It's been a month and a half since I posted The rationalist community's location problem, and there's been a ton of good discussion both in the comments and elsewhere. In this post I hope to summarize the discussion so far, provide additional data, and give my take on what I think we should do.
It's worth noting that there are two related discussions going on in parallel. The first is this one: rationalists as a whole have become less tied down geographically during the pandemic, and it seemed like a good time to reassess whether the Berkeley hub was the optimal setup. The second is internal to MIRI (though it's been posted about publicly a bit) – like everyone else, MIRI has had to experiment with new and nonstandard working setups in 2020, and that's also put them in a place of wanting to reassess whether being in Berkeley is optimal.
Note: I work for MIRI, but these are entirely my own opinions and not MIRI's – I've been thinking about this problem since long before I joined MIRI and don't necessarily see eye-to-eye with most people there on this question.
The discussion so far + data
People have shown the most serious interest in college towns (e.g. Austin, Ann Arbor, Oxford) and several European cities (e.g. Berlin, Tallinn, Prague, London). Options in New Zealand and Australia have been floated but appear to not be under very serious consideration. MIRI's main options under consideration are (1) the Toronto area, or (2) rural New Hampshire, within a couple hours' drive of Boston.
After looking at all the comments I thought, "Wow, there sure are a lot of places people are interested in and a lot of different things they care about. This seems like a great time to use a spreadsheet!" So here it is. You will quickly notice that it is not all the way filled out; this is because it has nearly 1500 cells and I was doing it alone. It is publicly editable, so please contribute a bit of your time to fill out a column or two if you think it's valuable to have this data!
The spreadsheet is informative, but it doesn't make the decision for us. In order for it to be useful, we have to figure out what we value.
The rationalist community is many things, but fundamentally, it exists to carry out the project of rationality, broadly defined. As such, our goal in choosing a location should be to preserve (or improve) the ability of individuals in the community to do productive work on this project.
I've come to believe that the two main cruxes for our ability to advance our goals are:
- Ensuring the existence of a thriving in-person community
- A sense of safety
Apart from those, I think people will have a huge range of idiosyncratic preferences leading to competing access needs. For example, some people really want low cost of living while others really want high salaries, but the two are generally anticorrelated. Or, some people strongly prefer winter to wildfires and some people strongly prefer wildfires to winter. These are considerations that every individual will need to weigh for themselves, since there's no objective right answer. On the other hand, I think the in-person community and safety considerations are more universally important and easier to evaluate objectively.
Habryka suspects that the rationalist community, writ large, basically wouldn’t exist without the Berkeley hub. If we go with that assumption, we need to be really careful about how and whether we move away from Berkeley, because it's very possible that if there's no hub in Berkeley, the in-person community will essentially cease to exist.
(Note: Pre-pandemic, I estimated that there were at least 400 rationalists living in the Bay community. I've seen and participated in several just-for-fun efforts to draw out all the social and professional connections between people in the community, and it always ended up as a connected graph. Point being: That's a lot of people, and you can't neatly remove just a subset of them without disrupting the whole social graph.)
I think that MIRI has been underweighting this concern, given their interest in moving to places like rural NH and Toronto, where it would be hard for others to follow. Whatever MIRI's internal goals, it is de facto very linked to CFAR and LessWrong – two organizations very focused on building and nurturing the rationalist community – and it might be logistically difficult to move MIRI without the other two (e.g. MIRI and CFAR share Anna Salamon, and LW depends on CFAR for much of its admin). Moving CFAR and LW away from the center of the community strikes me as an obviously wrong move. I also think that MIRI itself should want to be where the community is for reasons of recruiting, idea exchange, and social well-being.
Why is it important?
Why is having an in-person community important for the mission? So many reasons, according to me! Here are some bullet points:
- Eliezer wrote in the Sequences about the motivating power of being physically near other people who share your goals and values.
- High-context, face-to-face interactions are really important for exchange of ideas, working together on a team, and founding organizations (see Elizabeth's lit review on distributed teams).
- Talent flow between organizations
- Habryka and I have each worked at four different rationalist/EA orgs (not the same four), a level of mobility only made possible by the critical mass of rationalists in the area.
