本文作者深入探讨了浪漫关系的真正价值所在。在反思了之前关于关系价值主张的观点后，作者发现自己仍然觉得关系的核心价值未被完全揭示。通过与不同观点的对话，作者排除了“被接纳自身缺陷”、“有人‘罩着你’”以及“承担照顾责任”等可能为关系带来价值的因素。最终，作者提出了一种新的假设：大多数人渴望被爱、被接纳，认为自己“足够好”，而作者本人则更看重伴侣能共同成长、变得更强的关系。这种对“成长”与“接纳”的权衡，或许是理解人际关系价值的关键。

🔑 **价值的探索与未竟之处：** 作者在对浪漫关系价值的探索中，发现自己对现有模型（如“愿意变得脆弱”）仍感不足，认为仍有更深层的价值未被触及。这促使他进一步挖掘，期望找到关系的核心驱动力。

🙅 **“缺陷接纳”并非核心价值：** 作者对“伴侣被自身缺陷所吸引”的观点持否定态度，认为这暗示伴侣希望自己变得更糟，而非变得更强。他认为，真正的关系应是相互促进成长，而非固守或纵容缺点。

🤝 **“我罩着你”的有限性：** 作者承认“有人‘罩着你’”的概念在某些特定领域（如AI对齐、解决衰老问题）具有极高价值，尤其当伴侣能在这些“真正困难”的领域提供支持时。然而，他指出，现实中能满足这些高门槛需求的伴侣极少，且这种价值并非普遍适用于所有关系。

⚖️ **“被需要”与“提供价值”：** 作者认为，自己对“被需要”的需求已通过AI领域的工作得到满足，甚至希望能减少这方面的负担。他更看重的是“提供价值”的能力，即便没有AI项目，他也能在其他领域创造价值，因此“被需要”并非关系价值的决定性因素。

🚀 **成长与接纳的二元选择：** 作者提出了一个核心观点：伴侣要么让你觉得“足够好”（被接纳），要么推动你“变得更强”（促进成长），而这两者在同一领域内难以兼得。他认为自己更偏好后者，并寻求能与他一同成长的伴侣，这可能解释了他与多数人对关系价值的认知差异。

Published on August 10, 2025 1:28 AM GMT

Meta:

This was written for myself to clarify various thoughts; if you’re seeing it then I thought other people might find value in it or might provide valuable-to-me responses, but other people are not really the audience.“We” = “I”; it’s a dialogue amongst parts of myself.

Voice 1: A couple months ago I wrote The Value Proposition of Romantic Relationships. On a personal level, the main motivation for that post was… I saw people around me who seemed to really highly value their relationships, and when I looked at those relationships, the value people put on them just didn’t add up. Like, when I looked at those relationships, they seemed usually pretty marginal-value at best, and very often just outright net negative (usually moreso for the males). Yet the people in those relationships seemed to think they were super high-value. So I figured, hmm, they must be getting some sort of massive value out of these relationships which I haven’t seen or understood for some reason. What the heck is that value?

So I talked to some people, synthesized what I heard with general background models, and The Value Proposition of Romantic Relationships was the result. Overcompressed answer: the main value proposition of romantic relationships is a cluster of stuff downstream of willingness to be vulnerable, including emotional support, play, forming a tiny high-trust community, and ability to communicate tough things. Under the model, these seemingly-disparate things cluster because they’re downstream of willingness to be vulnerable. And I’m still not sure that model quite hit the nail on the head; many commenters thought it was almost-but-not-quite pointing to the thing (though they all pointed to different alternatives). But based on the responses to that post, it seems at least pretty close.

Which brings us to today’s topic: when I look at the model from The Value Proposition of Romantic Relationships, and consider how I’d feel in a relationship which had all the aspects which that post talks about… I mean, it would be good, don’t get me wrong, but it still doesn’t feel that valuable. More than just marginal, but not enough that it would be the unambiguous largest value-contributor of a relationship. Presumably I am still missing something.

And I do have various signs that I’m still missing something big in a general cluster to which this topic belongs, so it seems worth digging into.

Voice 2: Ok, so, let’s start with a standard hypothetical. I’ll use Falkovich’s wording from one of his Second Person writings:

What’s your deepest insecurity or shame around dating? An unfulfilled aspiration, a personal flaw you could never fix, an obsession too cringe to share? What would it feel like if someone not only accepted it, but was specifically drawn to it?

