social dynamics has a lot of stuff going on that is more complicated than you’re naively imagining.
I would like to register disagreement not with the technical truth value of this statement, but with the overall sentiment that I believe this statement (and your essay as a whole) is getting at: Social dynamics is really complicated and nuanced, and not getting the difficult parts right results in a lot of problems and overall suffering.
My belief is that just getting the basics right is usually enough to avoid most of the problems. I feel that a lot of the problems the rationality community has with social dynamics is basically due to a failure to simulate other minds (or simply a choice not to). For example, constantly violating small boundaries like being 2 hours late to an appointment. In what world does this behavior count as rational? If you predict yourself not to be able to make an appointment at a given time, why make the appointment at all? If something genuinely unexpected occurs that delays you, but you predict that the other person would be upset with you not informing them, why not tell them? This behavior to me sounds more like deontologically following first-order selfishness, the mistaken belief that rationality is defined as “being first-order selfish all the time.”
When our parents first teach us Basic Morality 101, they often say things like “Put yourself in that person’s shoes”, or “Do unto others” and things like that. And how we usually learn to interpret that is, “imagine yourself as that other person, feeling what they’re feeling and so on, and don’t do things that you imagine would make you unhappy in their situation.” In other words, simulate their experience, and base your actions upon the results of that simulation (TDT?). But then rationalists sometimes claim that socially skilled people are doing a bunch of fast System 1 intuitive calculations that are hyper-complex and difficult to evaluate with System 2 and so on, when it’s not really the case. Simulating another mind, even if you’re just approximating their mind as your mind but with the environmental details changed, isn’t that hard to do (though I imagine some might argue with that), because at the very least most of us can recall a memory of a similar situation. And if you can’t simulate it at all, then you’re probably going to have a tough time being rational in general, because models are what it’s all about. But you don’t need to be a neuroscientist to run these kinds of simulations.
Social dynamics is hard to understand with your system 2 (i.e. deliberative/logical) brain. There’s a lot of subtle nuances going on, and typically, nerds tend to see the obvious stuff, maybe go one or two levels deeper than the obvious stuff, and miss that it’s in fact 4+ levels deep and it’s happening in realtime faster than you can deliberate.
But I haven’t seen that many examples of social dynamics that are 4+ levels deep. I imagine that even showing that such levels exist would be really hard. I think it’s more likely that nerds are not even getting the “obvious” stuff right. I attribute this tendency to nerds’ desire to understand the world symbolically or through abstract theories when doing so would be much harder than just making a model of the situation in your head by predicting things you’ve already observed to happen. And often, nerds won’t do the obvious stuff because they don’t see why it’s strictly necessary within the abstract model they have available. The things nerds typically get flak for in social situations, such as (not an exhaustive list) - shyness, focus on extreme details or highly specific topics, immodesty (in the sense of speaking condescendingly or matter-of-factly), bad hygiene, eccentric hobbies or entertainment—are either things that are irrelevant or non-problematic in the presence of other nerds, or things that can be easily fixed without going 4 levels deep into the intricacies of social dynamics. This is probably an issue of being too much of a Decartesian rationalist when more empiricism is required.
It’s true though, that some of us would like to move higher up in the world which requires less eccentricity or weirdness than is allowable in our closest social circles. In this case, a better understanding of social dynamics would be helpful. But I think that in the case of trying to manage and organize within the rationality community, we’re probably overcomplicating the task a bit.
Thanks. I’m mulling this over, but I think off-the-cuff response is:
a) I agree a lot of nerds mostly need to get the basics right, and bother modeling other minds at all.
b) The thing that this post was a reaction to were naive attempts by rationalists to change culture without understanding what was going on. The most salient example was highlighted in Malcolm’s post (he’s quoting in turn from another post without attribution)
“As Tell Culture was becoming more popular in Berkeley, due to people mostly being excited about the bit in the name, it felt a good deal like I’d had Crocker’s Rules declared upon me at all times without my opt in.”
