Midwest EA’s Next Steps Retreat Post-Mortem
Thanks to Jessica McCurdy, Olivia Jimenez, Lenny McCline, Leilani Bellamy, and Harry Taussig for giving us feedback while we were planning this retreat.
The goal of this post is to summarize a Midwest EA Retreat run April 8-10th, 2022. The retreat hosted promising and engaged members from four university groups (Northwestern, UChicago, UMichigan, and UW–Madison) who recently completed an intro or in-depth fellowship (hereafter, ‘applicants’).
We expect this post to be most valuable to (a) grant-makers looking to assess the impact of retreats like this and (b) other organizers thinking about running similar retreats.
Our main takeaways were:
Results: Across all attendees (applicants and invited guests), survey results suggested the retreat was valuable. Attendees were 87% likely to recommend the retreat to a peer and everyone said the retreat was at least as valuable a use of their time as their counterfactual. Average attendees # of connections in EA and sense of belonging increased. There were only minor fluctuations in likelihood of pursuing a high-impact career path. All organizers agreed this retreat – the first of its kind for the Midwest EA retreat – was valuable and would be keen to run it again (if no other organization is running a retreat in the Midwest area)
While planning, talk to retreat organizers as well as reading retreat write-ups/planning documents
Collaborating with other organizers can make organizing a retreat feasible, if few/none of the groups have sufficient capacity alone
Retreats can help cultivate a regional network
Spend time clarifying strategy and priorities after significant planning shifts
Have excess operations capacity
Thoroughly plan for worst-case scenarios (e.g., a positive COVID case)
TL;DR: We had four main goals, listed here in descending priority: increasing the number of highly-engaged EAs (HEAs), identifying the most promising HEAs, and providing a practical and engaging follow up to the fellowship(s).
We had three main goals. We have outlined them below in descending order of priority, along with how we thought this retreat could fulfill those goals and relevant metrics.
Increase the # of Highly-Engaged EAs (HEAs)
Theory of change: Most of our applicants had little exposure to the EA community beyond their university group. We hoped that introducing them to students from different schools and working professionals would lead them to see EA as a wider community and update applicants towards viewing EA as a serious commitment and community that could support them.
Similarly, we thought that a retreat could improve intra-group social cohesion, making it likelier that each group’s members would bond and stay engaged.
Metrics: Changes in self-reported rankings to connection to an applicant’s university’s EA community, connection to the EA community, and the number of people in the EA community they would feel comfortable asking for a favor from.
Identify the most promising HEAs
Theory of change: By definition, applicants had had less time to demonstrate their promisingness. We thought that a fairly intensive weekend could help us better assess the most promising members within and across our groups.
Metrics: There was no explicit metric for this as it hinged on organizers updating their intuitions.
Provide a practical and engaging follow-up to the fellowship(s)
Theory of change: Some of our members felt like there was no clear next step for them after a fellowship. Relatedly, there were concerns that the EA community was too focused on discussion over action.
We thought both of these could be addressed by career planning but we had also heard concerns that it wasn’t clear how to apply EA principles to career planning, or even how to begin with career planning. Thus, we wanted to provide members with concrete questions (e.g., from an 80k workshop) and action items (e.g., filling out a worksheet) to help them begin career planning.
Metrics: Change in self-reported estimates of their likelihood of pursuing a high-impact career path; Self-reported value of the retreat compared to their counterfactual
Our priorities shifted during the planning process towards identifying HEAs but we were too slow to realize this, resulting in us not updating our goals accordingly.
We would have benefitted from (a) articulating what success looked like for our retreat at the beginning, and (b) reassessing our priorities when we decided to invite existing HEAs as guests.
TL;DR: ~40 attendees, with a 50⁄50 split between applicants (the most promising members from Midwest EA groups) and special guests (other university organizers and working professionals).
Originally, we only planned to invite the most promising but new members from our groups (‘applicants’). We expected most of these members to have recently co-facilitated a fellowship or completed an In-Depth Fellowship. Since some groups had not run an In-Depth Fellowship, however, most of our applicants ended up being recent Intro Fellowship alumni.
About halfway through our planning process, we spoke more with Olivia and Jessica and updated towards having a 50⁄50 split between our applicants and members of the EA community who were already highly-engaged (existing HEAs). This led us to expand our attendee pool to other university organizers and working professionals (special guests). Based on their advice, we anticipated inviting 25-40 people. This seemed like a manageable number for a team of six, particularly since only half had any experience with EA retreats.