This fits the “value downstream of willingness to be vulnerable” model very directly, and the way it asks makes it clear that Jacob expects people to find this hypothetical very high-value. So how do we feel about it?

Voice 1: Yeah, my immediate reaction to that one is “yuck”. Someone who’s specifically drawn to something which I myself am ashamed of? That would be a reason to not date that person; unambiguous negative value add.

Unpacking that instinctive response… I think the core thing here is that such a person wants me to be a weaker/worse person by my own lights. This is not a person who would make me stronger, who would push me to be stronger, at least in this one way. This is a person who would encourage me to embrace my own mediocrity, in whatever way it is that I’m mediocre, and just be happy with it.

… Actually, going back and rereading the prompt, my response here perhaps has some nonobvious assumptions built in. Like, presumably many peoples’ answers to that prompt would be sexual kinks - probably not even very unusual ones, just relatively-standard stuff like nonconsent or furries, where it seems outside the overton window if you don’t already know that there’s whole communities around these kinks. And I don’t think I have anything like that; I’m perfectly comfortable sharing my own sexual kinks^[1] and the like, even if they’re arguably somewhat taboo.

As a general rule, if I’m ashamed of something about myself, it’s something I consider a shortcoming by my own lights and I want to fix it, not embrace it. (Though I might not necessarily be allocating effort to fixing it right now; not everything can be top priority.) I’m generally reflectively stable about my own values (of which sexual kinks are a special case), and not ashamed of those. And yes, reflective stability about my own values includes basic stuff like “needing to not work literally all the time”, which novices have a tendency to overlook to their severe detriment.

I don’t think there’s anything even remotely in the vicinity of this prompt where I’d feel the way Falkovich seems to expect me to feel.

Voice 2: Ok, then let’s try a different direction. Here’s David’s response to the Value Proposition of Romantic Relationships post:

I think at least the "willingness to be vulnerable" is downstream of a more important thing which is upstream of other important things besides "willingness to be vulnerable." The way I've articulated that node so far is, "A mutual happy promise of, 'I got you' ".  (And I still don't think that's quite all of the thing which you quoted me trying to describe.)

So, mutual happy promise of ‘I got you’. How does that one sound? How do you feel about a relationship with a mutual happy promise of ‘I got you’?

Voice 1: First, I’ll note that I expressed skepticism of that one at the time. My main reason for skepticism was that I know at least one person who strongly agreed with David’s take, but is themselves in a relationship in which their partner unambiguously does not have their back (for lack of ability, not lack of will), yet that person still thinks they’re getting the main value proposition of relationships, whatever that may be.

Anyway, trying on the hypothetical… my knee-jerk reaction is, like, “ok I guess”? Like, I just… don’t particularly need someone to have my back? I have my shit together enough that I outright fall apart very rarely, and I have structured my life to generally have ample slack, so even if I do fall apart it’s nondisastrous.

But let’s check if there’s anything nearby which clicks more…

[...]

Ah. Oof. Ok, so, this is gonna require a little background. Setting aside the extent to which this is true, the way I feel about the field of alignment right now is that lots of people are trying to get other people to solve the problem, and lots of people are hoping to get LLMs to solve the problem, and the LLMs mostly just repackage things which were already written online, so nearly the entire field is entities outsourcing the work to other entities, until it grounds out in like two guys in an office somewhere who are actually working on the damn thing themselves. With the obvious two guys in question being me and David.

Point is: it would be absolutely amazing for someone else to have my back, when it comes to handling alignment of strong AI. That prospect was what I moved to the East Bay for, it was what I poured so much time into training people for. On the days (or weeks, or even months) when David and I are stuck, it would be great to know that someone else is actually doing the thing.

And if that person was also a romantic partner? Yeah, I’d give the world for that. That would actually feel like facing the world together with another person.

I could easily imagine feeling the same about someone on track to solve aging. That’s what I’d be doing if AI didn’t look like the higher priority, and I sure do feel the weight of failing to pursue it. I look at the field around it, and much like alignment, I mostly see clueless people larping; the last substantive unit of progress I know of was a decade ago. Man, it would be great if someone actually had my back there, if I could actually reasonably expect the problem to be solved without me having to pay attention to it. And if a romantic partner had my back in that department? Again, I’d give the world for that.

