Related: On the importance of Less Wrong, or another single conversational locus

One thing that seems to be a pattern across the history of human organizations, projects, and even social scenes is that schism begets schism. In other words, if there is a large and central space, once people start splitting off from it this can often lead to the "floodgates opening" and lots and lots of new groups forming - often in a way that even those who initially wanted to change things dislike!

Perhaps the most obvious example of this is Protestantism. Martin Luther did not want to start his own church, told people not to call themselves "Lutherans", and disagreed quite aggressively with many of those who are now lumped in with him as "Protestant Reformers". However, once challenges to the authority and unity of the Church got started, they were hard to stop, and Luther soon found that those around him had at times gone in directions he did not want -- and now there are over 40 different Lutheran denominations in North America alone, to say nothing of all the other Protestant groups!

However, such a trend is not only limited to religious groups. Political movements sometimes have a similar scenario befall them -- for instance, the Republican side of the Spanish Civil War was substantially harmed by internal schisms, subfactions, and disputes. To use a less consequential example, I've been a part of online communities that have been hurt by repeated schisms over moderation policy -- once people get fed up with the moderation in one place, they start another with much the same purview but different moderators.

Once schism gets going, it can be hard to stop - and once things get split, much of the benefit of a single conversation locus begins to degrade. Indeed, the "LessWrong diaspora" quite harmed this project -- we still haven't fully recovered from having the community split as much as it did, even though things have been improving a bit on that front more recently.

Now, some will say that perhaps splits are good -- perhaps one space isn't right for everyone, and it would be better to have a diverse range of norms that can appeal to different interests. Hence we see things like the archipelago model of community standards, which aim to set up a situation where one broader community contains many subgroups with their own rules and systems.

In practice, though, I claim this doesn't work, because schism begets schism. People say "if you don't like it, go make your own space" -- but they say that because it's an easy dismissal, not because it would actually be better! In point of fact, if everyone who was told such did go and make their own space, the central body would not survive -- "if you don't like it, go make your own" works as a dismissal precisely because it won't be followed! The world where everyone "goes and makes their own" at the drop of a hat is substantially worse and it is substantially harder to form a productive coalition and get things done under those norms.

In point of fact, doing important things often requires coordination, teamwork, and agreeing to compromises. If you insist on everything being exactly your way, you'll have a harder time finding collaborators, and in many cases that will be fatal to a project -- I do not say all, but many. Now, it's possible to get around that by throwing a lot money at the problem -- people will agree to a lot of eccentricities if you pay them enough, as they did with Howard Hughes -- and it's possible to get around that by throwing a lot of charisma at the problem -- Steve Jobs was able to be extremely perfectionist thanks to his personal charisma and (in?)famous "reality distortion field" -- but if those options aren't available, you're going to have to make some compromises, and if the norm is "if the way things are locally doesn't work for you, leave and make a new space!" that's going to be very difficult.

Indeed, once you start allowing this sort of "take my ball and go home" behavior, where does it stop? First you have one person who thinks they are being mistreated, and they go and start their own group to work with their rules. Then they try to enforce their rules, and now they drive someone else off, and so on and so on. Pretty soon you have lots and lots of petty little fiefdoms, each composed of just a few people and none of which are getting all that much done. It is better in my view to try to prevent even the first schism and keep things unified.

Yes, this means you'll have to work with people who don't fully agree with you at times, and yes, this means that there will have to be some agreements on how best to use shared institutions and spaces -- but the way I see it, the history of schisms indicates that that is far better than the alternative!

New to LessWrong?

New Comment
25 comments, sorted by Click to highlight new comments since: Today at 12:50 AM

My two main opinions are:

1. This post seems to take for granted that schisms are bad without actually arguing why. Even if it is the case that schisms beget more schisms, that in itself (or that + pointing at Protestantism) is not actually an explanation of why that's actually bad; it just claims so. It does imply that the badness is in the coordination costs that are increased by a diaspora over a centralised location, but if a diaspora happened in the first place that is strong evidence that whatever central place it spawned from was not only incapable of making this level of coordination happen but also its members judged it was incapable of changing into a place that can do coordination.

2. Two of your main examples are religion and politics, where one can't really belong to multiple subgroups. For things like the LW diaspora or Discord servers or w/e, schisms aren't schismatic – one can belong to multiple social groups and Discord servers and what-have-you (which can even be desirable, as you yourself have argued in your post about diversifying your friendship portfolio).

