This whole thing started with various crazies on bluesky writing hit pieces on the rationalist and AI alignment communities, as one does. One of their main beats was to note social ties between the AI alignment community and the Red Means No (RMN) consensual non-consent sex party community, then gesticulate wildly and claim that the AI alignment community is a dangerous sex cult. As one does.
It felt to me like the responses I saw to those hitpieces from my own in-groups were… not consistently candid. It seemed obvious to me that the hitpieces were partially pointing to real things, even if those real things were wildly overblown and not actually problematic in practice. And I was pretty frustrated at my in-groups’ refusal to acknowledge what seemed to me obvious kernels of truth; it felt like my own friends were playing public relations games rather than being honest.
I have now updated. I think I was wrong, there are actually fewer kernels of truth to the attacks than I thought, and in fact I was pretty surprised at just how few. This post is mainly about the story of that update, what I previously believed obvious, and what I now believe.
The Story
My own involvement began when Ozy put up a post touching on the topic of RMN and its purported value for professional AI Safety networking. My judgement agreed with Ozy’s core point: RMN isn’t any better for networking than most other events in the extended rationalist community, and in fact most other events are probably better for networking purposes. Nonetheless, I did feel like Ozy’s take (like many others) was downplaying a grain of truth, so I dashed off a quick comment trying to be more honest about the situation as I understood it.
The crazies jumped on the comment, not unexpectedly.
More unexpectedly, 24 hours later I got word that Dustin Moskovitz was freaking out about the comment and my funding was at risk. At one point the suggestion was made that I delete the comment and write a retraction, but I certainly was not about to do that with a comment whose contents I still believed were basically true. 72 hours after the comment, I had received the official email: my funding was canceled.
(I will probably be fine; this post is not about my funding.)
I let it cool for a couple days, admittedly in part because I had caught a cold and was napping a lot. And then, I wrote up a post to properly explain my models, much better than that one quick comment on Ozy’s blog had. I shared that post with Gretta, who is an RMN orgynizer and knows far more about it than me[1]. Gretta convinced me that some foundational facts I had thought obvious were, in fact, probably false. This was a very big surprise to me!
So now it’s time for the real meat: what were the surprises?
What I Previously Believed
Here’s a rough outline of what I thought previously:
RMNs are in large part people adjacent to the rationalist community - at least a large minority, probably a majority (though I was not very confident).
Being high status in the extended rationalist community, including the rationalist-adjacent parts of the AI safety community, made one more likely to be invited to RMN if one applied.
Attending would at least sometimes have a positive effect on some peoples’ careers, in the same way networking normally does: people make friends, and down the line friends sometimes help each other get jobs or get into professional programs or whatever.
Attending an RMN does send some weak signal of intellectual openness and anti-conformism, which is useful for alignment work (though far from the most important thing)
None of the above was inherently bad or a big deal in practice; the effects were real but weak.
Most of the above was very obvious: it’s just how communities normally work regardless of whether sex/kink are involved, it’s what we should expect on priors, it would be surprising if it weren’t true.
Being very obvious, at least some people had picked up on the weak-but-true fact that RMN parties could be good for one’s career or social status, and therefore felt pressure to get into it even though they weren’t actually into consensual non-consent. That, everyone AFAICT agrees, is bad.
By downplaying and denying those weak-but-true facts, my friends were making it seem more like there was something to hide, and therefore increasing the felt pressure. To decrease the pressure, we needed to be honest that there was some real effect there, but it was really weak. That’s what I was aiming to convey.
But after talking to Gretta, I now think the above picture is (very surprisingly to me!) wrong.
What I Now Believe
As an RMN orgynizer, Gretta has way more info than I do about the people involved, and also reached out to the other orgynizers to check the facts.
My new median guesses are that:
Actually, there have been zero cases where high status in the rationalist community resulted in someone being accepted to RMN who would not have been otherwise.
Actually, there have been zero cases where attending RMN resulted in someone getting an AI Safety job or getting into an AI Safety professional program or whatever who would not have otherwise.
I find these pretty surprising! This is not what I would expect, a priori, from any community! What gives?
Without divulging too much confidential information… the numbers involved just aren’t that large, there aren’t that many people who’ve gone to RMN at all and it turns out my estimate of the overlap with rationalists was too high (it’s probably closer to 25% than 50%). So actually there haven’t been that many opportunities for it to come up. And those RMN attendees who are in the rationalist community, or do work in AI safety, generally got into the rationalist/AI Safety community well before attending RMN (including myself).
Much of the discussion Gretta and I had was going through a couple specific people I knew who seemed like they might have been examples. We went into moderate depth on those two examples, and I was surprised by the answers in both. We then talked about more patterns across many people, and it all generally added up. I’m not going to share details, because it is very much all private. But to those who are skeptical as I was, I will say: I previously would have been very skeptical of these claims, I was willing to walk away from my funding in order to be honest about it in public, and I have been swayed.