- Serendipitous encounters
- Habryka points out that it's very unlikely that he would have started LessWrong 2.0 if he hadn't known Vaniver socially. Both Habryka and Vaniver(?) initially moved to the Bay because that's where MIRI/CFAR was. I don't think this is the only such story but it's the only one I feel confident telling.
- Finding close friends and life partners
- Being happy and stable is good for your productivity and motivation, and you gotta have good social relationships in order to be a happy and stable person. (Arguably happiness is intrinsically good as well :P).
So what do I concretely think we should do? Basically, I think if we move away from the Bay, we need to optimize the new location for ease of following. When my housemates and I went through a list of all the rationalists we knew to see how many might follow MIRI to rural New Hampshire, we were only able to think of five, and that seems really quite bad.
Here are some things that I think go into ease of following:
- Dating/friending/employee pool
- If the area is isolated or even just if there isn't a very strong existing intellectual culture there, it will be hard to find friends or people to date. This is a dealbreaker for a lot of people. Those people for whom it's not a dealbreaker will likely be mostly older, married people who are looking to settle down, and I don't think selectively moving all those people away from everyone else would be good for anyone.
- Anywhere with a reasonable pool of intellectual young people would likely meet these needs – i.e. an urban center or a college town.
- Employees are also a big concern. This is a strong argument for being near a tech hub or a tech-focused university. e.g., it's easy to have a continuous influx of people into the community and the organizations in Berkeley, because smart and talented people are always coming to the area for tech jobs and to study at UC Berkeley.
- Job availability – Even if people are willing to move to a particular new location, they need to be able to financially support themselves in order to actually live there.
- Literal availability – Somewhere like Gowrie Park in Tasmania may have many points in its favor, but it probably doesn't have enough jobs for 200 extra people, let alone the kinds of jobs that most rationalists would be looking for.
- Salaries – Many people do long-term financial planning and would not be willing to move somewhere if it meant a major pay cut (or a major tax increase).
- Visas can be a major barrier that will prevent people from following to a non-US location, even if there are well-paying jobs there suited to their skillset. This makes me concerned about MIRI's Canada plan. (Sorry for the US-centric viewpoint; I know it's hard to immigrate here, it's just that so many of us are already here.)
- Lines of retreat
- If MIRI moved to a rural area, I would be quite worried about an 'all-in or all-out dynamic', wherein there would be nothing there for rationalists except MIRI. This means you'd need a high level of commitment to go there at all, and that it could be quite scary to consider leaving. I think this leads to a lot of bad incentives and dynamics.
- A well-put point from Adam Scholl: "The Bay has lots of high-paying jobs (especially for programmers) and hence provides many folks more lines of retreat, I think, than one is likely to find elsewhere. (In the sense of e.g. being able to get a new job without also having to move/get a new social network)."
- This is another point in favor of urban areas.
- Ease of travel to the location
- Most people like being able to see their family and friends, at least once in a while. Many people just like traveling. Sometimes you (either as an organization or as an individual) want people to be able to travel to you. So it would be pretty hard to establish a hub many hours away from an international airport, since traveling from arbitrary other locations would be such a huge hassle. (This is the case for some areas under consideration, though not many). This narrows the search space to Europe and North America, within ~2 hours' drive of an international airport.
- Ideally a person would not need to make an extra special trip to the location of the community – there should at the very least be food and lodging nearby. Also, driving is garbage, and a location that's "within a 30-minute drive of" a place where people might actually want to be is still isolated.
- Modern conveniences, medical care
- Being in an isolated area makes it generally more difficult to pay other people to do tasks for you (e.g. restaurant delivery, grocery delivery, childcare, housekeeping). You also don't want to be too far from good medical care. The latter is only a problem at extremes of isolation (e.g. rural Tasmania, but not e.g. an average US suburb). I think a lot of people value these types of modern conveniences highly, especially insofar as automating or hiring away many daily tasks allows them to be more productive. But I'm sure there are people who think the quiet peace of rural life is worth the trade-off, so this point isn't as strong as the others on this list.