We were not as systematic in inviting special guests (i.e. already engaged members of EA community). For university organizers, we drew upon our existing networks and prioritized organizers from bigger and/or more active university groups. For professionals, we focused on professionals in the Midwest, partly out of convenience and partly to build the Midwest EA community. Finally, we reached out to 80,000 Hours for help with planning our career programming and they offered to send us an advisor (Matt Reardon).
We are especially grateful to Matt Reardon from 80,000 Hours, who led our two career workshops. He even followed through despite getting COVID and attending virtually!
We aimed for roughly ~5 applicants from each university, for a total of 20 applicants. This was determined in two stages, which are briefly outlined below. If organizers would like more detail on our selection process, please message us and we’d be happy to share.
Selection Process reflection
A few organizers felt that the retreat could have been more valuable if the average attendee was more engaged, but not all organizers agreed on this. Setting a higher bar for engagement would have meant fewer attendees, and it’s not sure that the benefits of more engaged discussions would outweigh the possibility that you shift newer members from early EA to highly-engaged EAs.
However, as with past retreat organizers, we share the reflection that even a few people who are not taking the retreat and EA ideas seriously can significantly detract from the whole vibe.
Racial and gender diversity was low
We found inviting incoming organizers was especially valuable, as it gave them the chance to connect with future regional other group leaders (both present and future), and discuss collaboration going forward.
Planning + Rough timeline
We met 1x/week for one hour, for four months (December 30th, 2021 to April 7, 2022). Though we were only in ~16 hours of meetings, we spent significantly more time (probably closer to ~80 hours) preparing for the retreat, such as creating programming or purchasing supplies.
Rough timeline below. Note that retreats have and could be planned much faster, particularly if the organizers are not dispersed throughout schools.
December 2021: Midwest organizers agree to run a joint retreat.
Early January 2022: Teams created & roles delegated out. Timeline and future action items planned out, based on Yale EA’s example timeline.
Late January: Tentative list of venues created, tentative invites sent to special guests. Marketing materials (e.g., application form, website) are started.
February: Venue booked & dates finalized, applications sent out to group members. Midwest EA decides to push the retreat back from mid-March to mid-April, after EAGxBoston finalized their dates.
March: Applicant decisions finalized and communicated, agenda finalized. What was really useful was walking through our draft agenda and doing a ‘vibe check:’ when we pictured ourselves experiencing our programming, did it feel like what we wanted?
April: Logistics confirmed (e.g., travel to the venue finalized for all attendees), COVID policy updated post-EAGxBoston. (EAGxBoston was also the first time Midwest EA met in-person!)
Deciding on Programming
Defining goals for each day
We adapted the original Next Steps Retreat schedule, but added an extra day. We organized programming based on the goal(s) of each day and overall retreat flow.
Friday: Getting to know each other
Saturday: Career programming and deep work
Sunday: Tools for sustainable progress (e.g., rationality, self-care), next steps
Each piece of programming was then evaluated on how well it achieved both our daily and our overarching retreat goals.
Emphasis on social programming
We emphasized social programming, especially on Friday, for two reasons:
This was the first collaborative event between all midwest schools, and most attendees had <1 year of EA exposure. Thus, attendees were largely unfamiliar with each other, so we expected they’d need more time to establish the trust that would enable them to feel comfortable and friendly with each other.
We thought forming bonds with EAs from other schools would create a sense of community that can serve as the basis for engaging more with EA.
Emphasis on 1on1s
We emphasized 1on1s because they achieve two separate goals:
Connecting attendees to HEAs. We wanted to make sure attendees came away from the retreat with at least one valuable, lasting connection and we think these typically come from 1on1s. Newer EAs are often hesitant to approach professional EAs and/or don’t know who is most relevant to them, so we wanted to match them (at least once).
On-site Career Guidance. 1on1s with relevant HEAs can provide concrete next steps, feedback on career plans, and connections to other professionals.
Emphasis on deep work and sprints
We thought that providing a space for personal, intensive career planning would work to “Provide a practical and engaging follow-up to the fellowship(s)” and “Increase the # of Highly-Engaged EAs (HEAs)”.
Career workshops generate numerous action items, which are often forgotten about or de-prioritised when a retreat concludes. To mitigate this risk, we budgeted in-retreat time for attendees to get started on these next steps.