I think Brienne’s original Tell Culture post contains some warnings about how to do it right, but those warnings weren’t the part that travelled the fastest. I’ve also seen several other rationalist-types attempt to implement game-theoretic-models of how social-circles should be able to operate, while missing important steps. I think this people tended to also have somewhat atypical mental architecture, which makes “just bother to model other people at all” an insufficient solution (i.e putting yourself in someone else’s shoes doesn’t really work if you and they are sufficiently different people).
Basically, the problems that motivated this post were mostly people trying to be cultural pioneers beyond their skill level.
I think this people tended to also have somewhat atypical mental architecture, which makes “just bother to model other people at all” an insufficient solution (i.e putting yourself in someone else’s shoes doesn’t really work if you and they are sufficiently different people).
Exactly what I wanted to write. Thank you!
The problem with modelling other people by imagining how I would feel in the same situation is that this method keeps getting the results consistently wrong. I would actually say that my interactions with other people improved when I stopped imagining them as “me, being in a different situation”, and that accepting this was one of the most difficult lessons in my life.
My explicit models are often wrong or don’t give an answer, but they are still better than “putting myself in someone else’s shoes” which was almost guaranteed to return a wrong answer. Other people are not like me; and I kept resisting this piece of information for decades, resulting in a lot of frustration and lost opportunities, probably including the opportunity to develop my social skills a different way, without having them constantly sabotaged by the unconscious desire to imagine other people being “myself, in a different situation”. (Because I would love to find more people who actually are like myself.)
To prevent a motte-and-bailey response, sure, there are many things that me and other people have in common, mostly at or near the physiological level: we love pleasure, hate pain, usually enjoy being smiled at (unless we suspect some ulterior motive), hate being yelled at, etc. Yes, there are thousand similarities.
But there are also many differences, some of which are probably typical for the nerdy or rationalist people, and some may be just my personal quirks. I prefer to know the truth, no matter how good or bad it is. Other people prefer to be told things they already believe, whether correct or incorrect; new information is only acceptable as a form of harmless entertainment. I like to be told where I am wrong, assuming that it comes with a convincing explanation, and is not done as a status move. Other people hate to be told they are wrong, and they are quite likely to punish the messenger. I hate status hierarchies. Most people seem to love them, as long as they are not actively abused. I prefer to cooperate with other people, but many people get competitive—and this is the most difficult part for me to grok—over absurdly unimportant things. (I have already partially updated, and keep expecting that people would stab me in the back if e.g. million dollars would be at stake. But I still keep getting caught by surprise by people who stab me in the back for one dollar; luckily, not literally.) Etc.
So my model of most people is: “cute species, physiologically the same as me, but don’t get confused by the superficial similarities, at some level the analogy stops working.”
Life would be so much easier and my “social skills” much higher if modelling other people as “me, in a different situation” would approximately work. There are a few people with whom this approximately works. With most of the population, it is better (although far from perfect) to model them explicitly as—more or less—someone who wants to be smiled at and hates to be contradicted, and that’s all there is. When I follow this simplistic model, my social skills are okay-ish; I still lack scripts for many specific situations. But imagining other people as “me, in a different situation” is a recipe for frustration.
I believe that other people also have a wrong model of me. (But they probably don’t care, because… what I already wrote above.) At least, when they tell me about their model of me explicitly, I don’t recognize anything familiar. Actually, different people have completely contradictory models of me. Actually, I have a nice, almost experimental-setting experience of this—once I participated at a psychology training, where at one lesson we had to provide a group feedback to each member. A few classmates were missing. The group has just finished giving their group feedback to me, when the missing classmates came and apologized for being late. So the teacher told the other people to shut up, and asked the newcomers only to provide a group feedback to me. What they said was almost exactly the opposite of what the previous group said. -- In other situations different people described as a militant atheist, fanatical catholic, treacherous protestant, brainwashed jehovah witness; a communist, a libertarian, pro-russian, pro-american; a pick-up-artist, a womanizer, an asexual, a gay. My most favorite was a conspiracy theory someone made after reading my blogs (at different place, at different time), concluding that I am not a real person, but instead some weird online psychological project written by multiple people using the same account (apparently that was their Occam razor for how a single blogger could simultaneously care about math, cognitive science, and philanthropy)....
