I think it's going to have to be a structural change. Encouraging exit from social media (which I've just done myself) and building vibrant discussion communities (e.g. increasing engagement at LessWrong itself) seem like good moves.
"Everyone on this site obviously has an interest in being, on a personal level, more rational."
Not in my experience. In fact, I was downvoted and harshly criticized for expressing confusion at gwern posting on this site and yet having no apparent interest in being rational.
The problem is with incentives.
angry people partitioned into angry groups, yelling at each other and confirming their own biases
A part of this is irrationality, which could be fixed by teaching rationality, at least to people with sufficiently high IQ, somehow, maybe...
But another part is rewards. As you said, the shouty passionate people tend to dominate. And that doesn't only mean winning an abstract argument, but also... getting power, getting resources, getting laid.
Without changing the reward system -- without dramatically reducing the rewards for (the right kind of) insanity -- people are not going to embrace sanity. Why should they?
On the other hand, the problem is partially solving itself. For example, and average lobbyist is probably more sane than an average politically mindkilled person screaming in the mob... and the average lobbyist also happens to have more power. Yeah, this solution is far from perfect, but it still works better than having everything decided by the best screamer and their followers.
An alternative solution seems to be democracy all the way down, like in Switzerland. If you give people opportunity to vote about local things, if they fuck up something, they are personally going to suffer, but they are probably going to survive it, and the next time they may be smarter. On the other hand, if you only give them voice on national scale, when they fuck up something, everyone is going to suffer, and then someone else will try to fix the problem -- so the feedback is not targeted on those who need it most, and the stakes are too high. When you think about it, it's insane that we have people who most of the time have zero power, and then once in a few years we give them power to turn the whole country upside-down; how are they supposed to learn their power responsibly? (Most people can't think rationally, but most of them are able to pattern-match to things that burned them in the past.)
Rational debate is nice -- I share your aesthetics -- but what is the incentive to join it?
Rationally designed institutions would make people first prove themselves on a smaller scale, and only then give (some of them?) power on a larger scale. Then we would still have many technical problems to solve (such as how to prevent people cheating, how to prevent them achieving short-term success with long-term failure, etc.), but at least we would go approximately the right direction. Feedback, and practice. Existing examples are: Switzerland-style democracy, army, meritocratic open-source software project, and to some degree lobbyism. (I am not endorsing any of them specifically, just pointing towards the thing they have in common.)
When designing your own institution -- even if it is a local LW group (and perhaps especially there; where else should we practice our art of rationality?) -- you should consider this. Make people prove themselves in small things before you give them a voice in large things. Somehow. For example, create two types of membership: the inner circle which either prepares the meetup room and cleans it afterwards, or prepares lectures on topics interesting to other people; and the outer circle which has no duties, and also no voting rights. People will willingly sort themselves out. (Related: "Require contributions in advance".)
This may feel unfair. But giving people who don't give a fuck about something (and therefore will optimize for something else, such as being popular or getting lulz on expense of someone else) equal decision power, is insane. Especially if those people will make decisions according to their whims or applause lights, and then someone else will have to do the hard work anyway.
The unfairness happens when people get locked out of this system artificially. For example, if you set up the system so that in theory "by proving yourself in small things you will be given power in greater things", but in practice some people are actually not allowed to prove themselves in the small things. -- For example, if someone who personally hates you, also happens to evaluate how you do the small things, and fails you regardless of your work. Or if doing the small things also requires resources that you don't have (and instead of "this guy didn't have resources to try doing the small things" the system labels you "this guy failed at the small things"). -- If you care about fairness, fix this, but don't throw away the whole idea of rewarding responsibility. And yeah, it is difficult to do this the fair way. Some people have genuine disadvantages; some people only pretend to have disadvantages to avoid having to do their homework; etc. You will never be 100% sure. Still, on average this system works; and if you try to avoid it, either your project will collapse, or an alternative shadow system will emerge.
I have a lot of thoughts on this, many of which are not suitable for discussion on LW, but one thing I strongly believe is that people (esp. LWers) should not discuss politics except insofar as it informs a potential decision to emigrate.