Sense of safety
A sense of safety is fairly necessary for getting productive work done. If you don't feel physically, socially, and financially safe, you're more likely to be anxious and distracted. As an extreme example, you don't want to be a penniless refugee alone in a war zone with no family or friends.
I think people are indeed treating this as a major consideration, tacitly or otherwise. So let's dig into it.
Factors that go into this are crime (particularly but not solely violent crime), laws, and the general feel of the place (maybe this just straightforwardly correlates with crime, I don't know).
One example when it comes to laws is the fact that male sodomy is a crime in Singapore. Singapore has other good things going for it, but we don't want to move somewhere where the gay members of our community will have to worry about being arrested just for living their normal lives. (Note that there's some disagreement as to how much of a problem this would actually be, but in terms of feeling safe I think it matters quite a bit either way.)
As for general feel, I'd count things like street homelessness, trash and feces on the sidewalk, and busy roadways cutting through residential areas as negatives; and things like clean and well-maintained parks, lots of children running around, and people systematically smiling at you on the street as positives.
I think some people probably are attracted to rural living because they like having control over their environment – in that they both have more property and are less restricted in what they can do with it. I can see this contributing to a feeling of physical safety. However, I think buying and building on rural land probably takes quite a lot of time and attention, and after really taking that into consideration I don't feel as excited about it as I initially expected to.
A lot of people's desire to move seems to be motivated by nebulous fear of the US political situation. I am not one of those people, and I'm frankly a little confused about their epistemic state. Despite inside view on the situation, widespread violent political unrest in the US continues to seem incredibly unlikely to me.
It makes sense to me that we might want to move out of the US to a country with generally saner governance. That makes sense to me if you think that political polarization and general societal decline is going to proceed apace in the US, and want to establish yourself somewhere saner and stabler sooner rather than later.
However, that doesn't seem to be the goal – the goal seems to be just "flee the US." And I don't think we're at a point where this is a reasonable goal.
For an interesting example, let's look at Wei Dai's story about how his grandparents ended up trapped in communist China. Very importantly, his grandparents did not fail to see the warning signs of a hostile takeover. They knew things were bad, but they stayed because they hoped that they would be favored under the new system – and got burned when that turned out not to be the case. This just does not apply to our current situation, and it seems really unlikely that it will suddenly become impossible for us to leave, without significantly more warning signs beforehand than we're currently seeing.
Something that's even harder for me to understand is the "flee to a rural area of the US" plan. If we are worried about violent political unrest, it's not clear to me that rural areas are that much safer. Yes they're less densely populated, but they're often very politically polarized and have relatively high rates of gun ownership. Even if we grant that violent political unrest is more likely in cities, the historical likelihood of dying in this manner in the US is statistically zero; and if the situation looks bad, having a car and a passport seems good enough to be able to get you out in plenty of time.
The only situation where you really want to get out of all urban centers is if you're expecting a nuclear attack, which seems vastly less likely than political violence in general. My models here aren't super explicit, but given the fact that despite the entire Cold War the only time nuclear weapons have been deployed against enemies was in 1945, priors on nuclear attack are just incredibly low. (ETA: A commenter makes the reasonable point, "What about anthropics? There've been a lot of close calls.)
For more on the subject of political unrest, see kdbscott's question post from last week.
Social safety is a much tougher nut to crack. At a smaller scale, everything I said above about having a rationalist community with good lines of retreat seems important. But what about social safety at the broader societal level? The prominence of cancel culture and its effects on people such as Robin Hanson and Steve Hsu have put some rationalists on edge, especially those who write (or act) publicly in a way that's traceable to their real identity.
But, is there anywhere, physically, that one can go to escape cancel culture? My instinct is no, but I didn't do a ton of thinking about this. As a toy example of the type of thing that might matter, maybe European universities are less (or more) likely than their American counterparts to fire a professor for saying something that's not PC. Also, I guess cities are maybe worse on the cancel culture dimension because if you're hidden in the middle of nowhere it's harder for people to credibly threaten to physically attack you. But again, this is all very off-the-cuff and not grounded in strong models. I'd welcome other people's thoughts here.
This one's pretty straightforward. All else equal, you want high salaries, low cost of living, low taxes, and good job security and job availability. Obviously many of those trade off against each other, but bottom line, more money is better.
