I just wanted to say that you have my vote of confidence on this. It makes the intuitions behind the idea more salient as well.
Yeah, the 'membranes' term is a bit awkward in my mind, but vastly preferable as a choice for this technical jargon we are trying to establish as compared to the already too-many-meanings / overloaded term of boundaries.
I do think that some people mean something more like an actively defended border when talking about their social boundaries (not that I am sure I approve of this generally). I think they are trying to say that they have drawn social lines in the sand, which, if you cross them, they will consider that to be a boundary violation such that they will take action against you, such as 'cancelling' you (criticizing you, and refusing to allow you to be part of their social circle / ostracizing you / requesting that their social allies also ostracize you, perhaps taking legal action against you if there is grounds for a lawsuit). In some cases, this social boundary setting may be even more similar to a defended national border, in that the person setting the social boundaries makes a precommitment to use nonlawful coercive force / violence against violators. For instance, an angry individual saying "if you say <x> one more time, I will shoot you with this gun, even though it means I will likely be sentenced to life in prison." My point is that there is something of a gradient, and there is a clear 'this is a social boundary' in some cases and in others it's being used more like a metaphor, trying to make the weaker thing seem like the stronger thing, in a Motte & Bailey sort of way. I agree that focusing on the stronger meaning is the important thing, I just wanted to point out that it's a tricky line to draw sometimes.
I do think that some people mean something more like an actively defended border when talking about their social boundaries
To be clear, I don't think social requirements and membranes are unrelated. Actually, I think for anyone who is sufficiently conscious, if you violate (or attempt to violate) their «membranes/boundaries», then they will treat that as a violation of their social requirements (colloquial 'boundaries'). In the limit of courage, there is that convergence.
I think they are trying to say that they have drawn social lines in the sand, which, if you cross them, they will consider that to be a boundary violation such that they will take action against you, such as 'cancelling' you (criticizing you, and refusing to allow you to be part of their social circle / ostracizing you / requesting that their social allies also ostracize you, perhaps taking legal action against you if there is grounds for a lawsuit).
I wouldn't naturally call this «membranes». Now, it is of course in any individual's power to define how they interact with others (because they have «membranes/boundaries»). Sovereign agents can totally have their own preferences and choose to enforce them. For example, "I'm only going to talk to people who don't yell at me when they get mad"— but this takes an extra step, and it's different than «membranes/boundaries», which (I claim) is actually universal (ie: observable by all(?) observers) in a natural way.
In some cases, this social boundary setting may be even more similar to a defended national border, in that the person setting the social boundaries makes a precommitment to use nonlawful coercive force / violence against violators. For instance, an angry individual saying "if you say <x> one more time, I will shoot you with this gun, even though it means I will likely be sentence to life in prison."
If this were allowed in the model, then I could "draw my boundary" to include you, and shoot anyone who I don't like who tries to talk to you.
But no, I don't naturally have sovereignty over you and who you talk to. (E.g.: you'd probably fight me back)
The sovereignty (autonomy) idea is important. I'll have to write more about that in the future.
Also, how do I know that someone else is a sovereign agent? Well, because I can't control them. I can control rocks just fine. I'll run into trouble if I try to control other sovereign agents (eg people), though.
My point is that there is something of a gradient, and there is a clear 'this is a social boundary' in some cases and in others it's being used more like a metaphor, trying to make the weaker thing seem like the stronger thing, in a Motte & Bailey sort of way. I agree that focusing on the stronger meaning is the important thing, I just wanted to point out that it's a tricky line to draw sometimes.
I think it can be tricky, but I think there's something real here. («membranes» are distinct from preferences)
I'm not sure I get what you concretely mean by this boundary/membrane, but it sounds a lot like what is between a system and its environment, as discussed by Luhmann, i.e., that which differentiates.
I had a quick chat with ChatGPT about it:
What does "the difference that makes the difference" (Luhmann) mean in the context of the separation between a system and its environment?
ChatGPT-4:
"The difference that makes the difference" is a concept that originates from Gregory Bateson, a British anthropologist, and social scientist, but has been adopted by other theorists like Niklas Luhmann, a German sociologist known for his systems theory.