As a framework for grounding a discussion, this has a lot of advantages. Most obviously, the decision if and where to emigrate is a near mode question, since most of us could actually emigrate if we wanted. Secondly, it grounds the discussion in solid facts, instead of airy conceptual speculation ("is Trump a racist?" becomes "will Trump hurt or help the economy?"). Emigration decisions are less judgmental: if I think Singapore is the best country, and some else says Sweden is better, then we simply conclude that I should move to SG and he should move to SE, not that one or the other is a bad person. Rationalists who spend a lot of time discussing sociopolitics are mostly failing to solve a bounded-rationality problem, since most of us will never influence the rest of society, so the time we spend thinking about it is largely wasted. Finally, at least for Americans, it reminds us that in spite of the apparently catastrophic state of our politics, the US is still a pretty good country overall.
I agree with this:
in spite of the apparently catastrophic state of our politics, the US is still a pretty good country overall
But I disagree with this:
people (esp. LWers) should not discuss politics except insofar as it informs a potential decision to emigrate
and this:
Rationalists who spend a lot of time discussing sociopolitics are mostly failing to solve a bounded-rationality problem, since most of us will never influence the rest of society, so the time we spend thinking about it is largely wasted.
It seems to me that the culture is influenced by discussion (for example, it is hard for me to believe that the discussion that took place on social media over the past year had no influence on the US election outcome). By recommending that LWers refrain from participating in political discussion, it seems to me that you are just removing a set of (relatively) rational, reflective and dispassionate voices from a discussion that badly needs that type of voice.
Being the calm, rational person in a fight doesn't always work, though. It only takes one party to want a fight to have one, after all.
Looks like you're conflating two things: being a rational person in a fight and being a rational person who want to avoid the fight. I would argue that being a calm and rational person in a fight is usually a good thing. I would also argue that avoiding the fight is not always the right choice. Sometimes a curb-stomping is very much needed.
what is needed is a good dynamic tension between the two sides
We have dynamic tension -- it's tension because things are tense :-) and it's dynamic because it just moved right in front of our eyes. As to "good", I don't know what you mean -- by what criteria do you want to decide whether it's good or not?
strategies for promoting useful discussion
Discussion is all well and good, but it's rather incomplete without doing. The "chattering class" label is not a compliment.
Everyone on this site obviously has an interest in being, on a personal level, more rational. That's, without need for argument, a good thing. (Although, if you do want to argue that, I can't stop you...)
But...
As a society, we're clearly not very rational, and it's becoming a huge problem. Look at any political articles out there, and you'll see the same thing: angry people partitioned into angry groups, yelling at each other and confirming their own biases. The level of discourse is... low, shall we say.
While the obvious facet of rationality is trying to discern the signal above the noise, there's definitely another side: the art of convincing others. That can swing a little too close to Sophistry and putting the emphasis on personal gain, though. What we really need to do is outreach: promote rationality in the world around us. There's probably no-one reading this who hasn't been in an argument where being more rational and right hasn't helped at all, and maybe even made things worse. We've also all probably been on the other side of that, too. Admit it. But possibly the key word in that is 'argument': it frames the discussion as a confrontation, a fight that needs to be won.
Being the calm, rational person in a fight doesn't always work, though. It only takes one party to want a fight to have one, after all. When there's groups involved, the shouty passionate people tend to dominate, too. And they're currently dominating politics, and so all our lives. That's not a status quo any rationalist would be happy with, I think.
One of the problems with political/economic discussions is that we get polarised into taking absurd blanket positions and being unable to admit limitations or counter-arguments. I'm generally pretty far on the Left of the spectrum, but I will freely admit that the Right has both some very good points and a role to play: what is needed is a good dynamic tension between the two sides to ensure we don't go totally doolally either way. (Thesis, Antithesis, Synthesis etc.) And the tension is there, but it's certainly not good. We need to be able to point out failure modes to ourselves and others, encourage constructive criticism.
I think we need ways of both cooling the flames (both 1-on-1 and in groups), and strategies for promoting useful discussion.
So how can we do this? What can we do?