Propagating That Update
If the foundational assumptions are wrong, that changes everything downstream.
Imagine someone who felt pressured to get into consensual non-consent to advance their social status or career amongst the rationalists. I expect they would see everyone denying or downplaying the link, and feel kinda gaslit, or like people are hiding something. That’s probably how I’d feel in their shoes. And in worlds where there is a nontrivial kernel of truth to the connection between RMN and career/status… well, it probably helps those people more to be honest about it, not downplay it, but also be honest about how small the connection is.
But I no longer think we’re in that world. If the connection is in fact so weak as to be nearly nonexistent, so weak that it’s probably been a deciding factor for literally zero people… then honesty means saying that the connection between RMN and career/status has been a deciding factor for literally zero people.
Honesty does include admitting that this claim is surprising! It is not what one would expect a priori, it merits some evidence. That’s the part which I think connects to our hypothetical person who’s feeling gaslit: you’re not crazy, I too thought the connection was obviously there (though weak), it’s surprising if it’s not. And sadly, the nature of RMN means we can’t discuss most of the evidence in public. But again, for whatever my word is worth: I too thought the connection was obviously there, and I have been convinced otherwise. I’m still not confident, but my median guess is that social status in the rationalist or AI Safety community has never actually been a deciding factor in someone attending RMN, and attending RMN has never actually been a deciding factor in a hiring or promotion or acceptance to a professional program.
An Apology
Having updated, the last step is to call the traditional social API for publicly changing one’s mind: apologizing.
I was wrong, I was publicly wrong, and the people who were most unjustly inconvenienced as a result were the RMN orgynizers. To the orgynizers: I’m sorry. You do great work, you get all sorts of shit unjustly heaped upon you in public, and as a result of being wrong I unjustly contributed to that heap.
Gretta in particular was involved in the discussions from multiple angles. Some of that I don’t intend to apologize for; I’m sorry for being wrong, not for causing a ruckus on bluesky or angering Dustin Moskovitz. So rather than apologizing for those other parts, I will say: thank you Gretta for taking on this shit, thank you to the others involved in various discussions for taking on this shit, and thank you to Gretta in particular for going through the object-level evidence with me. You weren’t obligated to do that, but as a result of doing it I think you made my models, our community, and the world a little better.
Lastly, David, who was also on the grant that got cancelled. I also can’t honestly apologize to David; he knows what he signed up for by working with me. But I sure am grateful to David for working with me anyway, and for continuing to be a great research partner through it all. Our work is frustratingly slow these days, but it would be so much slower without David.
Also Gretta was one of many people involved in the behind-the-scenes arguments about the whole situation. Also I’m dating Gretta. So she had pretty much all the context on everything and was a pretty natural person to discuss with.
I think parts of how you approached your initial comment on Ozy's post were kinda dumb, but I also think your candor is really admirable and I respect the decision to acknowledge the (believed) kernel of truth a lot.
This post is about whether or not RMN is good networking, but your original comment was about mostly whether it was a signal of desirable traits for a researcher. You list this in your prior beliefs but not any update on it.
I have been to RMN more than once, and I think it is mostly rationalists actually. But it's close enough to 50/50 that I'm not confident.
Also I personally absolutely do upgrade my assessment of someone's AI alignment qualifications if I find out they go to RMN. In order to do useful thinking on AI alignment/safety, I claim it is a strict requirement that one is able to notice places where mainstream society is getting triggered and being stupid, and unflinchingly dismiss the mainstream's stupidity. Attending RMN is one of many possible strong signals that someone can dismiss mainstream stupidity in cases where the mainstream is being stupid. But there are many other signals which can substitute for it.
This whole thing started with various crazies on bluesky writing hit pieces on the rationalist and AI alignment communities, as one does. One of their main beats was to note social ties between the AI alignment community and the Red Means No (RMN) consensual non-consent sex party community, then gesticulate wildly and claim that the AI alignment community is a dangerous sex cult. As one does.
It felt to me like the responses I saw to those hitpieces from my own in-groups were… not consistently candid. It seemed obvious to me that the hitpieces were partially pointing to real things, even if those real things were wildly overblown and not actually problematic in practice. And I was pretty frustrated at my in-groups’ refusal to acknowledge what seemed to me obvious kernels of truth; it felt like my own friends were playing public relations games rather than being honest.
I have now updated. I think I was wrong, there are actually fewer kernels of truth to the attacks than I thought, and in fact I was pretty surprised at just how few. This post is mainly about the story of that update, what I previously believed obvious, and what I now believe.
The Story
My own involvement began when Ozy put up a post touching on the topic of RMN and its purported value for professional AI Safety networking. My judgement agreed with Ozy’s core point: RMN isn’t any better for networking than most other events in the extended rationalist community, and in fact most other events are probably better for networking purposes. Nonetheless, I did feel like Ozy’s take (like many others) was downplaying a grain of truth, so I dashed off a quick comment trying to be more honest about the situation as I understood it.