In Luhmann's systems theory, the world is seen as a vast network of interrelated systems. A system, in this view, is defined by a boundary that separates it from its environment. This environment comprises everything not included within the boundary of the system, which may in turn include other systems. The boundary is what allows a system to maintain its identity and function in a specific way, distinct from its environment.
The phrase "the difference that makes the difference" in this context essentially refers to the critical factors that distinguish a system from its environment and make it a separate entity. It's about the unique qualities or elements within a system that give it its distinct identity and enable it to function and adapt in its own way.
In Luhmann's sociological systems theory, for example, these "differences" might be the rules, norms, or social structures that separate one social system (like a legal system or an educational system) from another or from the broader societal environment. It's these differences that allow the system to process information and manage complexity in a way that's specific to that system. This concept emphasizes the idea of self-organization and autonomy, suggesting that systems actively construct their own boundaries and manage their interactions with their environments.
To summarize, "the difference that makes the difference" refers to the characteristics and factors that define and separate a system from its environment, enabling it to function and adapt in its specific way. This is a crucial aspect of Luhmann's systems theory, reflecting the system's autonomy and self-organization.
1but it sounds a lot like what is between a system and its environment
yes. and i think i particularly mean this to mean only boundaries that are 'natural' in some way. probably homeostatic/self-maintaining.
Looked at ChatGPT blurb- Yes this seems extremely related. Thank you, I wasn't aware of his work and i'll have to look into it! Let me know if you do think of any good resources
these "differences" might be the rules, norms, or social structures that separate one social system
this i might disagree with a little. Ie: I wouldn't call the "difference" of a cell from its environment the specific ion channels… I'd call the 'difference' the force of agency that's constructing and maintaining the boundary. (I agree that the rules or norms are differences, though. ugh, terminology)
A quick summary of Luhmann's systems theory can be found on his Wikipedia page: https://en.wikipedia.org/wiki/Niklas_Luhmann#Systems_theory
I recommend his seminal work Social Systems, though I can't say how good the English translation is. I have read the German original, and it is difficult to read because its approach to its quite abstract subject is to explain and explore from many angles and refer to the body of prior work and their terminology without the benefit of math (though math wouldn't help much I guess at this stage of the field).
I totally agree with the potential for confusion here!
My read is that the LessWrong community has too low of a prior on social norms being about membranes (e.g., when, how, and how not to cross various socially constructed information membranes). Using the term "boundaries" raises the prior on the hypothesis "social norms are often about boundaries", which I endorse and was intentional on my part, specifically for the benefit of LessWrong readership base (especially the EA community) who seemed to pay too little attention to the importance of <<boundaries>>, for many senses of "too little". I wrotr about that in Part 2 of the sequence, here: https://www.lesswrong.com/posts/vnJ5grhqhBmPTQCQh/boundaries-part-2-trends-in-ea-s-handling-of-boundaries
When a confusion between "social norms" and "boundaries" exists, like you I also often fall back on another term like "membrane", "information barrier", or "causal separation". But I also have some hope of improving Western discourse more broadly, by replacing the conflation "social norms are boundaries" with the more nuanced observation "social norms are often about when, how, how not, and when not to cross a boundary".
Reasoning is a membrane separating the physical world from the acausal jungle. Dangerous simulacra can crawl into reality through sufficiently capable cognition.
I think there is a boundary concept useful for alignment, not letting things from beyond through AI's reasoning rather than focusing on using AI's reasoning to channel specific things. Sufficiently good prevention of hallucinations, in LLM chatbot sense but more broadly construed, would only let an AI instantiate what's already in the world (in some unclear sense), not something new.
Just in case you haven't seen it: «Boundaries/Membranes» and AI safety compilation, «Boundaries» for formalizing a bare-bones morality. But you seem to be talking about this as a membrane insulating cognition, which is something I haven't thought of before... it's an interesting idea i think, i don't know what to make of it. Do let me know if you get more thoughts on it:)
a membrane insulating cognition
Cognition is the membrane, its sanity and alignment insulating the physical world, its capability providing the option of having scarier things pass through. It's an example of membrane vs. boundary distinction, because the membrane is a physical machine, the AI, not some line in the sand. And if it lets through what it shouldn't, the world dies (metaphorically for the world, literally for the people in it), so there is reason to maintain it in good condition. But it's a weird example, because the other side of the membrane looks into the platonic realm, not into another physical location, and it selectively lets through ideas/designs/behaviors, not physical compounds. An analogous example would be a radio, a device made out of atoms that selectively listens to electromagnetic signals.
