Feb 08, 2018
Disclaimer 1: I take responsibility for all opinions expressed here and note that they do not necessarily reflect the views of the people I interviewed or of my various employers.
Disclaimer 2: In this post I sometimes make claims on behalf of LW2. I am under the impression that this is mostly okay, but I want to clarify that I don't work for LessWrong.
LessWrong meetups have been taking place for almost a decade, and CFAR has a long history of encouraging people to start their own groups. But despite their popularity and staying power in the rationalist community, no one has yet solved meetups beyond the object level - that is, I haven't seen anyone give explicit models of what meetup groups are supposed to accomplish, nor even systematically examine what they've accomplished in the past.
I became interested in this a year ago, when I started my own meetup group and found that I was basically completely directionless, having not had much of a goal beyond 'start a meetup group'. In the intervening months, I’ve become increasingly interested in the theoretical underpinnings of the whole endeavor we call 'meetups', so in an attempt to figure out what the heck is going on, I interviewed members and organizers of eight rationality meetup groups from around the world and collected anecdata from dozens more.
I embarked on this project with the goal of answering two overarching questions:
I feel that I've gone some of the way towards answering both of these questions, but that I'm definitely not all the way there yet. Though I think this write-up captures the most important parts of what I've learned, there is a lot of other potentially valuable information that I left out for the sake of brevity, so if you think any of my claims seem unsubstantiated, I encourage you to ask for further details.
I intend to turn this post into a mostly static reference document for organizers, but there are hundreds of perspectives that I was not able to capture here, particularly from people who have many years of experience with this issue, so it’s far from complete. I would be very interested to hear from people with experience running or attending meetups who disagree with any of my claims, think I’ve missed something critically important, or have other suggestions on how this can be improved.
Note: Because I'm trying to answer two separate questions, the following document contains both object-level advice for meetup organizers and higher-level discussion. Apologies if this is displeasing.
As I discovered firsthand, the purpose of a rationality meetup group is not always immediately clear, and different groups may serve very different purposes. It’s important to figure out what the purpose of your particular group is, because you can’t optimize without something to optimize for, and if you don't decide on a purpose ahead of time your group will be pulled in many contradictory directions and probably end up ineffectual (h/t spiralingintocontrol).
So decide: Are you there to learn about the world? To become better at thinking, or to help others become better at thinking? To make progress towards your goals using applied rationality? To just hang out with like-minded people? And are some of these goals better than others – either in the sense that they will produce stabler groups, or in the sense that groups with these goals will have more impact on rationality and existential risk?
It seems that different groups of people may just want different things out of a meetup - and that’s fine, insofar as meetup groups are a social thing that exist primarily to serve the needs of their members. But I don’t work on meetups just because I want people to have friends (although that’s definitely a nice side-effect); I work on them because it seems that they have historically been able to produce people and outputs that have maybe marginally contributed to us being less likely to go extinct within the next couple of decades. So let’s try to figure that out.
I asked people about the counterfactual impact that participating in or running a meetup group has had on their lives, and more generally, what value they think their meetup creates. I list some of the most significant answers below.
Creating new rationalists/EAs
I will define this category to include people who became actively involved with the community and the ideas who:
It is unclear how common each of these scenarios is, but it seems plausible that this might account for a large part of the value of meetups, so long as the group retains the rationalists that it creates and guides them along the right path.
Retaining existing rationalists/EAs and helping them grow
Speaking from experience, rationality is something it’s pretty hard and sad to do alone, and it can be easy to lose sight of your motivation. People report that being part of a meetup group has made them more engaged with the rationality material, and has helped them sustain motivation for rationality/EA just by virtue of being around people who are doing the same thing.
For some people, being part of a meetup group also gives them space to make personal progress, either by regularly using applied rationality in a structured context, or just by being around people with similar cognitive styles who are motivated to help each other improve.
For organizers in particular, running a meetup gives them quick feedback loops on coming up with ideas, and can also help them get better at social modeling.
Sending people to CFAR workshops
Meetups send a lot of counterfactual people to CFAR workshops, which is probably good for their personal growth, increases their engagement with the community, and may cause some of them to work on existential risk.
