Thoughts about additional ramifications of this (not optimized much for readability).

Background on Epistemic Effort:

I'm of the belief that if I'm proposing a major idea that I'm hoping people will take action on, I should think seriously about the idea for... N minutes. N varies. But the key is to look into the dark, accounting for positive bias: What the ways an idea might not succeed? What consequences might it have that didn't fit as prettily into the narrative I was crafting?

In my personal experience, this takes something like 30 minutes at least. In my original Epistemic Effort post I suggested 5 minutes, but I've found 15 minutes is barely enough to finish searching through existing thoughts already in the back of my mind. 30 Minutes is how long it takes to get started trying to think multiple steps into the future and general novel concerns.

This process is somehow very different from the process that generates my original blogpost.

I've noticed that now that I know it takes at least 30 minutes, I'm a lot more hesitant to even try to take 5. (I almost went ahead and posted this post without doing so, and then flagging "didn't think for 5 minutes about how it might fail", and then that felt embarrassing and it seemed important enough to do so that I went ahead and did it. But it might bode poorly for the idea)

"What About the Babyeaters?"

A failure mode of canonical Archipelago-ism is altruism + "think of the children." If The Other Group is focused on something actively harmful, and you don't trust people to be able to leave, or you think that the harms take root in childhood before people even have a chance to choose a civilization for themselves (say, secondhand smoke in the privacy of people's own homes, or enforcing strict gender norms from early childhood...

...then the "to each their own" concept advocated here doesn't sound as persuasive.

This is a different issue with Social Subculture Archipelagism. Even if there are no literal children, you may worry about pernicious, harmful ideas taking root that you expect to be memetically successful even though they are dangerous.

It's very conceivable to me that the outcome of a "successful" Archipelagism taking root would be Bad Ideas Winning and the whole thing to end up net negative.

My current take is that some kind of "Bad Ideas Being Successful" thing is likely to happen, but that the overall Net Harm/Good will be positive. I don't really have a justification for that. Just a feeling.

Observations of what's happened so far....

Competing Norms

During the Hufflepuff Unconference, I ran into issues of how norms collide. I wanted people to either firmly commit to coming, or to not come. This failed in a few ways:

• I held the event in a public space known to the surrounding community, which meant people with no idea of the norms ended up coming regardless.
• Som people who were scrupulous and recognized that they couldn't follow the rules chose not to come. Some people who didn't care about respecting the rules just came anyway, creating a mild asshole filter. (I only have evidence of this affecting maybe 2-4 people total in either direction, but it was noticeable)
• People who earnestly wanted to come and follow the rules ran into an issue where other people who weren't firmly committing to things prevented them from making a firm commitment. (i.e. someone had a long lost friend visiting for the week, and the friend wasn't sure of their schedule, and Person-A definitely wanted to make time for their friend if they could, but definitely wanted to come to the Unconference if their friend would be busy).

This last issue eventually resulted in my changing the rules to "please respond with an explicit estimate of how likely you were to come, and some of the decision-relevant things that might affect whether you come or not." I think this worked better.

I don't have evidence that this is that big a problem (I tried one experiment, it didn't work as well as I wanted, I came up with a solution. System working as intended). But it implies future issues that one might not foresee

It's Hard To Make Spaces

I have attempted to create a few spaces (at different times in different cities). And in general, it's harder to create a new space dedicated to a particular thing than I'd have thought (in particular, finding enough people who care about a thing to seriously try out novel norms). In New York, it was hard because there weren't that many people. In the Bay Area, it's been harder-than-I-expected because although there ARE enough people that I expect to flesh out subcultures, those people have more things competing for their attention.

I (currently) expect to be able make things happen, but it won't be as easy as hanging out a shingle. 45 people came to the Hufflepuff Unconference, but I spent 2 months and several blogposts hyping that up. (More recently, I tried to get an Epistemic Unconference happening that'd have a different set of norms, and I couldn't get a critical mass of people interested. I didn't try very hard - it's Solstice Season and I need to conserve my "hey everyone let's all do an effortful thing!" energy for that. But this clarified the degree of difficulty I'd have attracting interest in things)

I expect to have an easier-than-average time getting people interested in things, and it to still require a couple enthusiasm-driving-blogposts per individual thing.

