At the start of this month I co-organised Saddle Point, a three-night & 30-person retreat. (Duncan, my co-organiser, reviewed this post before I published it, but all opinions are mine alone.) Overall, things were a success! Average satisfaction was 8/10; no participant thought they would have preferred to do their counterfactual activity instead; the median willingness to pay was 50% higher than the price (£300 vs £200); we generated around £10,000 of surplus (excluding costs of organiser time).
When preparing for the weekend, I tried searching for blog posts about what running a retreat is like, but couldn’t find much that was relevant. I’m hoping that this somewhat fills the gap – it’s a mixture of practical details about the way we did things, personal reflections on how the experience differed from my expectations, and what I’d change next time. You can also read a Claude-written summary of the anonymous feedback we received after the event here (we managed to get a 93% response rate, mostly thanks to individually messaging people who we thought hadn’t filled it out) – it’s interesting to see which aspects were criticised in both directions (e.g. too much & too little structure, price too low & too high).
Outside of the weekend of Saddle Point itself, organising was actually quite a bit easier than I was expecting – I estimate that I spent around 15 hours total working on it pre- and post-event, and we were in a position to start accepting applications only five days after I first had the idea (which was six weeks before the weekend took place).
If you’re considering organising a retreat or other event like Saddle Point, I’d be more than happy to chat, so do feel free to reach out! Kristi has put together a very handy list of things to bear in mind – and look into getting event liability insurance too.
Advertising, applications, attendee diversity
- The vast majority of attendees heard about the event directly from Duncan or me, and were known personally to (at least
one of) us. We got the word out by posting in Discord servers & WhatsApp groups for the communities we’re part of,
and speaking directly to some individuals we were especially keen to have join us.
- Malaika made a good comment about this: it would be better to phrase messages in a less promote-y
way, as if it looks too much like bulk copy-pasted advertising people are liable to just ignore whatever you’ve sent.
- It’s definitely true that I could’ve written a more personal-sounding message with a little. more cognitive/emotional work in about the same amount of time, and that this would’ve piqued interest more.
- That said, I don’t think we suffered too much from not doing more engaging promotion – because most of the people seeing the advert already had some kind of connection to Duncan/me, they were predisposed to read what we said, and we didn’t have to try especially hard to catch their attention.
- I did write more personalised messages when DMing people as opposed to posting in a group, and I think this was sensible.
- Having “spikier” copy for the adverts might have also helped us attract especially cool potential attendees from outside our existing networks – although the information doc we wrote up was designed to get across the vibes correctly, the advertising messages were not, despite being people’s entrypoint to the brand.
- Malaika made a good comment about this: it would be better to phrase messages in a less promote-y
way, as if it looks too much like bulk copy-pasted advertising people are liable to just ignore whatever you’ve sent.
- In large part due to the sorts of places we advertised, practically everybody who expressed interest in the retreat
was linked in some way to the rationality / EA / AI safety ecosystem. This meant that our attendees
were a lot less diverse – demographically and intellectually – than ideal.
- Among attendees, the female:male ratio was 1:5. This was roughly in line with the applicant proportions, though we had two women (and no men) who’d been invited drop out, which skewed the ratio further. I was aiming for no worse than 1:2, and will target this again next time (also working towards 1/3 non-rationalists).
- Ages ranged from 18 to mid/late twenties; we didn’t accept any minors to avoid potential safeguarding complications (there might have been no problem but there would’ve been some amount of hassle & risk involved).
- We were significantly oversubscribed, and there were many people who we had to turn down for lack of space.
- After the first few days went by with fairly weak interest, I fretted that £200 was too expensive, but things all turned out fine.
- However, I suspect the price may have disproportionately put off women and non-rationalists.
- I would’ve liked to have had another co-organiser who was female and/or had a substantial network outside rationality-adjacent spheres.
I did try asking three people to fill this role, but none was interested/available.
- Stakeholder engagement is important, and I don’t have a great intuition on what language/phrasing might be reassuring or worrying to different groups of potential attendees.
- Anecdotally, one hesitation from women I talked to was that they didn’t like the idea of going to stay in a house
with strangers. It’s obviously not possible to run a retreat without this being a feature, but there being a female co-organiser might
have reduced the concern at least somewhat.