Moving to Major Hubs
I know meetups causing people to move to the Bay is a controversial topic, but from my perspective, moving to a major hub is one of the best things a person can do in terms of expected impact on the existential risk landscape. It gives people the opportunity to work at aligned organizations, and to be around hundreds of like-minded people, which (in addition to its social benefits) allows people to find collaborators with whom to start new projects and organizations.
Some local groups have projects or reading groups dedicated to learning about problems in AI alignment. I do not know enough about these groups to guess at their impact.
Due to different groups having different purposes, a lot of elements are highly variable, such that one group may have completely opposite needs and preferences to another along some axes. The following are things that seemed to be critically important for all groups, regardless of how little else they had in common.
At least one organizer who is very committed and cares a lot
It seems pretty clear that no meetup group could exist at all without the efforts of one committed person. But while just one person is sufficient to get a group started, a group that relies entirely on the efforts of one person is in a very precarious position. Almost all successful meetups have at least two people in organizer-type positions, which makes the group much more robust to organizers’ personal setbacks.
As a meetup organizer, you should identify someone to be your second in command, who can run the meetup for a week if you’re sick or away. You should also try to give anyone who will listen the opportunity to take a leadership role of some sort. In addition to reducing your workload (and therefore reducing the chance that you will burn out), handing off meaningful responsibilities to other people makes them more invested, strengthening the group as a whole.
A successful organizer should…
It is important for someone to be explicitly in charge of the group, so that that person has the authority to make unilateral changes when needed, to steer discussions, to address interpersonal conflicts before they blow up, etc. By default all such responsibilities fall to the meetup organizer, but this often fails because since they are in charge ‘by default’, organizers often don’t feel that they have much real authority.
The Seattle community solves this by having people with designated roles, such as Moderator or Welcomer. The New Hampshire community, which is much smaller, has a unanimously elected dictator.
It is very important for the long-term health of a meetup that those in charge be able to deal effectively with people who create problems, which may include asking them to leave the group. This issue came up in almost every single interview I conducted and has been raised elsewhere as well. Most organizers struggle with this because it feels mean and wrong to exclude people, and they often feel like they don’t have the power or right to do so because meetups are not a formal institution.
Why it’s important to eject some people:
Three classes of solutions to this kind of problem:
Intentionality of discussions
It is easy for a formal meetup to be wasted if no one has control of the conversation. You need an official moderator, someone who has the authority to interject if things are going off the rails and steer the discussion back in a more productive direction. By default, moderation falls to the organizer, but moderators can also be designated separately, as they are in the Seattle community.
One thing all successful long-term meetup groups have in common is that they often hang out together outside of the scheduled meetup time, or even live together. It can be useful to foster these friendships explicitly, especially if you’re the organizer. Go out of your way to set up one-on-one meetings or hangouts with new members, as well as making sure to talk to them at informal events.
Hold a combination of structured and unstructured meetups
If your meetup has only formal events, you are less likely to develop strong friendships with the other members and will probably miss out on a lot of interesting and valuable conversations you could have had in an unstructured context. On the other hand, if your meetup has only informal events, the group will probably not accomplish much, and you may find retention difficult.
Models for accomplishing this:
The Seattle Rationality Reading Group’s formal meetings are always followed by dinner. Other groups alternate week to week between formal/teaching events, and more casual social events. You can also mix it up by having events like Petrov Day (this one can be automated!) or Secular Solstice (this one can’t!) that are structured but also are fun, encourage personal bonding, and remind you what we’re fighting for.
It seems almost-tautologically true that any value that comes out of a meetup is due to the people involved. Members of a meetup should inspire each other and help each other grow. A meetup will most likely produce nothing of value if it doesn’t have any members who Actually Try, and meetups comprised of people who don't like or inspire one another are unlikely to last.
Other helpful but non-critical factors
At the beginning of any group’s existence, it is important that it be able to attract enough new members to establish a core set of people who will reliably show up, or the group will die. However, this is not a first-order factor, because once a group reaches a stable point, it does not need to constantly bring in new members in order to survive.