After drafting a schedule we verbally walked through the programming chronologically to evaluate it. Doing so we considered several factors:
Energy levels—are valuable events too late in the day?
General vibe—have we packed too much intense programming together?
Retreat Arch—does the programming “peak” where we want it to (in this case, Matt’s 80k workshop)?
Selected Programming Reflections
If you have questions on any programming not mentioned here, feel free to reach out to Michel. Note that all events after breakfast Sunday did not happen because someone tested positive for COVID and we ended the retreat early. While we have slides or ideas for these events, we don’t have feedback.
Paired 1on1s: 1on1s were probably our most valuable piece of programming. We expect significantly fewer and less relevant 1on1s would’ve occurred had we not made these mandatory. Planning pairs in advance (e.g. this organizer and this applicant have similar interests) led to better career and social connections, and made running the session considerably smoother.
80K Career Workshop: This was also highly valuable. We’re glad that an 80k advisor ran this program as it made it seem far more prestigious and was facilitated better than if by an undergrad. This event was conducted virtually since the 80K advisor got COVID, but this format still worked fine.
Sunset Walk: People like walks. This was an enjoyable walk, and it can be nice to split up the social scene at times and allow for more personal connections to form.
Lightning Talks: These didn’t happen. Predictable lesson learned: don’t plan programming from 9PM, even if you call it low stakes. People want to chill.
Closing Session: Our final session featured a short talk, followed by everyone going around and ‘shouting someone out’ (e.g. complimenting someone for a good conversation or a fun activity). People seemed to really appreciate the shout outs and the general atmosphere was very positive. We recommend organizers try to shout out attendees who haven’t been mentioned yet.
Running lots of social programming, early
After arriving Friday afternoon we found filling the rest of the day with entirely social programming quickly got attendees comfortable. By the end of the day, the vibes seemed generally good and most attendees had mixed well with students from other groups.
We may have tried to do too many events
Having lots of events has a few downsides:
Doesn’t leave enough space for spontaneous planned events
Takes more work to organize in advance
Takes more work to organize day of, which leaves organizers running around
Having to cancel an event that few people show up to may send the message that the programming isn’t valuable.
Despite emphasizing that every event was optional, we felt the packed schedule detracted from the emphasis on 1on1 spontaneous conversation, as attendees may have still felt they had something else to go to
Our schedule was pretty packed and a few events had to be canceled (in addition to those canceled because of COVID positive test), so we may have suffered from these downsides. But the point around the ideal amount of organizing is a nuanced one: you want enough programming so that attendees don’t feel driftless and engage with the key goals, but you also want to avoid the above downsides of too much programming.
Spontaneous events should be encouraged
We recommend that retreat organizers encourage attendees to think of and share any spontaneous ideas they have in a central communication channel (e.g. Slack channel). This should be coupled to a clear indication of whether this will happen (e.g., “if this message gets three likes I’ll start a Dungeons & Dragons game”).
Give all attendees clear expectations
One guest suggested we could have done better to “make sure that everyone knows what’s expected of them for a given time period or event.”It is important to give all attendees, including students and invited guests, ample time and information about the expectations for each retreat’s day. For example, on Friday night, clarify that the expectation is that people will just get to know one another; the expectation is not that we’ll begin solving career cruxes already.
People like to clean
Cleaning makes people feel like they’re helping. It can be bonding, and it makes things look nice. Feel free to budget this into the schedule. We suspect this generalizes to most communal projects like cooking, setting up a room for an event, etc.
Make sure that organizers are aware of this goal and that norms are shared with everyone to uphold inclusivity (e.g. mention “pacmanning,” which refers to holding conversation in a circle that leaves space for someone to join).
Before the retreat
Our criteria were Location, Beds, Dining options, Meeting space, Recreational, Price
You can see our spreadsheet here
Arrive at the venue early to get things set up. Some people have recommended getting there a day in advance but we found that having three people get there ~4 hours before was sufficient.
Ordered all vegan food, had all of the orders confirmed ~1-2 weeks in advance
We only ordered lunch and dinners and brought stuff like bagels/oatmeal/fruit for breakfast, which people seemed to enjoy
We tried using ezcater.com but it had a somewhat limited selection of vegan food. We had more luck doing broad searches for things like “thai food Lake Geneva” and calling restaurants directly
We re-used the list from the original retreat. This is a pretty generic list.