1. This post seems to take for granted that schisms are bad without actually arguing why. Even if it is the case that schisms beget more schisms, that in itself (or that + pointing at Protestantism) is not actually an explanation of why that's actually bad; it just claims so. It does imply that the badness is in the coordination costs that are increased by a diaspora over a centralised location, but if a diaspora happened in the first place that is strong evidence that whatever central place it spawned from was not only incapable of making this level of coordination happen but also its members judged it was incapable of hanging into a place that can do coordination.

I'm not sure that it's the case that people correctly judged that it was incapable of changing. Reforming an existing space is often harder than just taking your ball and going home, but the benefits are shared across the entire group.

2. Two of your main examples are religion and politics, where one can't really belong to multiple subgroups. For things like the LW diaspora or Discord servers or w/e, schisms aren't schismatic – one can belong to multiple social groups and Discord servers and what-have-you (which can even be desirable, as you yourself have argued in your post about diversifying your friendship portfolio).

Belonging to multiple social groups and Discord servers and the like is nice, but belonging to a bunch of different groups that are all sort of doing fundamentally the same thing (and claim that others are trying to do the same thing but doing it worse) isn't as good. When people found servers that are like "this is the same as <SERVER X> but with better moderation>" that tends to split the community (and indeed things like this have happened multiple times with Discord, online forums, etc.).


I don't like the way this post hints that schism and splitting are the same thing.

E.g. the split between CFAR and MIRI, and the creation of multiple other AI risk organizations (FHI, FLI, BERI, etc), fit the archipelago model, without being schism-like.

Hostility begets hostility, but agreeing to specialize in different things doesn't have the same tendency to beget further specialization.

Fair points. That said, I don't think the split between CFAR and MIRI is a destructive schism, they're still in very close alignment with one another and part of the same broader project. Same for FHI, FLI, BERI, etc. -- but if someone had founded a "new MIRI" saying that MIRI was failing and their paradigms were fundamentally destructive and everyone should withdraw their support of MIRI and back the new organization, that would be schismatic in the way I warn about.

How do you square this with the other post you made today about how if you don't like the way a group is going, you should spend more time with a different group? Isn't it true that if everyone followed your advice from the "Diversify Your Friendship Portfolio" post, any tendency a group took which some people didn't like would cause those people to spend less time with the group, which would reinforce that tendency (since the people who dislike it are no longer around), which would eventually result some form of a schism?

(I'm not saying I have a solution to this. It appears to be a sort of collective action/public goods problem among folks which don't like the new tendency.)

Another consideration is that people splitting off become explorers. If they don’t realize that, they’re very likely to fail or not to go very far. And I’d say overall explorers are very valuable. But if everyone is one, then that doesn’t work.

In point of fact, doing important things often requires coordination, teamwork, and agreeing to compromises. If you insist on everything being exactly your way, you'll have a harder time finding collaborators, and in many cases that will be fatal to a project -- I do not say all, but many.

This is true and important, and the same or a very similar point to the one made in Your Price for Joining.

But that post has a different standard than the one given by the OP:

If the issue isn't worth your personally fixing by however much effort it takes, and it doesn't arise from outright bad faith, it's not worth refusing to contribute your efforts to a cause you deem worthwhile. [emphasis mine]

Sometimes things are bad or (or much worse than they could be) in some group or community. When that's the case, one can 1) try and change the community from the inside, or 2) get a group of his/her friends together to do [thing] the way they think they should do it, or 3) give up and accept the current situation.

When you're willing to put in the work to make 2 happen, it sometimes results in a new healthier group. If (some) onlookers can distinguish between better and worse on the relevant axis, it will attract new members.

It seems to me that taking option 2, instead of option 1, is cooperative. You leave the other group doing it their way, in peace, and also create something good in the world in addition.

Granted, I think the situation may be importantly different in online communities, specifically because the activation energy for setting up a new online group is comparatively small. In that case, it is too easy to found a new group, and accordingly they splinter to regularly for any single group to be good.

Sometimes things are bad or (or much worse than they could be) in some group or community. When that's the case, one can 1) try and change the community from the inside, or 2) get a group of his/her friends together to do [thing] the way they think they should do it, or 3) give up and accept the current situation.
When you're willing to put in the work to make 2 happen, it sometimes results in a new healthier group. If (some) onlookers can distinguish between better and worse on the relevant axis, it will attract new members.
It seems to me that taking option 2, instead of option 1, is cooperative. You leave the other group doing it their way, in peace, and also create something good in the world in addition.