- I hypothesised that direct referrals from another woman would help make women more comfortable & interested in attending, but I didn’t get a sense of whether this actually happened.
- It was probably counterproductive for me to message people asking them to invite female friends and implicitly linking that to us having an imbalanced gender ratio. If someone told me that some event had X group in a small minority, I would not be especially inclined to tell my X friends to attend, because I’d worry they might have a bad time.
- I think our expression of interest form
was great – high signal, low effort to fill out,
and not wasting people’s time by making them repeat information available elsewhere.
- It was very useful having the question asking “What activities would you be interested in running?”.
- directly relevant, actually useful, and fun to do for the sorts of people we want coming
- We made our acceptance decisions from the written submissions alone, but with hindsight it would’ve
probably been worth doing a 15min call with the people we didn’t know before inviting them.
- A few people who neither Duncan nor I knew decided that they would only put a LinkedIn profile URL into the “more about me” section, which was obviously not enough context for us to be able to invite them.
- It was very useful having the question asking “What activities would you be interested in running?”.
Communication and structure
- I think the info doc did a good job of communicating our intentions for the event – participant-driven,
fairly light on structure, an eclectic mix of activities.
- Counterpoint: in the feedback, some people described it as a “conference”, which was definitely not a word we ever used,
and has much more formal connotations than the atmosphere we were aiming for.
- The schedule each day slipped a fair bit, but I think this was totally fine. We had two-ish check-in sessions per day to get everybody on the same page, and these were very valuable.
- Counterpoint: in the feedback, some people described it as a “conference”, which was definitely not a word we ever used,
and has much more formal connotations than the atmosphere we were aiming for.
- It was naive of me to think that people would
actually read all of the info doc and participants’ agreement.
- I think several attendees may not have realised until they arrived that all food would be vegan, because this was kind-of buried in the communications (since, in a bit of typical mind fallacy, didn’t think that this was an especially important fact).
- If I wanted the participants’ agreement to have been a proper thing, it should’ve been circulated with DocuSign rather than a Google Doc
(since I could theoretically have edited it post-acceptance). But as a Claude-written pro-forma it was just pretty pointless, and a shorter
doc setting out norms in a few bullet points would’ve been better.
- It also didn’t explictly say what would happen in the event of boundary violations (e.g. removal from the event). Happily nothing came remotely near near this, but still should’ve included some disclaimer to this effect.
- We set up a WhatsApp group for communications but in retrospect Discord would’ve been more convenient (different
channels for announcements, quotes, session changes, etc).
- Though maybe a WhatsApp group is marginally more likely to stay active, especially if lots of attendees don’t usually use Discord.
- It might have been good to send out information on the expected guest list along with invitations, as a couple of people messaged me asking about this.
Culture
- Pareto efficiency is great, but (believe it or not!) so is having a nice social atmosphere where
nobody feels excluded.
- One feature of Saddle Point I didn’t like – and foresaw as a potential issue in advance – was the use of real-money financial markets to allocate resources. There was buying & selling of rights to comfier beds, an auction for places at an especially busy session, and bribing of others to substitute in for kitchen duty.
- Sure, these mechanisms might lead to an «optimal» outcome, but it creates an environment that I just don’t really find very comfortable or kind.
- At least six participants mentioned this as a negative in the feedback forms, and next time I would push harder to not have these kinds of trades at all, even as semi-joking entertainment.
- I don’t think there’s really any way to tell in advance from answers to the expression of interest questions
whether somebody is the sort of person who will contribute to rather than pollute the commons. I’m not sure how I’d know even from an interview.
- The vast majority of people who attended were not pro-active at helping tidy up – e.g.,
I think someone other than me emptied the bins only ~2 out of ~9 times; one time when I pointed out that some
mugs could be loaded into the dishwasher, nobody in the room actually acted and so I did it myself.
- I asked people to do a “2min tidy” once and should’ve done so more, but even having to do that is psychologically taxing.
- The other solution to the bystander effect is singling out individuals and asking them to do a task, but my guess is that often people would get in a huff about being arbitrarily chosen, and sometimes not complete it to an acceptably good standard.
- The vast majority of people who attended were not pro-active at helping tidy up – e.g.,
I think someone other than me emptied the bins only ~2 out of ~9 times; one time when I pointed out that some
mugs could be loaded into the dishwasher, nobody in the room actually acted and so I did it myself.