...shortly, if other people completely fail to simulate me, using the symetrical algorithm of imagining themselves at my place, why should I expect to achieve better results by using the same algorithm on them? Just because it mostly works when one average person simulated another? (And I am saying “mostly” here, because as far as I know, the interactions of average people with each other are still full of misunderstandings.)
I find what you said about the competitiveness aspect of other people the most compelling, but aside from that, I’ve had increasing success with imagining someone else as myself but in another situation. It takes breaking down what a “situation” is and entails, and how entrenched it is into our identity. Think of times in your life when you were vastly different. Maybe you were stressed about your first SO in high school—perhaps someone you can’t even imagine liking now. Can you successfully pretend, through the haze of time, being as obsessed over that person now as you were then? Is anything blocking that sensation, and if so, what?
Back when I was a sophomore in undergrad, I had a “bro”-like roommate. I was your typical college hipster, and he had a very abrasive approach to camaraderie. (He was not unlike the competitive type you describe.) I had the hardest damn time trying to figure out what was going on in his head, and one day he called me out for my introverted personality, misreading my reclusiveness as conceit. When he did that, I remember crying in my room, because there was this fundamental misunderstanding between us. He didn’t understand my approach or emotionality, and I didn’t get his. But over time I had a sensationally gradual realization that I could actually interact with this guy as “one of his own.” It started as mimicry, and I did “toughen up” a little through this event, but it also gave me an idea.
Before this time, I didn’t codify what it actually meant to feel like someone else. You can’t walk a mile in someone else’s shoes if they don’t fit your feet, right? When I was able to (in part) embody his approach to the world (while retaining my own on a backburner), it was easier to realize that this kind of sensational transition is necessary for everyone. It isn’t that we are fundamentally different, always, full-stop. It’s just that we can’t just start imagining someone else’s experiences a priori without forgetting who we are for a little while.
Don’t mistake this for spiritual chakra mumbo-jumbo or what-have-you, all I mean is that there is a mundane, demonstrable relationship between the body and all of the factors influencing it. You might be able to consider some of these factors a priori, but take care to recognize that your evaluation of their count and intensity may not at all be accurate, and that estimate could very well be blurred by an opposing sensation (or lack thereof) that your self currently retains.
But when you do become someone else, or rather, when your identity is changing shape to accommodate some grief or other perhaps more complicated sensation, take note of the nature of that change. Even if you can’t fully comprehend someone else’s motivations as your own, taking note of how you change shape and generalizing that this can happen to wildly varying degrees within our species (thinking about a civilian’s mental adjustment to being a soldier, accruing PTSD, etc), can be a very helpful placeholder for imagining the experiences of someone else’s life, and recognizing their actions as reasonable, given the possible constraints.
So my model of most people is: “cute species, physiologically the same as me, but don’t get confused by the superficial similarities, at some level the analogy stops working.”
I agree that in general people can differ pretty substantially in terms of preferences and interactions in a way that makes golden rule style simulations ineffective.
e.g. I seem to prefer different topics of small talk than some people I know, so if they ask me to, say, go into details about random excerpts of my day at work I get a bit annoyed whereas if I ask them mirror questions they feel comfortable and cared for. So both of us put the other off by doing a golden rule simulation, and we’ve had to come up with an actual model of the individual to in order to effectively care for the other person.
At the same time, some of these examples you give to me feel like inside view vs outside view explanations, in particular this line stood out:
I like to be told where I am wrong, assuming that it comes with a convincing explanation, and is not done as a status move. Other people hate to be told they are wrong, and they are quite likely to punish the messenger.