We made a separate list for organizers.
An organizer tested positive for COVID one day before the retreat began, which left the team operating at ~80% capacity. We were able to quickly pivot, including re-assigning programming that they were meant to lead and assigning another organizer from the same university.
We had two positive tests on Sunday morning and isolated both people. We moved everyone outside after that and cut all of our Sunday programming.
Important that all organizers know where key items are (e.g. everyone should know where the first aid kit and extra supplies are)
Internal slack channel for communicating behind the scenes worked pretty well but was sometimes difficult to check
To attendees & special guests
Sending messages out earlier could have been better
We think it’s valuable to reach out to participants to encourage them to follow up with any connections they made, perhaps twice.
During the retreat
Logistics can be divided into frontend (directing people, answering people’s questions) and backend (setting up events, making sure food arrives, etc). From our experience, both the front end and the backend logistics for a retreat of this size take up at least one organizer working essentially the whole duration of the retreat.
Two organizers arrived a few hours before the retreat to set up the venue. They brought in snacks and other supplies, labeled rooms, set up air mattresses, and did lots of other miscellaneous things like distributing toiletries in each bathroom.
There was one organizer each for backend and frontend logistics.
The organizer for backend logistics made sure that all the supplies were set up correctly, coordinated with the restaurants to get food, and made trips to grocery stores and restaurants to pick up the food.
The organizer for frontend logistics greeted attendees upon entry and gave a summary of what was happening and answered any questions that they had. The spiel upon entry was roughly:
A short greeting conveying excitement that they were here
Giving them a covid test
Their room assignments and directions to their room
A summary of the schedule for today, and giving them the goal of getting to meet people and form new friendships with people they hadn’t met before.
Asking if they had any questions
We did a lot of planning ahead of time, which was very useful. However, many last minute changes required us to change a lot of our logistics: we had several cancellations and additions. This caused us to need to swap around the room assignments and 1-1s pairings during the beginning of the retreat.
We had half the response rate for the post-retreat survey (30 responded to the pre-retreat survey and 15 to the post-retreat survey), partly because we were not prepared to deal with the COVID case that occurred during the retreat. While we have reported key results below, these should be taken with a big grain of salt.
The below data was only collected from applicants.
Before the retreat (n=30)
After the retreat (n=15)
|What’s your current likelihood of pursuing a high-impact career path?|
5% (1) reported 20-40%
24% (5) reported 40-60%
33% (7) reported 60-80%
38% (8) reported 80-100%
0% reported 20-40%
43% (3) reported 40-60%
14% (1) reported 60-80%
43% (3) reported 80-100%
[From 1-5] How connected do you feel to…
Your university’s EA community: 3.9
The EA community: 2.9
Your university’s EA community: 4.3 (10.3% increase)
The EA community: 3.4 (17.2% increase)
[From 1-5] How much do you enjoy being part of…
Your university’s EA community: 4
The EA community: 3.7
Your university’s EA community: 4.1 (2.5% increase)
The EA community: 3.9 (5.4% increase)
|How many people in the EA community would you feel comfortable asking for a favor from?||6.4||12.5|
We also had the following non-comparative metrics, which applied to all attendees. We have reported them as an aggregate; for a more detailed breakdown, feel free to contact us.
On average, attendees were 87% likely to recommend the retreat to a peer.
Everyone said the retreat was at least as valuable a use of their time as their counterfactual. 7% (1) said the retreat was about as valuable, 20% (3) said it was 1-3x, 40% (6) that it was 3-10x, 27% (4) that it was 10-30x, and 7% (1) that it was >30x.
Interestingly, the most valuable aspects of the retreat changes between the qualitative and quantitative responses.
Qualitatively, 1:1s were universally the most valuable. Some participants suggested having more time for 1:1s and strongly encouraging attendees to schedule 1:1s.
Other particularly valuable experiences included the career workshop professional panel, the evening socials, the paired 1:1s, and the paired dinner groups.
Experiences that were considered least valuable included the career workshop, professional panel, and feeling excluded from group conversations.
Quantitatively (from 1-5),
Paired 1:1s were rated an average of 3.6
The opening session was rated an average of 3
The career planning workshop was rated an average of 3.3
The professionals panel was rated an average of 4
The career sprints were rated an average of 3.5
‘What’s next for you and your school?’ was rated an average of 3
The closing session was rated an average of 3.2
We think this may be because participants found more value from spontaneous 1:1s than from the pre-paired 1:1s.
