long term, either as more engaged members or by becoming local group organizers^[8].
• Both being engaged in the group over the long term and becoming a group organizer will, in turn, make the person more likely to take more impactful actions^[9]. 

3.2 Other value outcomes

Retreat attendees also frequently cited obtaining more object-level knowledge or taking cause areas more seriously as positive takeaways. Less frequently, they described making more progress in their career planning, becoming inspired by connecting with external speakers, or feeling more excitement about EA. Some tangible outcomes included making project plans, booking career counseling, applying to EAG(x) with other attendees, and deciding to found a new field-building or EA organization because of the retreat.

3.3 Impact stories

The most impressive impact stories I came across in the 12 different retreats I assessed in-depth through interviews and feedback forms were:

• AI safety hire: The EA Hungary Retreat 2022 likely helped an attendee get more involved in EA and connect with a future employer.  The person ended up being hired by an AIS organization later.
• Founding a new field-building group: An AI Safety Group was started because of the EA Midwest Next Steps Retreat 2024 for universities.
• Improved self-confidence: An external speaker complimented a socially anxious, highly engaged EA on their social skills at the EA Midwest Next Steps Retreat 2024, resulting in a significant increase in social confidence.  

Further stories:

• 80,000 Hours calls: 7 out of 16 attendees applied for 80,000 Hours calls because of the EA New Zealand Connections Retreat 2023 and the EA Midwest Next Steps Retreat 2024 (I did not confirm how many were accepted or had the call).
• Veganism: 3 attendees reported that they were changing their diets to vegan after the EA New Zealand Connections Retreat 2023 and the EA Midwest Next Steps Retreat 2024.
• Internship: One attendee (with zero previous EA exposure) was invited to the EA Hungary Retreat 2022 and is now starting an internship with a national EA organization.

4. Diminishing returns

Retreats typically have higher reported value outcomes for group members who are quite new and/or haven’t attended a retreat before, as they offer lower barriers to engagement, more object-level EA knowledge, and the opportunity for personal connections that more experienced group members are more likely to have made. Repeat attendees seem to experience diminishing returns. The caveats here are that: 

• Some of the value new group members gain comes from connections with more experienced, engaged EA members.
• Retreats with more focused objectives (e.g., making career progress) can provide significant value for attendees who have previously attended a casual retreat^[10]. These retreats also seem to self-select for attendees who think they will get value from that specific theme.

5. Counterfactual time

Organizers for the national groups would generally have spent the time on other EA organizing activities, as many of them are paid through their national EA organizations (e.g., CBG). 

For uni retreats, the large majority of both attendees and organizers said that the time spent at the retreat was much more useful than how they otherwise would have spent it. Uni retreat attendees generally reported that they would have spent time on academics or taken the weekend off. Uni retreat organizers reported the same, as well as focusing on other EA organizing activities. For organizers, the retreat generally negatively impacted their academics. 

Compared to an EAG(x) or other EA events, the counterfactual use of time for attendees at uni retreats was likely far more impactful than they otherwise would have spent it. This is because EAG(x) attendees may have jobs that are impactful, whilst students are less likely to be doing impactful work day-to-day. 

6. How much do the retreats cost?

One-third of the GSF budget is allocated to retreats^[1]. By sampling 15 retreats, I determined that the median GSF budget^[11] per uni retreat was $2061 and the average uni retreat budget was $2464 with a standard deviation of $2824^[12]. For national retreats, the median GSF budget per retreat was $6202 and the average was $2755 with a standard deviation of $5758. The distribution of GSF funding for retreats is heavy-tailed. An important caveat is that almost half of these retreats received funding pre-FTX. 

Compared to EAGx costs from 2022-2023, the GSF budget per attendee is between 16% and 48% of the EAGx cost per person^[13]. Forty percent of the sampled retreats partially self-funded their retreat with participant fee. The average participant fee was 1.70% of the total retreat budget (or $44.5 with a standard deviation of $36.4).

The largest budget items are the venue costs, which on average amount to 60% of the total budget.

7. Which heuristics can we follow to increase the cost-effectiveness of retreats? 

Below are some practices that seem especially promising. This is not an exhaustive list, and represents my opinions after working on this project during my summer internship. 

I recommend:

• Setting clear objectives and designing the retreat accordingly. If your objective is for attendees to make progress on their career plans, find the target audience that best fulfills this criterion and plan the content to optimize for it.
• Ensuring there are actionable next steps. Content like workshops aimed at clear, tangible outcomes can be very motivating for attendees and might nudge them to think more deeply about which role EA could play in their lives. Example: A career-planning workshop with defined ‘next steps’ to follow up on after the retreat, e.g., 80,000 Hours application session. 
• Optimizing for (inexpensive) coziness. The venue’s cost matters less than the atmosphere. Prioritize ways to make a cheaper venue feel homey (e.g., light some candles) instead of going for the more expensive option.
• Creating opportunities for deeper connections. If the retreat is focusing on social outputs, prioritize activities that encourage authentic connections. Ensure that activities are inclusive to allow participants to engage at their own comfort level, and plan a mix of scheduled and unscheduled social events that promote meaningful conversations or shared experiences.

8. Best Practices for Organizers: Groups Resource Center

As part of this project, I have updated the Groups Resource Center page about retreats with practical advice and best practices for organizers planning their retreat. It also has a new section dedicated to templates, both for content at the retreat and for the logistical aspects of running the retreat.

Does your EA Group want to apply for Group Support Funding?

__

 Reactions and ideas are welcome! Feel free to post a comment or reach out to unigroups@centreforeffectivealtruism.org. 

I want to give a big thanks to Sam Robinson for all their help during the summer. I also want to thank Joris Pijpers and Jessica McCurdy for providing feedback on drafts, Heidi Basarab for helping improve my writing and to Cian Mullarkey for helping me set up the project — and to the rest of the Uni Groups Team for good vibes!

1. ^{^}

   2023 budget

2. ^{^}

   The user interviews were with (11 male and 4 female) group members, covering 11 different retreats globally. :

   • UK: 3 organisers, 2 attendees
   • Mainland Europe: 5 organisers
   • Asia: 1 attendee
   • Africa: 1 organiser
   • US: 1 organiser, 1 attendee
   • Australia & NZ: 1 organiser
3. ^{^}

   The feedback surveys were gathered from the individual retreats, and so were not standardised, and were generally conducted on the last day of the retreat.

4. ^{^}

   This is data that is collected from the EA groups after the GSF grant period has ended.

5. ^{^}

   These are usually in LMICs.

6. ^{^}

   Three out of 15 interview participants highlighted, without prompting, how volunteering and cooking together helps them connect more deeply with others.

7. ^{^}

   Two out of four interviewees who attended national retreats reported this.

8. ^{^}

   This I infer from: i) user interviews and retreat feedback reports of attendees who want to become more involved with the community and ii) impact stories with attendees becoming organisers partly because of inspiration at a retreat.

9. ^{^}

   Groups Resource Center: How do EA Groups produce impact?

10. ^{^}

    Four of 11 of the retreats I looked at fit into this category.

11. ^{^}

    Amount of budget funded by GSF — several of the retreats are partially self-funded.

12. ^{^}

    Using exchange rates from August, 2024

13. ^{^}

    EA Forum Post: How much do EAGs cost and why?