So with that in mind...

Predictions and Gears-Thinking

(IMO this is the hard part of a "think seriously for 30 minutes" thing. This will be most stream-of-conscious-y of the sections)

First, I guess my most obvious prediction is "doing this at all is harder than I was hoping, and barely anything happens."

Futher predictions are sort of weird, since the act of saying some of them out loud might make them come true. (It occurs to me I could secretly make predictions and see if anything happens in a year. I may do that but am not doing it yet)

I notice that the default way my brain is attempting to generate thoughts here feels like the Social Modesty Module running.

The second thing my brain's doing is listing the things I hope will happen, and then see how my internal-surprise-o-meter feels about it.

The third thing, that I will actually record here, is listing things that seem like they might happen, that I want to happen or am afraid might happen, and not list my particular predictions yet but at least get the predictable-in-theory-ideas out there:

How many people will actually attempt to start a subgroup or change norms at one they already control as a result of this blogpost?

How many people will end up involved with those subgroups?

How many groups will happen secretly or privately? How many public?

How many will try experiments past the Valley of Discomfort?

In a year, and in 5 years, how many people will feel that those subgroups were useful?

How many novel social norms will be developed?

How many times do I expect that I'll be surprised by something that happens as a result of this blogpost?

How many times do I expect that I'll be confused by something that happens as a result of this blogpost?

How many attempted social norms will clash in actively bad ways?

Will I end up regretting this blogpost (separate questions for "will I think it turned out not to work but was still the right thing to push for at the time", and "will I think, in principle, that I should have spent my time and social capital doing something else?"

Will people end up more socially isolated, less, or, neutral, as a result of this class of experiment?

(Huh, result of this: "generate hypothesis you can test without stressing about actually deciding on your predictions" was a suprisingly useful technique - I notice with several of the above that I have a least some intuitive sense of how it will play out, and in others, I notice I expect things to fail by default, but that I immediately see ways to make them less failure prone, if I chose to spend the time doing so)

It would be interesting to have a place to collect information about the subgroups. Monthly newsletters from non-secret ones on their progress and lessons learned?

I'd happily write about my group creation leanings.

I am pessimistic about this. Not just because of the potential for Successful Bad Ideas; that's already happened within the rationalist community without an explicit decision to archipelagize (uncontroversially, Leverage, and, more controversially, postrationalism). But also because a fractured community is vulnerable, and I expect this to cause or accelerate fracture. We're already using different norms in different places; Tumblr/Twitter/LessWrong 1.0/Facebook, Bay vs. non-Bay, Berkeley vs. SF vs. South Bay. I don't think this needs more pushes to diverge.

In fact, to the extent I'm working on community issues it's trying to find means of strengthening ties. My pro-epistemic holiday creation project is trying to find things to reinforce the shared features without tying us to any subculture. I think this is more needed at the present moment.

First, thank you so much for writing this!

It is important to notice that the "island" doesn't have to be fully built from start. "Let's start a new subgroup" sounds scary; too much responsibility and possibly not enough status. "Let's have one meeting where we try the norm X and see how it works" sounds much easier; and if it works, people would be more willing to have another meeting like that, possibly leading to the creation of a new community.

I am afraid I can't give this topic the time and energy it deserves, so I will mostly recycle some old thoughts: I believe that a community designed for long-term survival needs to be "eukaryotic" -- to allow subgroups with different ideas of how to achieve the stated common goal, and different levels of commitment; yet treating all subgroup members as valid members of the group. (That means, a subgroup going against the goals of the whole can be excommunicated, just like an individual going against the group. It's just that neither mere membership in a subgroup, nor a lack thereof, is considered a violation of group norms. It is perfectly okay to join one or more subgroups, just like it is perfectly okay to ignore them.)