We also had some optional sections.
Two bottlenecks mentioned by respondents who put their likelihood of pursuing a high-impact path as <80% were uncertainty and alienation around EA, and worry over sacrificing happiness/fulfillment to pursue a stressful career that they don’t enjoy.
Reported concerns and uncertainties around EA were
Feeling overwhelmed by trying to maximize impact (1)
Being concerned about not having many concerns with EA (1)
Representative, e.g. gender (1)
Dogmatism & ‘cult mentality’, e.g. around longtermism (2)
Overarching Goal Reflection
Increasing the number of highly-engaged EAs (HEAs)
We did a poor job collecting data on this. Anecdotally, participants enjoyed the retreat. Organizers are around 75% confident that, on balance, members who attended the retreat will be more engaged in next semesters than they would be if they hadn’t attended.
Expanding applicants’ EA action space following fellowship
The Next-Steps retreat proved to be a viable follow up to the fellowship. Post-fellowship attendees made meaningful career plan changes, and heavily engaged with EA content.
Identifying the most promising HEAs
In the end we did not optimize for this, mostly due to a reliance on pre-existing material from organizers who may not have had this goal. While lots of interaction with attendees may have somewhat helped to identify especially promising members, the retreat was not significantly better than 1on1s in this respect.
Takeaways for Future Retreats
Developing a regional network
Both attendees and organizers made connections and friends with EAs from nearby universities. These connections improve the diversity of attendees’ networks and may increase the likelihood that organizers will coordinate with other regional groups. Generally increasing attendees’ connections also fosters a valuable sense of community within EA as a whole.
In addition, incoming leaders from UMichigan, UW-Madison, UChicago, and Northwestern had the chance to meet, share resources, and discuss future collaborations. These leaders are planning on running a similar retreat next year.
Have extra floating operations people
We had about one organizer for every six guests, and could have benefitted from more floating support. These extra operations people don’t have to be deeply involved in organizing the retreat or running the programming—just having someone who can consistently take on routine tasks that make the retreat run smoother.
Plan for worst-case scenarios (e.g., positive COVID cases)
We would have benefited from a more thorough COVID protocol, especially one which lists a detailed plan for if an attendee tests positive during the retreat. Retrospectively, we think planning for a worst-case scenario was too aversive and thus never discussed in detail. Organizers should make a point to overcome this.
Consider the dates surrounding your retreat
If possible, have your retreat at the beginning or ~4 weeks from the academic term (ideally, the first or second weekend). Students are generally busier in the middle of the term. Additionally, try to avoid dates that are close to other EA events. Our retreat was sandwiched between EAGx Boston and EAG London. This meant many attendees and organizers had back to back EA weekends. Our retreat was fairly chill, but we still think that leaving buffer time between EA events is weakly preferable.
Should you do a retreat?
We agree that University Groups Should Do More Retreats.
Type of retreat can vary. Generally, the amount of work required for a retreat scales with how inexperienced the attendees are (and the organizer) and how big the retreat is. There’s a version of a retreat that’s really small with the core group of EAs already taking ideas seriously that’s still really valuable but not as much work. Nevertheless, many people require a more entry level retreat to first start taking EA very seriously.
It seems important to consider how many other similar opportunities to engage with EA there might be for members (e.g. EAGx events). However, we still feel that retreats often create a more personal, friendly environment that is unique from EA conferences.
If you’re very time constrained, consider running your retreat at a dedicated retreat center or hiring people to do a significant amount of the logistics for you.
Consider collaborating with other nearby university groups
Running a retreat can be a good way to test fit for operations work for organizers
Google Drive, which includes folders for
Forms (we haven’t made templates of these but feel free to ask for generic versions)
Both are adapted from the original Next Steps Retreat Notion, which is similar to this Notion Template
If you are organizing a retreat, feel free to message us for any other resources (e.g., organizer packing list, email templates) that would be useful.
This retreat was inspired by Olivia & Harry’s Next Steps retreat.
Organizers: Northwestern (Kasey, Nick), UChicago (Miranda), UMichigan (Thomas), and UW-Madison (Max, Michel)
However, the end of the survey result had a low response rate so this should be interpreted with caution.