I think a fairly common failing is for people not to consider closely how the other will feel when some information is related to them (I know I personally am often less considerate in my words than is warranted). I think it’s not so uncommon to feel attacked by someone who was merely inconsiderate rather than attacking, partially because it really is hard to be sure of which the other person is until it’s too late (being considerate towards an ambiguous person who was attacking often opens up a substantially larger attack surface).
The failure of “put yourself in their shoes” seems similar to the failure of “do to others as you’d have them do to you”. You have to be hyperaware of each way that the person you’re modeling is different from you, and be willing to use these details as tools that can be applied to other things you know about them. This is where I actually find the ideas of guess/ask/tell culture to be the most helpful. They honestly seem pretty useless when not combined with modeling, precisely because it turns into “this is the one I have picked and you just have to deal with it”.
To prevent a motte-and-bailey response, sure, there are many things that me and other people have in common, mostly at or near the physiological level: we love pleasure, hate pain, usually enjoy being smiled at (unless we suspect some ulterior motive), hate being yelled at, etc. Yes, there are thousand similarities.
Even something simple as “hate pain” isn’t as straightforward. There are plenty of masochists out there who choose to act in a way that painful to them. Pain can make people feel alive.
So my model of most people is: “cute species, physiologically the same as me, but don’t get confused by the superficial similarities, at some level the analogy stops working.”
With most of the population, it is better (although far from perfect) to model them explicitly as—more or less—someone who wants to be smiled at and hates to be contradicted, and that’s all there is. When I follow this simplistic model, my social skills are okay-ish; I still lack scripts for many specific situations. But imagining other people as “me, in a different situation” is a recipe for frustration.
I might be closer to normal people in terms of mental models compared to other people here on LW. Or the people I’m around at school may also be more interesting than average. (Or many other possibilities...)
But the point seems to be that I find that neither of these models tend to match my interpersonal experiences.
To be clear, I think there exists models of abstraction where it might be more convenient to treat people as NPC’s and/or driven by simplified incentives.
For the most part, though, my face-to-face interactions with people has me modeling them as “people with preferences that might be very different from mine, perhaps missing some explicit levels of metacognition, but overall still thinking about the world”.
This doesn’t preclude my ability to have good discussions, I don’t think.
In conversations, I’ll try to focus on shared areas of interest, cultivate a curiosity towards their preferences, or use the whole interaction as an exercise to try and see how far I can bridge the inferential gap towards things I’m interested in. And this ends up working fairly well in practice.
(EX: rationality-type material about motivation seems to be generally of interest to people, and going at it from either the psychology or procrastination angle is a good way to get people hooked.)
I’ve found that people who are “on the clock” so to speak (that is, are at that moment talking with you solely because of a job they’re doing) are almost always easier to interact with when treated as NPCs with a limited script that is traversed mostly like a flowchart.
Police officers pulling you over, wait staff at a restaurant, and phone technical support representatives (to name some examples) are sometimes literally following a script. It can be helpful for both of you to know their script and to follow it yourself.
In conversations, I’ll try to focus on shared areas of interest, cultivate a curiosity towards their preferences, or use the whole interaction as an exercise to try and see how far I can bridge the inferential gap towards things I’m interested in. And this ends up working fairly well in practice.
Yeah, this is pretty much what I do, too. (Well, when I remember to do this.) But “how far I can bridge the inferential gap towards things I’m interested in” usually doesn’t get far before things get rounded to nearest cliche.
The problem is to keep remembering that I cannot say anything “weird”. Which includes almost everything I am interested at. Which includes even the way I look at things other people happen to be also interested about.
As an approximation, when I successfully suppress most of what makes me me, and channel my inner ELIZA, sometimes I even receive feedback on having good social skills. Doing this doesn’t feel emotionally satisfying to me, though.
I would like to register disagreement not with the technical truth value of this statement, but with the overall sentiment that I believe this statement (and your essay as a whole) is getting at: Social dynamics is really complicated and nuanced, and not getting the difficult parts right results in a lot of problems and overall suffering.