- Enforcing rules is hard, as is deciding whether you should.
- I could have gone around telling people to be quiet after 11pm as they had agreed to, but I think overall letting people do what they wanted did lead to higher social welfare. ~24/30 people were still up chatting at 11:30pm on the Saturday and they were having a good time; shushing them would’ve forced that to stop.
- Similarly, it simply didn’t seem worth nagging people about doing their washing up.
Observations from during the retreat
Food
- This was the source of the greatest number of complaints by a long way, I think to a rather unreasonable extent.
On the first night, I asked people to look over the proposed menu and make any requests for changes – nobody did,
but that didn’t stop dissatisfaction with what was served later on.
- For the first two meals in particular, people took more than had been planned for and so we ran out of servings.
There was never a shortage of food in aggregate: we had plentiful snacks and also around a dozen ready meals in the fridge
(only two of which were eaten during the retreat), but people did feel upset about this.
- Since it’s impossible to buy exactly the correct quantity of food, we should’ve explicitly told people as they were helping themselves what the expected portion sizes were and not to have more than that until seconds.
- More protein was another common request, but apart from the first meal (pesto pasta with roast vegetables),
I think there was a perfectly normal amount of protein:
- Falafel wraps with hummus, cucumber, carrot, olives, lettuce, diced tomatoes
- Veggie bangers and mash with peas, grilled asparagus, and caramelised onions
- Brunch (hash browns, veggie sausages, beans, mushrooms, roasted tomatoes, scrambled
tofueggs, bread) - Veggie chilli (kidney beans & mince) with rice avocado, tortilla chips, tomato & red onion salsa
- In the end, there was a meat-eaters’ revolution, where some people independently ordered in takeaway and eggs / beef mince / pork sausages. The mood at mealtimes did appreciably improve afterwards, which I found frustrating & a bit upsetting even though in a respect positive.
- For the first two meals in particular, people took more than had been planned for and so we ran out of servings.
There was never a shortage of food in aggregate: we had plentiful snacks and also around a dozen ready meals in the fridge
(only two of which were eaten during the retreat), but people did feel upset about this.
- It’s good that we had so many snacks, because seriously running low on food would’ve made people very grumpy and really
should be avoided.
- Unfortunately, the asymmetry between excess & deficit means that it’s optimal to aim for an expected surplus of food, and so you ought to anticipate some going to waste.
Venue and equipment
- It’s extremely difficult to get a feel for a venue without physically being there, so if you’re running an important event
I’d recommend you go visit first. (We didn’t, though I think this was the correct decision.)
- I’d heard the spiel about how important a venue is from the Lighthaven folks before, but
didn’t fully appreciate it previously.
- There were no good communal spaces for people to have 1:1 conversations, and I think this is a big part of the explanation for why so few happened. People could’ve chatted in bedrooms, but were understandably a bit reluctant to.
- One bedroom adjoined directly onto the kitchen, which meant that it couldn’t be used for night-time conversations without disturbing the people (including me) trying to sleep. I could’ve figured this out from the floor plan but it just didn’t occur to me.
- It being the winter, outside was not an attractive location to do things, so everybody was rather cooped-up and the flaws of the house became especially clear. Running this in summer would take off a bit of that pressure, and also deal with the fact that it was just quite cold indoors too.
- I’d heard the spiel about how important a venue is from the Lighthaven folks before, but
didn’t fully appreciate it previously.
- Having access to a car for running errands was incredibly useful.
- People might not bring things that they’d predictably need/want (e.g. ear plugs, antihistamines), and as an organiser it falls to you to go fetch them.
- I ordered about £500 of stationery (flipcharts, whiteboards, post-its), blankets, and musical instruments (harmonicas and kalimbas) from Amazon which failed to arrive. This was nowhere near the catastrophe I would’ve predicted it would be, but I was glad we could go into town to buy some replacement paper & pens.
Activities
- I was personally disappointed that the musical & more creative sessions we had pencilled in didn’t happen,
but by and large people seemed to enjoy the sessions that ran.
- Duncan tried to organise a sea shanties session twice but both attempts were dead in the water; I planned to do karaoke but the market-making game preceding me just had too much demand. The keyboard I brought didn’t get any use, either.