From all this, I think it's pretty clear that we need to be in either a college town or a major metropolitan area (though not necessarily in the heart of one).
My primary recommendation is that the community – and MIRI – should stay in the Bay. This seems like the surest way to not destroy the large (and, pre-pandemic, thriving) in-person community we've built. It's pretty likely we could move MIRI and/or a sizable contingent of people to the South Bay without disrupting things too much, if we wanted to. If some MIRI researchers want a more secluded, less urban lifestyle, they can move to the suburbs (e.g. Moraga, if the bulk of the community stays in Berkeley), or even just North Berkeley. That way they can have a quieter environment but still be within easy commuting distance of everyone else.
If we decide that we need to leave California but not the US – for example, because CA passes laws that make it unfriendly to tech and business, or because wildfire season appears to be getting significantly longer every year – we should move to the Boston/Cambridge area. It of course ticks the boxes of both being in a metropolitan area and being near a good college. MIT is the best technical university in the world by most measures, and the Boston area has plenty of tech jobs. There are trade-offs in terms of climate and costs (e.g. the Bay has higher rents but Boston has higher taxes), and the cultural milieu is certainly quite different, but overall I'm optimistic about the possibility of building a strong community in the Boston area.
If we decide we need to leave the US, I think we should probably move to Oxford. Oxford is home to FHI and Nick Bostrom; many rationalists have spent time there and enjoyed it; and both these things contribute to the fact that it's had a stable x-risk/rationality culture for many years. It also ticks the box of being a university town, and it's an hour's train ride from London (commuting distance by Bay Area standards!), which is not only a huge metropolis but also home to 80,000 Hours.
I have not actually looked into the practical aspects of living in England very much. For example, I don't know how hard it is for US or EU citizens to get work there, and I'm very ignorant of Britain's political situation (beyond having heard of Brexit and Dominic Cummings). Insight into these or any related questions would be much appreciated :)
No matter where MIRI ends up, I think it would be good in the abstract for the community as a whole to pay more attention to and nurture secondary hubs (such as Prague) and enclaves (such as the EA Hotel). I say "it would be good in the abstract", because I don't really have any concrete plans for this. At the very least we should acknowledge that each provides things the Bay can't – e.g. the quiet focus and financial support of the EA Hotel, or how much easier it is for someone in the EU to move to Prague than to the US.
Some examples of beneficial exchange between hubs: CFAR has been running several workshops a year in Prague and training a whole cohort of Czech CFAR instructors. Buck has previously suggested 'EA residencies', where employees from Bay Area orgs go live near different local groups (e.g. Yale EA) for a couple months. Scott Alexander and six other Bay rationalists/EAs took a road trip last year to visit a bunch of East Coast EA groups and SlateStarCodex meetups.
While the pandemic has eliminated most opportunities like these due to the drastic reduction in face-to-face interaction, it's also untethered many people from their physical locations, a change which if it persists will make inter-hub exchange much easier. I also expect that there's plenty of low-hanging fruit in this domain, since not much attention has been put on it up to this point.
Thank you all for reading; please let me know your thoughts in the comments, and help fill in the location spreadsheet if you're so inclined!
 I'm not sure I agree with him there, since if there wasn't one large hub, I can imagine a world where local meetup groups became more important, as in the early days of OBNYC. Lots of places had and still have fairly successful local groups, and they might be even stronger without the "Berkeley brain drain" problem.
 I counted ~30 rationalist group houses in Berkeley and ~8 in San Francisco; the calculation assumes that a group house has an average of 5 people and that about half of people in the community live in group houses. Also, this is probably a low estimate since we sold nearly 300 tickets for the 2019 Winter Solstice. Additional data: Open-invite SSC or LW meetups and parties usually get about 150 people, and it's definitely not the same people every time.
 I did not write this bottom line beforehand – I started this discussion to see what the correct answer was, and I just think this is probably it. I'm pretty sure I dislike living in the Bay more than the average Bay Area rationalist, and I've often personally dreamed of moving. If I could wave a magic wand and move the whole community to Oxford with no roadblocks I probably would. But in the end I value the continued existence of the in-person community more than my own personal feelings about the city.