So, ok, maybe the ‘got your back’ thing has some legs to it. It’s just that the places where I feel enough chance of failure that I actually want someone to have my back are, like, actually hard. I don’t need someone to have my back for easy things.

(Also important to note: for both of these particular high-value possibilities, there probably just does not exist any candidate romantic partner who would fit. I would very likely have heard of them by now. But part of the point of this exercise is to explore the possibility space, so hopefully these two examples point to other high-value possibilities.)

Voice 2: I note that those examples both sound like finding a highly agentic equal partner. And people do seem to find a lot of value in other patterns besides that? People do have kids and pets, and there are definitely lots of happy relationships which are basically like that.

Voice 1: My knee-jerk reaction is “man, I do not need another dependent”. Like, clearly a lot of people want to be needed by someone, want to be responsible for other people. Whatever instinctive need there is for being-needed, mine is entirely saturated already by the whole AI thing. I would be fine with being needed less, if anything.

To be clear, that’s not to say I’m unwilling to take on more responsibility for other people. That’s definitely a pattern with my casual partners, for instance. And I’m fine with that, in the context of an overall-worthwhile broader relationship. But I wouldn’t consider it an upside, for me.

(Actually, on reflection I don’t think it’s that my need-to-be-needed is saturated by the AI thing. I think the real need is not so much to be needed, as to provide ample value to other people. And I think my need-to-provide-value-to-others would still be saturated even without the AI thing; even working at something else I’d create plenty of value.)

Voice 2: Ok, how about another pattern? Uli brought up the Comet King’s wife in Unsong as an example. Someone who is smart and competent and agentic (in the sense of figuring out how to achieve goals rather than just following a social script), but takes a support role for someone else. (And a concern we’ve heard from Uli and others is that the rationalist culture might just dismiss such a person, or accidentally alienate them, or some such.) Would that sort of relationship in fact be high value for us?

Voice 1: There’s definitely versions of that which would be very high value to me, and it’s a type of value-contribution which I’ve pretty explicitly tried to encourage in the past (e.g. the post How To Play A Support Role In Research Conversations), and a skillset I’ve explicitly tried to cultivate myself to some extent.

That said… the most common versions look like pretty marginal value, and the high value versions are very rare AFAICT. Unpacking that would stray pretty far from the point of this post, though. This axis (you might call it the corrigible vs sovereign axis) seems largely orthogonal to whatever the main value proposition of romantic relationships is supposed to be. So it’s probably not the main thing I’m missing here, one way or another.

Voice 2: Cool, then let’s try to get back to the main thread.

Voice 1: Here’s a thought, starting from the Falkovich thread above and thinking about how it would look in the context of my previous long relationship: it seems like a thing the large majority of people want is to be good enough, to be loved as they are, to have a partner who thinks they’re perfect, that sort of thing. And in practice, a partner who acts like I’m basically not importantly imperfect is a partner who does not push me to grow stronger or help me to grow stronger. On the flip side, if a partner is pushing me to grow stronger, then they’re not treating me like I’m already perfect. There is no growth without recognizing some way in which one is less than perfect.

Another way to put it:

If a partner makes you grow stronger and better, that means they’re not acting like you’re good enough as you are.If a partner acts like you’re good enough as you are, they’re not pushing you to grow stronger and better.

In any particular domain, you can choose at most one of those two things, you can’t have both. And I think most people crave the “good enough as you are” option (even if they profess otherwise), while I’m hitting the “grow stronger and better” option over and over again and have been for a long time and don’t ever expect to stop. And I definitely want a partner who at the bare minimum will help me continue to grow better and stronger, and ideally will grow better and stronger with me. Which means neither of us acting like the other is good enough as they are.

This also feels like it maybe ties into many responses to the empathy posts? Like, maybe the core thing nail-in-head-girl wants on a gut level is for her partner to act like she’s good enough as she is, like there’s nothing wrong with her which needs fixing, or something like that? Then my “yuck” response to her is similar to my “yuck” response to the idea of a partner being drawn to my flaws (and therefore this model maybe compresses some of my observations about how my instinctive responses differ from other peoples’ instinctive responses).

I dunno, this seems like a frame/hypothesis which could maybe account for a bunch of things by hypothesizing this particular delta between myself and a typical human. But I’m not entirely convinced that it’s the main thing here, or that it’s the right frame. Maybe it will prompt useful refinements.

1. ^{^}

   go ahead, ask in the comment section, I have no problem talking about this publicly

Discuss