One important bit of context is that Duncan has already left LW because he didn’t like the moderation policies (it’s sort of an awkward grey area where other people post and link to his stuff and talk about it).
I don’t currently have bandwidth to respond in more detail about the main point you’re making. I think that having some kind of principled way to address either this sort of post (possibly), or ones that are not too far removed from it, is very important. And I don’t think we have such a principled approach yet, and that we should.
We’ve definitely thought about things in that vein – alongside things like John Maxwell’s comment on this post about asking for feedback if a post isn’t frontpaged or curated.
The issue is mostly that there’s a lot of potential, similar settings we want to give authors. And a serious problem is how to managed complexity-creep: if you give authors a lot of choices it can feel overwhelming (and, in fact, one of the reasons we removed the “submit to frontpage” option is that ever since LW 1.0, and still on 2.0, people have complained that the choice is confusing, and feels like they’re either sacrificing the visibility of frontpage, or putting something they’re unsure about on frontpage and worry that they’re demanding too much attention for an unpolished idea, and it’s a bit of a nerve-wracking choice for many people)
We’ll eventually give this issue some thought and figure out how to go about it, but it’ll require a fair bit of thinking. Meanwhile, I’m not sure how often the “author didn’t want a post on Frontpage, but the mods put it there” issue will come up.
Cool – yeah I agree all these things sound great. One of the places we were currently looking at for this year’s Bay Summer Solstice is was actually pretty accessible all around (i.e. even in a wheelchair I think you could actually arrive fairly easily), and I’ll keep that sort of thing more in mind.
I’m interested if you have thoughts on “embodied fun” that would make for good examples for people with a wider variety of body types and capabilities – I generated the list mostly via “what would I find fun?” and agree it’d be good to include a wider variety of examples.
Is your premise that everyone taking this seriously should figure out their comparative advantage within an AI risk organization because they contain many non-researcher roles
Yes, basically. One of the specific possibilities I alluded to was taking on managerial or entreprenerial roles, here:
So people like me can’t just hand complicated assignments off and trust they get done competently. Someone might understand the theory but not get the political nuances they need to do something useful with the theory. Or they get the political nuances, and maybe get the theory at-the-time, but aren’t keeping up with the evolving technical landscape.
The thesis of the post is intended to be ‘donating to MIRI/CHAI etc is not the most useful thing you can be doing’
Curated both for being concretely helpful advice to advice givers, as well as for highlighting an overall rationality-failure-mode that’s easy to fall into.
Curated for succinctly creating some useful handles for two concepts that have implicitly been coming up a lot. i think this has already been helpful to me when thinking about some confusing/challenging conversations.
It was – but as Zvi says, it keeps getting gutted.
hmm. I can fix these links, but fyi if you clear your browser cache they should work for you. (If not, lemme know)
Yeah, I was stretching/abusing the definition of double crux here. I’d edit the original but I’m actually not sure how to quite phrase what I meant.
There’s a concept I’ve been thinking about lately I’ve been internally calling “aesthetic doublecrux” or “deep doublecrux.” In an in-person-conversation, I’d expect it to take at least a full day of discussion, quite likely much more.
The OP would essentially be the *first* stage of the discussion. (In person, it’d actually be an interwoven with the two people trying to explain all their background assumptions and mash their worldviews together. Online, in essay format… well I don’t know exactly how it’d work, but there’d need to be at least four stages of Essay/Response/Counter-Response/counter-counter-response (and it’d only end there if the counter-counter-response was “ah, I agree with your counter-response”).
In the spheres where Duncan has been commenting (which doesn’t include LW), I have noticed the pattern you point to (i.e. he hasn’t been engaging with the three major criticisms), and yeah this post does not earn the term “deep doublecrux” until he actually does that, and I think it makes sense to be wary about the fact that he hasn’t.
The version of my remark I’d endorse is something more like ″whatever you want to call it, the original post was a huge amount of effort, and I think fairly successful at being the first stage of an extended disagreement, of a class that normally doesn’t even get to the first stage. And it makes sense to hold up the bar that says ‘you aren’t actually done until we get to the end’, but I think it’s also important to at least acknowledge the effort so far.″
(The situation on how to consider the post and subsequent LW is a bit confusing, since Duncan didn’t post it here)
This would be totally fine if anyone else were making that point.
It seemed to me that other people were making the same point, AFAICT. (This comment by Dagon, this one by Vanessa). (I have more thoughts on this matter but focusing that response into another comment, see elsethread)
Is Duncan Right?
[taking off my mod hat, insofar as I honestly can]
I think the essay makes a lot of good points. I don’t like Punch Bug – I think it was the wrong choice of symbol to carry the argument forward, in particular because “no-punch-back” rules on non-opted-in games seem like bullshit to me, and importantly so. I’m not certain about roughhousing in general or the threshold of “punch” being the correct line.
But I think the essay is well within bounds. It is an important case study for attempting a deep/aesthetic double crux on a topic that normally would involve completely talking past each other.
And as much as I think no-punchback-rules are bullshit, I think they are importantly not on the same level as pogroms.
It so happens you can more easily paint a visceral picture of how no-punchback-roughhousing-games leads to pogroms, than, say, how bad economic policy might lead to pogroms. But I think bad economic policy is probably more relevant (or at least tied). Economic policy is also hard to get right, and there’s a lot of room for i.e. people arguing for and against minimum wage pointing at each other and calling each other monsters, and that clearly isn’t the environment you want for figuring out good economic policy, and I don’t think it’s the environment you want for figuring out interpersonal societal norms either.