- Only very rarely did we have more than two sessions running simultaneously, which I think was partly to do with the venue size
and partly to do with energy levels being slightly low.
- It was great when we had a swarm of people up at the unconference board sticking up Post-it notes, and ideally it would’ve been like that every time.
How it felt
- The weekend itself was far more tiring & draining than I was expecting, and I didn’t really have the energy to enjoy it as a holiday at all.
My main goal from organising was to get a feel for what running events is like (& to add some excitement into my winter), but if I had primarily been hoping to have a fun trip then it would’ve been far better to arrange something low-key with a handful of friends.- There’s a huge amount of mundane work that needs to happen for things to go smoothly during an event, and most of the time
you just don’t notice it. (I say this as a two-time ESPR alum, now with an even-greater admiration for the ops team.)
- Somebody has to set up the nice tasty snack bar, and clear it away when the space is being used for something else.
- Somebody has to collect up the plates & mugs to stack the dishwasher, and throw away food wrappers scattered around.
- Somebody has to remind whoever is on the cooking rota to be there on time, and not to disappear afterwards before the washing-up is done.
- I think I was doing about 80% of the ops, and this was pretty much a full-time job. Duncan did the bulk of the remainder, and
was spending most of his time running sessions of the sort we really wanted to happen.
- For comparison, ESPR camps have around 50% more attendees and a team of ~6 dedicated ops staff
- So, although I played a couple of games and chatted to people a little, I didn’t fully participate in (let alone run) any sessions or have proper 1:1s.
- There’s a huge amount of mundane work that needs to happen for things to go smoothly during an event, and most of the time
you just don’t notice it. (I say this as a two-time ESPR alum, now with an even-greater admiration for the ops team.)
- I felt a strong sense of responsibility for others’ enjoyment, though I did manage to let go a bit.
- At times when it seemed like people might not be having a good time, I felt physically tense & stressed, and almost embarrassed to show my face (worrying that I was being judged by everybody for things going badly)
- There were a few occasions I went into the living room and saw several people working silently on their laptops – I couldn’t
figure out whether this was what they most wanted to be doing or they just didn’t see anything on they wanted to join.
- As an organiser I felt like I should’ve checked whether they needed anything or wanted help/encouragement to run a session, but I was too self-conscious to actually do that and instead just semi-guiltily walked away, reasoning that there was enough going on that I couldn’t really be at fault if they didn’t have anything they fancied doing.
- I have a tendency to fixate on the parts that (seem to) have gone poorly, and take them as personal criticisms. This was especially true in the case of the meat-eaters’ revolution, because I care a lot about animal welfare and so the gradual erosion of the vegan menu at an event I had brought into existence was emotionally upsetting.
- It was kinda crazy to me that some of the ~24yr-old adults attending didn’t know how to stack a dishwasher / soak a pan / peel an avocado.
- I can get quite curt & impatient when I’m frustrated and feel like others are acting foolishly. Or at least, I end up internally exasperated – I’m not sure how much it shows.
- The food waste that’s inevitably associated with events like this makes me sad, and I didn’t like how some attendees seemed totally unbothered by it. I know that «rationally» the impact of £100 of wasted food is miniscule, but it just feels like the kind of thing you ought to want to avoid, virtuous character-wise.
All of that said, I am glad that I organised Saddle Point. It feels extremely gratifying to know that without me taking the initiative and putting things in motion, none of it would’ve happened. I really like being busy, especially when what I’m doing seems valuable, which the positive feedback helped a lot with.
Other changes for a future iteration
- Run some more getting-to-know you sessions at the start
- Potentially ask everybody to fill out a “Names + Faces” biography + photo slide in advance and get everyone to introduce someone else
- Hotseat in groups of 4-5 on the first night, maybe also 2x pre-planned 1:1s
- Operate at a slightly higher budget and price point
- Find a larger venue, and in particular one that has more nooks & side-rooms for small conversations
- Order in takeaway food rather than cooking, or if budget permits (though I expect it won’t) hire a private chef
- Each meal took around 15 hours of person-time, and also led to a lot of disruption for when sessions would run
- Encourage a norm of not talking about any given topic (e.g., AGI) for more than 20% of the time, borrowed from ESPR’s prohibition on conversational “pits of despair”
- Anticipate that some people will apply late, and factor this in to calculations about the number of people to invite in the first tranche.