I agree that option 2 can be cooperative, but I want to point out that taking option 1 is also cooperative. If the other group or community is, as you say, much worse than it could be, helping to improve it from the inside makes things better for the people already involved, while going and starting your own group might leave them in the lurch. In general I think you should probably at least initially try to reform things, though if it doesn't work well there's a point where you might have to say "sorry, the time has come, we're making our own group now".

Granted, I think the situation may be importantly different in online communities, specifically because the activation energy for setting up a new online group is comparatively small. In that case, it is too easy to found a new group, and accordingly they splinter to regularly for any single group to be good.

Yeah, I think online the cost of just creating another site is importantly too low. On Discord it takes like 10 seconds to make a new server, these days you can set up a basic web forum very quickly without even having to pay for hosting, and so on. In real life it's harder to create new organizations, events, etc. in a way that can actually help avoid splitting communities.

If the other group or community is, as you say, much worse than it could be, helping to improve it from the inside makes things better for the people already involved, while going and starting your own group might leave them in the lurch.

Sure. When everyone (or at least a majority) in the initial group are on board with your reform efforts, you should often try to reform the group. But very often there will be a conflict of visions or a conflict of interests.

In general I think you should probably at least initially try to reform things, though if it doesn't work well there's a point where you might have to say "sorry, the time has come, we're making our own group now".

I certainly agree with this, though it seems plausible that we have different views of the point at which you should switch to the "found a splinter group" strategy.

One thing I find interesting, as an example that may be particularly pertinent to some rationalists, is how effective altruism has, in spite of everything else, been robust to the kinds of schisms you're talking about. In spite of all the differences between different factions of EA, it remains a grand coalition/alliance (of a sort). Each of the following subgroups of EA, usually built around a specific, preferred cause, in total has at a few hundred if not a couple thousand adherents in EA, and I expect would each be able to command millions of dollars in donations to their preferred charities each year, including:

  • high-impact/evidence-based global poverty alleviation (aka global health and development)
  • AI risk/alignment/safety
  • existential risk reduction (inclusive of AI risk as a distinct and primary subgroup, but focused on other potential x-risks as well)
  • effective animal advocacy (focused on farm animal welfare)
  • reducing wild animal suffering (focused on wild animal welfare)
  • rationality
  • transhumanism

While none of these subgroups of EA is wholly within EA, it's very possible the majority of members of these communities also identifies as part of the EA community as well. An easy explanation is that everyone is sticking around for the Open Phil bucks, or the chance of receiving Open Phil bucks in the future, as a cause area's increased prominence in EA is moderately-to-highly correlated with them receiving => $10^7/year within a few years, when before each area's annual funding was probably <= $10^5. Yet there isn't a guarantee, and the barriers to access to these resources has been such that I've seen multiple of these subgroups openly and seriously consider splitting with EA. If any or all of these causes could sustain and grow themselves such that one or more of them might do better by investing its own resources into growing outside of EA, and securing its independence. However, as far as I can tell, there has never been a single, whole cause area of EA that has 'exited' the community. As the movement has existed for ~10 years, it seems unlikely that this would be the case if there wasn't other factors contributing to the cohesion of such otherwise disparate groups.

This is very close to on how on magic color philosophy white sees red as dangerous, that order is required to keep chaos at bay. And lot of the of the standard counterarguments apply, red can argue that compromise stiffles freedom and having a to find a single solution that works for everyone takes forever compared to making quick local solutions to the persons that are affected by them.

For example it's probably not a error state that the globe has multiple states in it instead of a single super country. Instead of intolerable chaos it's more like diversity to be celebrated.

There is the argument that if you are a pack hunter then you need to be compromising enough to be able to form the hunt group instead of everyone hunting for themselfs. But then again people form villages and instead of doing a single super village around 150 they split of into multiple independent groups (this hints at hidden costs of unity).

Unity is not an unconditional requirement. Some things need / benefit from unity but the need is often finite or only in regards to some aspect. And it might be possible to solve the same need with some other method rather than unity.

I'm very "MtG white", yes. (Well, white-blue, but yeah.) :P

It seems like the Martin Luther case is an example of open disagreement begetting schism. If LessWrong can't deal with open disagreement, what's it even doing?