My belief is that just getting the basics right is usually enough to avoid most of the problems. I feel that a lot of the problems the rationality community has with social dynamics is basically due to a failure to simulate other minds (or simply a choice not to). For example, constantly violating small boundaries like being 2 hours late to an appointment. In what world does this behavior count as rational? If you predict yourself not to be able to make an appointment at a given time, why make the appointment at all? If something genuinely unexpected occurs that delays you, but you predict that the other person would be upset with you not informing them, why not tell them? This behavior to me sounds more like deontologically following first-order selfishness, the mistaken belief that rationality is defined as “being first-order selfish all the time.”
When our parents first teach us Basic Morality 101, they often say things like “Put yourself in that person’s shoes”, or “Do unto others” and things like that. And how we usually learn to interpret that is, “imagine yourself as that other person, feeling what they’re feeling and so on, and don’t do things that you imagine would make you unhappy in their situation.” In other words, simulate their experience, and base your actions upon the results of that simulation (TDT?). But then rationalists sometimes claim that socially skilled people are doing a bunch of fast System 1 intuitive calculations that are hyper-complex and difficult to evaluate with System 2 and so on, when it’s not really the case. Simulating another mind, even if you’re just approximating their mind as your mind but with the environmental details changed, isn’t that hard to do (though I imagine some might argue with that), because at the very least most of us can recall a memory of a similar situation. And if you can’t simulate it at all, then you’re probably going to have a tough time being rational in general, because models are what it’s all about. But you don’t need to be a neuroscientist to run these kinds of simulations.
But I haven’t seen that many examples of social dynamics that are 4+ levels deep. I imagine that even showing that such levels exist would be really hard. I think it’s more likely that nerds are not even getting the “obvious” stuff right. I attribute this tendency to nerds’ desire to understand the world symbolically or through abstract theories when doing so would be much harder than just making a model of the situation in your head by predicting things you’ve already observed to happen. And often, nerds won’t do the obvious stuff because they don’t see why it’s strictly necessary within the abstract model they have available. The things nerds typically get flak for in social situations, such as (not an exhaustive list) - shyness, focus on extreme details or highly specific topics, immodesty (in the sense of speaking condescendingly or matter-of-factly), bad hygiene, eccentric hobbies or entertainment—are either things that are irrelevant or non-problematic in the presence of other nerds, or things that can be easily fixed without going 4 levels deep into the intricacies of social dynamics. This is probably an issue of being too much of a Decartesian rationalist when more empiricism is required.
It’s true though, that some of us would like to move higher up in the world which requires less eccentricity or weirdness than is allowable in our closest social circles. In this case, a better understanding of social dynamics would be helpful. But I think that in the case of trying to manage and organize within the rationality community, we’re probably overcomplicating the task a bit.
Thanks. I’m mulling this over, but I think off-the-cuff response is:
a) I agree a lot of nerds mostly need to get the basics right, and bother modeling other minds at all.
b) The thing that this post was a reaction to were naive attempts by rationalists to change culture without understanding what was going on. The most salient example was highlighted in Malcolm’s post (he’s quoting in turn from another post without attribution)
“As Tell Culture was becoming more popular in Berkeley, due to people mostly being excited about the bit in the name, it felt a good deal like I’d had Crocker’s Rules declared upon me at all times without my opt in.”
I think Brienne’s original Tell Culture post contains some warnings about how to do it right, but those warnings weren’t the part that travelled the fastest. I’ve also seen several other rationalist-types attempt to implement game-theoretic-models of how social-circles should be able to operate, while missing important steps. I think this people tended to also have somewhat atypical mental architecture, which makes “just bother to model other people at all” an insufficient solution (i.e putting yourself in someone else’s shoes doesn’t really work if you and they are sufficiently different people).
Basically, the problems that motivated this post were mostly people trying to be cultural pioneers beyond their skill level.
Exactly what I wanted to write. Thank you!
The problem with modelling other people by imagining how I would feel in the same situation is that this method keeps getting the results consistently wrong. I would actually say that my interactions with other people improved when I stopped imagining them as “me, being in a different situation”, and that accepting this was one of the most difficult lessons in my life.