The crazies jumped on the comment, not unexpectedly.
More unexpectedly, 24 hours later I got word that Dustin Moskovitz was freaking out about the comment and my funding was at risk. At one point the suggestion was made that I delete the comment and write a retraction, but I certainly was not about to do that with a comment whose contents I still believed were basically true. 72 hours after the comment, I had received the official email: my funding was canceled.
(I will probably be fine; this post is not about my funding.)
I let it cool for a couple days, admittedly in part because I had caught a cold and was napping a lot. And then, I wrote up a post to properly explain my models, much better than that one quick comment on Ozy’s blog had. I shared that post with Gretta, who is an RMN orgynizer and knows far more about it than me[1]. Gretta convinced me that some foundational facts I had thought obvious were, in fact, probably false. This was a very big surprise to me!
So now it’s time for the real meat: what were the surprises?
What I Previously Believed
Here’s a rough outline of what I thought previously:
But after talking to Gretta, I now think the above picture is (very surprisingly to me!) wrong.
What I Now Believe
As an RMN orgynizer, Gretta has way more info than I do about the people involved, and also reached out to the other orgynizers to check the facts.
My new median guesses are that:
I find these pretty surprising! This is not what I would expect, a priori, from any community! What gives?
Without divulging too much confidential information… the numbers involved just aren’t that large, there aren’t that many people who’ve gone to RMN at all and it turns out my estimate of the overlap with rationalists was too high (it’s probably closer to 25% than 50%). So actually there haven’t been that many opportunities for it to come up. And those RMN attendees who are in the rationalist community, or do work in AI safety, generally got into the rationalist/AI Safety community well before attending RMN (including myself).
Much of the discussion Gretta and I had was going through a couple specific people I knew who seemed like they might have been examples. We went into moderate depth on those two examples, and I was surprised by the answers in both. We then talked about more patterns across many people, and it all generally added up. I’m not going to share details, because it is very much all private. But to those who are skeptical as I was, I will say: I previously would have been very skeptical of these claims, I was willing to walk away from my funding in order to be honest about it in public, and I have been swayed.
Propagating That Update
If the foundational assumptions are wrong, that changes everything downstream.
Imagine someone who felt pressured to get into consensual non-consent to advance their social status or career amongst the rationalists. I expect they would see everyone denying or downplaying the link, and feel kinda gaslit, or like people are hiding something. That’s probably how I’d feel in their shoes. And in worlds where there is a nontrivial kernel of truth to the connection between RMN and career/status… well, it probably helps those people more to be honest about it, not downplay it, but also be honest about how small the connection is.
But I no longer think we’re in that world. If the connection is in fact so weak as to be nearly nonexistent, so weak that it’s probably been a deciding factor for literally zero people… then honesty means saying that the connection between RMN and career/status has been a deciding factor for literally zero people.
Honesty does include admitting that this claim is surprising! It is not what one would expect a priori, it merits some evidence. That’s the part which I think connects to our hypothetical person who’s feeling gaslit: you’re not crazy, I too thought the connection was obviously there (though weak), it’s surprising if it’s not. And sadly, the nature of RMN means we can’t discuss most of the evidence in public. But again, for whatever my word is worth: I too thought the connection was obviously there, and I have been convinced otherwise. I’m still not confident, but my median guess is that social status in the rationalist or AI Safety community has never actually been a deciding factor in someone attending RMN, and attending RMN has never actually been a deciding factor in a hiring or promotion or acceptance to a professional program.
An Apology
Having updated, the last step is to call the traditional social API for publicly changing one’s mind: apologizing.
I was wrong, I was publicly wrong, and the people who were most unjustly inconvenienced as a result were the RMN orgynizers. To the orgynizers: I’m sorry. You do great work, you get all sorts of shit unjustly heaped upon you in public, and as a result of being wrong I unjustly contributed to that heap.
Gretta in particular was involved in the discussions from multiple angles. Some of that I don’t intend to apologize for; I’m sorry for being wrong, not for causing a ruckus on bluesky or angering Dustin Moskovitz. So rather than apologizing for those other parts, I will say: thank you Gretta for taking on this shit, thank you to the others involved in various discussions for taking on this shit, and thank you to Gretta in particular for going through the object-level evidence with me. You weren’t obligated to do that, but as a result of doing it I think you made my models, our community, and the world a little better.
Lastly, David, who was also on the grant that got cancelled. I also can’t honestly apologize to David; he knows what he signed up for by working with me. But I sure am grateful to David for working with me anyway, and for continuing to be a great research partner through it all. Our work is frustratingly slow these days, but it would be so much slower without David.
Also Gretta was one of many people involved in the behind-the-scenes arguments about the whole situation. Also I’m dating Gretta. So she had pretty much all the context on everything and was a pretty natural person to discuss with.