The proposed alignment technique is guarding against hallucinations on the level of chatbot's personality rather than only of facts it voices, avoiding masks that have fictional personalities with fictional values. Not making up values strengthens the prior towards human values.
"Boundaries" term is bad because you can set fake boundaries and I don't think fake boundaries are worth talking about for safety purposes
We've been thinking about the core concept behind «boundaries» extensively, and going forward we're going to stop using the term "boundaries" (alone) but use the term "«boundaries/membranes»" instead.
Summary: We've found that whenever try to use the word "boundaries" with others in lay speech, our explainees have had trouble. We think this is because the word "boundaries" in English is overloaded with different definitions, some of which directly conflict with our intended meaning. Critch suggested "membranes" as an alternative term in his first «boundaries» post, and we suggest using that instead.
Difficulties with using the word "boundaries"
The definition of "boundaries" used in common social situations (preferences) means something different than «boundaries/membranes»
We've found that the common use of the word "boundaries" is very confusingly related to the «boundaries/membranes» concept.
For example, if you hear someone say, "Hey, you yelled at me and that crosses my boundaries!", this is very not what «boundaries/membranes» is about.
Basically, everything in this post — which describes people setting social rules about how to interact with them according to their consent or preferences — is exactly what we don't mean by the technical term. Another example: this post is not about the "boundaries" that we mean: This post talks about "boundary" rules being variable, which we'd likely call "preferences"— meanwhile we think the concept at hand is is independent of any particular agent.
Basically: "crossing (boundaries : social requirements enforced by someone)" does not imply "crossing «membrane-boundaries»".
And that's annoying because every time we explain the concept to someone using the word "boundaries", we have to qualify, "it's called boundaries, but it's not about the normal kind of boundaries, but it's also not unrelated, but please just ignore that for now" lest they get understandably confused. Ugh.
(Also, this exact conflation delayed Chipmonk's understanding of «boundaries/membranes» for the entire month of December!)
Note: Critch and we seem to agree that the concept is distinct from preferences. He says so in his Part 3b:
The definition of "boundaries" used to mean inert separation means something different than «boundaries/membranes»
There's one definition of the word "boundary" that means "a line that marks the limits of an area". However, in practice we find it useful to break this into two cases:
These are different: In the former case, there is autopoietic pressure (because if you don't defend your membranes, you die), in the latter case a line has been drawn in the sand. These are different, and doing this conceptual disambiguation whenever we use the word "boundaries" is annoying. Ideally, we'd just use a different term without this baggage.
Why use "membranes" instead?
The term "membranes" was originally proposed by Critch in «Boundaries», Part 1: a key missing concept from utility theory (bolding ours):
(Wikipedia: Membrane: "a selective barrier; it allows some things to pass through but stops others. Such things may be molecules, ions, or other small particles")
But Critch chose not to use "membranes" saying:
But we disagree with Critch on these last points, and so this is probably why we're willing to choose the term "membranes" even though he isn't.[1]
But we quite like the term. The word "membrane" is otherwise broadly used to refer to something that is independent of any particular agent (usually it points to something physical, eg a lipid bilayer). And we do think that concept at hand points at something that exists in more than just the mind of some set agents (e.g.: unlike human agreement or preferences).
And we don't think the word "membranes" comes with anywhere near the explanatory baggage that "boundaries" does. This isn't to say the word "membrane" is perfect, however— for example, it's not immediately obvious that the word "membrane" could refer to something nonphysical, eg an informational membrane/boundary.
Still, maybe there's an even better word (or frame) to use than "membranes" ("eg ownership"?, "sovereignty"?). Let us know.
For instance, I (Chipmonk) think «membranes/boundaries» are inherently autopoietic. For this reason, I'm pretty confused about where the «boundaries/membranes» might be in some of the examples Critch gives in «Boundaries», Part 2. E.g.: Where are the membranes in "work/life balance"?