Released: Feb 4 2022 (ongoing, incomplete thoughts, most recent update January 8, 2024)

The following is a template I've used occasionally to run workshops. It's far from perfect, and I often deviate a lot from this pattern. Still, I hope there are some ideas others find useful. Thoughtful suggestions for improvement or variation are welcome.

People: If wonderful people attend, and you create an atmosphere where they're comfortable sharing, it'll work. An ideal workshop is a group of people talking excitedly with a roof. And the roof is optional.

Only invite people you admire so much you want to hear them give a talk: I love emailing the entire attendee list and saying "If you'd like to speak, then I'd love for you to do so. But also: there's absolutely no requirement to speak". When you email everyone like this it creates common knowledge: you think they're all amazing with something huge to contribute (but with no obligation). This tells people both how much you admire them individually, and also how much you admire all the other attendees. They all belong there. This is especially important for junior people and people who may feel like outsiders. With the time structure described below, this invitation-to-speak limits you to about 20-25 attendees. Usually I find about 3/4 of people want to speak (with quite a bit of variation).

Do not, do not, do not invite people just because they're senior or you feel some obligation: Again: are you excited to hear them speak? If not, don't invite them. I don't care if they won the Nobel Prize. The anti-pattern here is: "We should invite so-and-so, because they're part of the crowd". It's understandable. But you are compromising everyone else's experience this way. That said: 20-25 people is not a lot, and sometimes you'll hurt the feelings of people you like and admire (a common pattern: you're inviting ~6 people from four communities, but you happen to know ~30 people in one of those communities). This is challenging to deal with. I know I sometimes compromise here, but I try to be pretty hard core about not doing so. I know I've made some people very angry by not inviting them to certain events they felt entitled to attend. That's a pity, and I certainly don't take any personal joy from it (quite the reverse), but: I made those choices to serve the event better.

Intertwining: A pattern I love – though it can be scary to carry out – is to bring together people and communities who should know each other but don't yet. I think of it as releasing latent potential. Ideally that latent potential is enormous. When you bring a pre-existing community together, the outcome is often fairly predictable. When you introduce communities or networks to one another, the unpredictable happens. I am very explicit about posing and answering these question, both with myself and with other organizers: which communities are we bringing together, and why? This is genuinely scary: an hour before the workshop starts I'm often in a state of fear, thinking "but what if they have nothing to say to one another? what if they hate each other? what if they're bored?" But instead, what (thankfully) has always happened is an outpouring of energy and excitement. Invite curious, generative, generous people, people who will excite and enable and amaze one another. No assholes¹; be very careful with the censorious; be very careful to observe if anyone is uncomfortable; and act swiftly if anyone is making others feel uncomfortable.

Time structure: Friday night and all day Saturday. I tend to start around 6pm Friday. An hour to mingle, for people who know each other to get reacquainted, and for people to begin meeting new people. Opening remarks + 2 talks takes us through to 8:30-9:00pm. Food tends to be nibblies before and maybe after, not a full meal (this keeps people standing and mingling, rather than seated). The sign that you've gotten the right group together: they won't want to go home. On some memorable occasions most attendees have still been chatting at 2am, and I've had to throw them out. Re-start no later than 9am. At least 90 minutes for lunch, 2 hours is better, plus at least two half hour breaks. I tend to leave people to themselves on Saturday night; depending on the mood, a large subgroup may go together for dinner.

Weekends: That said, weekday or weekend is a hard call. There are complicated issues with both. Know and be sensitive to the fact that whichever you choose will make it difficult for some people to attend. In general, try to be mindful of and sensitive to people's needs. You won't be perfect, but the effort can greatly improve things. Things to think about: differing income levels; the impositions of travel, especially from distant lands; family commitments; childcare; language barriers; alcohol; food needs; accessibility in general; religious commitments; age; physical needs; and many more.

Try to begin with two really eye-opening talks, out of left field: Ideally, people should sit up and think "my goodness, I'd never seen anything like that before". It's tough to source these. But it's worth working very hard to do so. When it succeeds it creates immense excitement, a sense that this is different, and a novel context which benefits everything else that happens. Note that the anti-pattern here is to start with the talks everyone is pretty much expecting. This is very common, and a genuine anti-pattern, putting everyone into a business-as-usual mode. You want people to feel that they are in new creative territory, not business-as-usual.

Talks: Many people will say "Oh, there should be less time on talks, more time to mingle". This misdiagnoses the problem. My preference is to ask people to prepare 15-20 minutes of talk, but expect to hold the floor for 30-40 minutes, with questions and discussion intermingled the whole way. When someone has spent years doing something to the best of their ability, I want them to have the floor to talk in-depth. You develop a deep, rich shared context that is then the foundation for wonderful in-person conversations. You can seed the intermingled questions by asking a few yourself early on; try to model good questioning, but not dominate. It's especially good to ask humble questions, revealing ignorance, or uncertainty about fundamental issues.

No swag: Part of me is just repulsed by the waste and excess. Also: if swag is important to an attendee, they're often not a person you really want there.

Prioritize making it easy to run workshops at the drop of a hat: Organizers often put a huge amount of effort into irrelevant minor details. Again, what matters is: getting wonderful people to attend, and creating an atmosphere where they feel comfortable openly sharing their understanding. If you can do this, nothing much else will matter. In some sense, a workshop is just a room, a well-crafted email invitation, and a list of great people to invite.

Work hard on the invitation email: What you want is a sense that this will be a special and memorable event; you also want to make clear what special thing the person you're inviting will contribute, and what special thing they will get out of it. I will sometimes go through a dozen drafts, trying to find just the right thing. Sometimes the result will just be one or two sentences, but they're carefully chosen to be unique and (I hope) striking. If you can't do this, then it's worth reflecting on why you want to run the workshop.

Opening and closing words: Organizers often drone on. Do thank and single out the people who made the event happen. And if you have something genuinely thoughtful and heart-felt and unusual to say, say it: people will be very responsive and thoughtful in response. But don't just speak to feel important, or because you feel you should.

Location: Find somewhere immersive, where people will enjoy themselves, feel welcomed, and little need to leave. A comfy home is good. Any space with special qualities is good: a thoughtful collection of books, or artwork. Unremarkable offices are not so good, though if that's all you have available, go for it – space shouldn't be a barrier. You can run a workshop in a tin shed.

Variations: Obviously, this is just one template. Some of the ideas carry over to different structures. Just three observations: (1) if you go much beyond 24 hours, increase the density of free and unstructured time; (2) changes in the number of people qualitatively change the nature of the event (with transitions around 10, 25, 60, 100, and 250); (3) increasing the length qualitatively changes the nature of an event (week-long, month-long, and 6 month-long events have very different capabilities).

Footnotes