There's a significant difference between "open disagreement" and "open disagreement, refusal to cooperate with the mechanisms for settling this, denunciation of the legitimacy of authority, and calling for others to leave the group". LessWrong can deal with open disagreement, but if the front page constantly had people denouncing the moderators and calling others to go and join their new community that would be something else.

Basically agree, but for every group there’s probably some size / some disagreement where it would be better off splitting. So may be I’d rephrase this as people / groups have a modern bias for splitting too early.

I'd love to see some quantification or heuristics for "what is the right number of groups", or "when is a split justified". Or the reverse: what's the cost function for splitting too early or too late?

I object to the framing "take my ball and go home" - at least in voluntary discussion groups a better metaphor is "Keep your ball, enjoy your game - I'm going to play this other game for awhile".

I'm not sure this generalization applies very well to groups that don't claim exclusive access to their members. Religions, states, and "movements" are generally somewhat jealous of their members' loyalties, and object to having them split.

I can be a member of many elements of the rationalist diaspora, and I can ignore those I like. This seems strictly superior to trying to force them all into one set of standards and expectations.

It may be superior for the individual, but it is often worse for the group's ability to coordinate and get stuff done. Similarly, when the diaspora first started, many people had the choice of whether to continue posting on LW or whether to create their own blog, gaining personal status/influence but damaging the unity of the group. The latter broadly won out, to the detriment of the project as a whole.

Now obviously making your own blog wasn't just defecting - there were serious issues with LessWrong's culture and standards that made posting there feel like a chore. But ideally we would have fixed that such that the locally incentivized behavior was that which was better for the project as a whole rather than that which helped the individual at the cost of the group. Sadly, we missed the opportunity, at least when the issue first came up.

It may be superior for the individual, but it is often worse for the group's ability to coordinate and get stuff done.

This may be a crux for our disagreement. I deny that the group has any moral weight, except as a sum of individual experiences. If it helps the individuals to find groups that fit better so they can coordinate better, that's good.

Now you can certainly argue that some leavers are making a tactical mistake, and they would be better off staying. I don't disagree, but I want to acknowledge that the opposite error also happens (staying when one should splinter), and it's extremely idiosyncratic which one to weigh most heavily in any situation.

I deny that the group has any moral weight, except as a sum of individual experiences. If it helps the individuals to find groups that fit better so they can coordinate better, that's good.

What about as a coordination mechanism that can make things better for individuals in the long term?

What about as a coordination mechanism that can make things better for individuals in the long term?

I think that falls under the "made a mistake" heading. The individuals are incorrect in thinking they're better off with a different set of groups and coordination mechanisms.

I do think that sometimes people stay in things that they should split off from. However, I've noticed that locally there seems to be a lot of praise for the "archipelago" and the like, and also that the community seems to have been seriously damaged by splintering too much and losing its unity and sense of shared progress. I think these things are connected and that at least around here, people should be more wary about splitting off than they are by default.

But ideally we would have fixed that

Was the splinter issue more one of 1) people going off and starting a new place by themself, as opposed to 2) bringing people with them/trying to bring everyone with them? (If an entire group moves to a new place, and starts over and it works, this seems like the same result as if they'd stayed in the same place and fixed things. I.E. A hundred islands is worse than ten.)

I was thinking about something similar the other day. I was wondering if, from a historical perspective, it would be valid to look not just specific sects, but all Abrahamic religions, as 'schisms' from the original Judaism. One thing is that religious studies scholars and historians may see transformation of one sect into an unambiguously distinct religion as more of an 'evolution', like speciation in biology, than 'schisms', as we typically think of them in human societies.

I think of the change from Judaism to Christianity as too significant to be viewed as merely a schism. Similarly, it would seem strange to classify Islam or Mormonism as "Christian schisms" in the same way that one would classify, say, the Old Catholic Church as a schism from the main Catholic Church -- it's certainly true that Islam and Mormonism both take Christianity and then add a new prophet and his book on top of it, but that seems too significant to qualify as just a schism. To me, schisms are often notable for the relatively small nature of the differences that they are splitting over, and by the time you're adding new holy books and substantially reinterpreting the past teachings you've gone beyond that phase.

(To give a nonreligious example, I would say "we're going to make a new forum with exactly the same purview, target audience, and board structure as the old forum but with different moderators" is a schism, while "we're going to make a new forum that addresses substantially different topics while still including some of the old stuff" is not.)