Having multiple groups
It seems to help stability to have multiple groups with different purposes and partial (but not complete) overlap in membership, because these different groups attract different members, making for a larger, more diverse community. For example, Berlin has a rationality dojo, a sequences reading group, social meetups, and circling, and used to have a machine learning reading group as well. These are not all run by the same person, which means that the community is not as vulnerable to the loss of one organizer, and all of the groups have slightly different cultures. This spread of groups is also good because each group has an explicit purpose.
There is an obvious catch-22 here where in order to have multiple stable groups, you need to already have a large enough community and good enough infrastructure to sustain multiple groups. However, despite the fact that only a community that is already pretty stable will have multiple groups, it does seem that the existence of multiple groups in turn works to strengthen the community.
Contact with other meetup groups
Meetup groups in Europe have benefitted from cooperation and collaboration, both at the annual LessWrong community weekend and in one-on-one group cooperation. I am not sure what the nature of the collaboration has been or what benefits have come of it, but more than one organizer has found collaboration valuable and has expressed the desire for more of it.
What things can meetups create, beyond what they already have?
I did not explore this question at all, but it seems important to answer for the purposes of people and organizations that encourage the creation of meetup groups with the goal of mitigating existential risk.
Where does most of the value of a meetup come from?
Anecdotally, it’s possible that attending just one or two meetups can light a fire under someone, and after that their agency is activated enough that they quit their soulless day job and move across the country to work for a rationalist organization. So sometimes the value is heavily front-loaded, because sometimes it’s enough to just be told what’s out there.
Other people stay with a community for many years before moving on, or are happy to just have their local community and have no intention of moving on at all. These people are probably getting something very different from meetups than the people in the previous paragraph. There are also all sorts of people between these extremes.
What would a successful system for running meetups centrally look like?
My goal is to make running meetups smoother and easier for anyone who is interested in doing so, but it’s very hard to know how to accomplish that. With the SlateStarCodex meetups, I tried to make it so that there was only one point of failure instead of dozens, by having the buck stop with me. However, this didn’t work very well because I was still relying on the other individual meetup organizers to respond to my requests for information, and also from the outside view this system was extremely unstable.
My ideas on this are not fleshed out, but I can imagine centralized coordination looking similar to other accountability systems - we could designate people within groups to hold one another accountable, or design incentive systems at a higher level. It is also possible we could help remotely with some steps of the process, possibly in a way that’s similar to what the Center for Effective Altruism is attempting to do with local EA groups.
Rationalist group houses
What does a single group house do for a community?
Non-Bay Area rationalist houses are not common enough to be easy to gather data for, but I would be interested to know how rationalist group houses interact with and change their local communities, as well as how, why, and when they arise.
Creating a meetups system on LessWrong 2.0
The new meetups system will be much more functional and easier to use than the system on the old LessWrong, and will aim to address the key points of failure in the system that have plagued meetup groups striving for visibility in the past. This will include the ability to set up recurring events, regular automated check-ins to ensure that the information is up-to-date, and a map that isn’t terrible, among many other features.
Centralized sharing of resources
It would be helpful to meetup organizers – especially new organizers or people considering starting a meetup – to have centralized resources that tell them what to do and what to expect. Currently, the resources for running a meetup group are scattered across the LessWrong wiki, the closed LW Organizers Facebook group, and various other private channels, making them much less useful than they could be. We hope to aggregate these into a centralized location on the as-yet-non-existent LessWrong community page.
In addition to static resources, there will be a section on LessWrong dedicated solely to the discussion of meetups. This will solve the ‘people want to talk about this but no one really knows where to write it or where to find it’ problem. Discussion that people don’t want to be public can still take place on the LW Organizers Facebook group.
Several organizers expressed the desire for some sort of lesson plans or ‘rationality modules’ they could present, because finding the time and mental capacity to create an activity or lesson plan every single week can be difficult and stressful, especially if you also have a full-time job.
The High Impact Network tried to create EA modules in the past and I've been told it didn’t go very well, but I know that a few meetup organizers have already created some materials that would help with this, so I haven't given up on the idea. I’m not yet sure how modules will be integrated into the meetups system, if at all.