In an adjaecent comment, Ozy describes BDSM consent norms, and notes that “a light tap doesn’t count as violence.” In the present-day world, I think it’s fair for people to note, if someone says “punch buggy no punch backs” and does a light tap, they are totally setting up a potential future situation where harder punches might happen.
But I can imagine a world, 20 years from now, where consent norms more firmly solidify, and someone hypothetically manages to invent “tap on the shoulder no tap back” without the historical baggage, and people react just as strongly as they are to the punching thing now. I think this would most likely be real bad.
And the argument here is that this has already happened, or is a about to happen. Humans adapt to treat pain signals as relevant depending on their context, and the feedback loop of treating smaller and smaller pain thresholds as suffering is bad – not actually reducing suffering on net, and meanwhile crimping a lot of important needs regarding touch, both gentle and rough.
I’m not sold on this argument, but it’s not obviously wrong.
The part where I think the essay is most wrong is where it doesn’t engage with the notion of actual abuse, with Punch-Bug-No-Punchbacks being a cover for bullying. But I feel like the correct response to this is more like “okay, this essay obviously misses this thing”, and then think about how to take the essay’s points seriously and do the thesis/anti-thesis/synthesis dance.
(I think this is mostly what’s been happening – pretty much every thread I’ve seen re: this post has included people bringing up those points, and at least some amount of synthesis, and I think this is basically correct)
[Moderator hat still on because it’s sort of dishonest to take it off, although this comment is much more off the cuff and not intended as a definitive LW Moderator Take, and the mod-hat in this case is more “I’m speaking as a guy with opinions on discourse which are informed by being a LW mod”]
First, while I stand by the “have a higher bar for invoking Nazis” guideline, the most important bit here is, as Zvi says in the Second Circle, remember to win. We are here to figure things out. I’m not confident I’m getting all the nuances here right, and making the right judgment calls is more important than having easy-to-follow guidelines. (With a further caveat that we do at least need good enough guidelines that there doesn’t always have to be a huge discourse when this sort of thing comes up)
With that in mind:
So, I see basically four ways to look at the situation:
1. Discussion/promotion of Punch Bug in particular, or anything relating to a combination of physical violence and opposition to BDSM style consent norms, is forever off the table – completely over the line.
2. It’s not intrinsically over the line, but it requires a lot of care and higher-than-average standards, and Duncan failed to meet those standards.
3. It requires care/higher-than-average-standards, and Duncan did meet those standards
4. LW (or related spaces) doesn’t have a strong stance on matters of consent or violence.
Options #1 and #4 both seem pretty bad to me. To be clear, #4 seems *worse* than #1, by a lot. Having a space for extreme truthseeking requires lower Maslow tiers to be satisfied.
But one of the most important points of LW, to me, is actually being able to say “hey, what if we’re wrong about our basic assumptions” even when it’s scary. We need to be able to have conversations that, if they go well, shape our future light-cone, hopefully for the better, and this will necessarily require resolution of meta-level conflict between people with very different values and frames. (If it didn’t involve such disagreements, this wouldn’t because “we picked the right frame and values”, it’d just be because we filter bubbled ourselves into an echo chamber)
It’d be convenient if that style of conversation could take place entirely in far mode without ramifications on the world we live in and the way we interact with each other on a day-to-day basis, but it doesn’t, and this means someone is going to feel threatened at least on occasion.
I can definitely imagine debating #2 vs #3, and I think adjaecent debates of ″what exactly are the meta-level norms that we can agree on that reasonably satisfice on as many people as possible feeling safe enough to have the kinds of conversations we need to have″ are important.
From what I can tell, you’re basically advocating for #1. (I’m also happy to debate #1, since, like, it’s a necessary part of the process. But, my current take is a strong “no”)
Yuppers. I definitely with the addition of REACH has a good chance of reducing or eliminating this problem – having lots of public facing events solves a lot of the issues.
Yeah, agreed that events that are “expecting effort” on the part of participants don’t usually have this problem.
The place where it seems most relevant are events that are sort of on the border between “hanging out with friends” and “hanging out with community” – house parties that play a large role in determining the overall social scene for Berkeley, where, say, 50-100 people get invited, but there 200 people in the area.
(This is not me saying anyone is doing anything wrong, just, it’s a thing to be aware of)
I think there is specifically a “work on x-risk” subgroup, which yes recruits from within Berkeley, and yes has some debilitating effects. I wouldn’t quite characterize it the way Zvi does but will say it’s not obviously wrong.
[Edit: I have mixed feelings about whether or how bad the current dynamics are. I think it actually is the case that x-risk desperately needs agents, and yes this competes with non-x-risk community building which also needs agents. I think it’s possible to make pareto-optimal improvements to the situation but there will probably be at least some tradeoffs that need to get made and I think reasonable people can disagree about where to draw those tradeoffs]
Ah, I think that makes sense.
I actually think this happens fairly frequently, although may be happening sort of invisibly:
I think it most concretely happened at the last Winter and Summer Solstice – in this case it was explicitly due to event insurance concerns and explicit attendee caps.
More often and more generally: I think, esp for medium-sized parties (basically any time it’s a private FB event, and the room ends up pretty full), I think it’s often the case that, before you got to the point where people notice and feel excluded, there’s a pre-emptive pass where only a smaller subset of people get invited in the first place. The competition is happening quietly in the social network.