This is important in long term, because people's capacity to participate in group activities change over time. For example, now that I have kids, I cannot spend the same amount of time on LW as I did before. (Generally, people with kids, but also e.g. currently changing jobs, will opt out of time-intensive group activities. If you kick them out, or just make them feel unwelcome, you are losing their potential contribution in future, when their situation changes again.) On the other hand, people with enough time and lot of agency will feel that more should be done, and you should provide them a valid way to do this, getting some other members involved, but allowing other members to ignore the project. Otherwise, they will probably leave the group and go to some other place where their contributions will be more welcome.

The obvious example of a very-long-living "eukaryotic" organisation is the Catholic Church. On the outside, it has clear boundaries (people generally know whether they are members or nonmembers), but actually very little is expected from the members, other than profess their membership and participate in a few rituals (which is technically costly signalling, but recently the cost is relatively low, e.g. some people just visit the church once in a year on Christmas). People who desire more intense participation can become priests and/or join some internal subgroup such as Jesuits. Both options are valid for a Catholic; the whole organisation works on the assumption that most people will choose the "easy" option and some will choose the "hard" option; without either group the organisation would fall apart. There can be multiple competing internal groups, such as Jesuits and Dominicans, as long as they signal credibly that they are subordinate to the whole.

I hesitate to use Mensa as an example of a successful organisation, but let's admit it has survived a few decades. Mensa is also "eukaryotic"; the officially recognized subgroups are called "special interest groups", and are defined by a shared interest of a few Mensans, e.g. playing chess. (When I was more optimistic about Mensa, I was thinking about creating a rationalist SIG within Mensa.) You can participate in one, or many, or none of the SIGs.

Then I have a personal example of when I was active in the Esperanto community. On outside, it is a community of people speaking the same artificial language; but the problem is that despite the shared language those people often lack a common topic. (Which usually defaults to meta, but using Esperanto to talk about Esperanto and the Esperanto movement can get boring quickly. Except to old people, who often love to talk about the past endlessly.) So with a few friends I founded a subgroup, with the goal to promote Esperanto using internet and electronic media generally. This didn't mean leaving the larger community, nor inviting everyone to the project; we were a well-defined subgroup. We created a few websites and multimedia products, and later merged with another subgroup.

So, if the rationalist community is to exist long enough, it should have similar structure. A clear boundary with simple but clear rules (e.g. "if you believe in horoscopes, you are not a rationalist, no matter how you wish to identify"). Active subgroups. And some authority (e.g. a council of high-status rationalists) that can authoritatively declare the boundaries, admonish and excommunicate heretic subgroups, talk to media, etc. (to prevent "if you say that Deepak Chopra is not an important member of the rationalist movement, that's just your opinion, man").

Dangling sentence at the beginnig:

While I obviously have some specific communities here, and will build off these clai

This is one major advantage of having separate sub-reddits instead of merely seperate tags. It allows each area of discussion to develop its own norms and rules of discussion and is a large part of why Reddit is so vibrant.

Probably the easiest place for many of these groups to develop would be online because it'll be hard to get the numbers elsewhere, but we need the technical infrastructure to make that easy. I mean, I could try creating a group on another site, but people are so much more likely to join when it is right here on Less Wrong.

In particular, thank you for pointing out that in social experiments, phenomenal difficulty is not much Bayesian evidence for ultimate failure.

If you have an event you're running, or an online space that you control, or an organization you run, you can set the norms. Rather than opting-by-default into the generic average norms of your peers, you can say "This is a space specifically for X. If you want to participate, you will need to hold yourself to Y particular standard."

Learning a new set of norms/standards and sticking to them in the right contexts is often not easy. Getting a bunch of other people to choose to do so seems likely to be harder. (Although that’s just my immediate sense of how it is; it may be completely wrong…)

Thus, I have a weak expectation that this will only work well when participants and organizers are fairly good at deliberate standard-following and standard-establishing, respectively. Which actually seems like something the rationalist community is pretty good at, but it might still be worth giving some thought to how to be better at it.

In practice, if you're building an organization, you may not have time to do "proper science" - you may need to get a group working ASAP, and you may need to test a few ideas at once to have a chance at succ

You’re missing the end of a word there.

Communities are crutches - we need them for reasons, and that's why they don't just fall to pieces. You could talk to people who already are doing this. Some of them, at least, will say that they don't like watchers (non-participants). And then how will you observe the thing you wanted to?