My explicit models are often wrong or don’t give an answer, but they are still better than “putting myself in someone else’s shoes” which was almost guaranteed to return a wrong answer. Other people are not like me; and I kept resisting this piece of information for decades, resulting in a lot of frustration and lost opportunities, probably including the opportunity to develop my social skills a different way, without having them constantly sabotaged by the unconscious desire to imagine other people being “myself, in a different situation”. (Because I would love to find more people who actually are like myself.)
To prevent a motte-and-bailey response, sure, there are many things that me and other people have in common, mostly at or near the physiological level: we love pleasure, hate pain, usually enjoy being smiled at (unless we suspect some ulterior motive), hate being yelled at, etc. Yes, there are thousand similarities.
But there are also many differences, some of which are probably typical for the nerdy or rationalist people, and some may be just my personal quirks. I prefer to know the truth, no matter how good or bad it is. Other people prefer to be told things they already believe, whether correct or incorrect; new information is only acceptable as a form of harmless entertainment. I like to be told where I am wrong, assuming that it comes with a convincing explanation, and is not done as a status move. Other people hate to be told they are wrong, and they are quite likely to punish the messenger. I hate status hierarchies. Most people seem to love them, as long as they are not actively abused. I prefer to cooperate with other people, but many people get competitive—and this is the most difficult part for me to grok—over absurdly unimportant things. (I have already partially updated, and keep expecting that people would stab me in the back if e.g. million dollars would be at stake. But I still keep getting caught by surprise by people who stab me in the back for one dollar; luckily, not literally.) Etc.
So my model of most people is: “cute species, physiologically the same as me, but don’t get confused by the superficial similarities, at some level the analogy stops working.”
Life would be so much easier and my “social skills” much higher if modelling other people as “me, in a different situation” would approximately work. There are a few people with whom this approximately works. With most of the population, it is better (although far from perfect) to model them explicitly as—more or less—someone who wants to be smiled at and hates to be contradicted, and that’s all there is. When I follow this simplistic model, my social skills are okay-ish; I still lack scripts for many specific situations. But imagining other people as “me, in a different situation” is a recipe for frustration.
I believe that other people also have a wrong model of me. (But they probably don’t care, because… what I already wrote above.) At least, when they tell me about their model of me explicitly, I don’t recognize anything familiar. Actually, different people have completely contradictory models of me. Actually, I have a nice, almost experimental-setting experience of this—once I participated at a psychology training, where at one lesson we had to provide a group feedback to each member. A few classmates were missing. The group has just finished giving their group feedback to me, when the missing classmates came and apologized for being late. So the teacher told the other people to shut up, and asked the newcomers only to provide a group feedback to me. What they said was almost exactly the opposite of what the previous group said. -- In other situations different people described as a militant atheist, fanatical catholic, treacherous protestant, brainwashed jehovah witness; a communist, a libertarian, pro-russian, pro-american; a pick-up-artist, a womanizer, an asexual, a gay. My most favorite was a conspiracy theory someone made after reading my blogs (at different place, at different time), concluding that I am not a real person, but instead some weird online psychological project written by multiple people using the same account (apparently that was their Occam razor for how a single blogger could simultaneously care about math, cognitive science, and philanthropy)....
...shortly, if other people completely fail to simulate me, using the symetrical algorithm of imagining themselves at my place, why should I expect to achieve better results by using the same algorithm on them? Just because it mostly works when one average person simulated another? (And I am saying “mostly” here, because as far as I know, the interactions of average people with each other are still full of misunderstandings.)
I find what you said about the competitiveness aspect of other people the most compelling, but aside from that, I’ve had increasing success with imagining someone else as myself but in another situation. It takes breaking down what a “situation” is and entails, and how entrenched it is into our identity. Think of times in your life when you were vastly different. Maybe you were stressed about your first SO in high school—perhaps someone you can’t even imagine liking now. Can you successfully pretend, through the haze of time, being as obsessed over that person now as you were then? Is anything blocking that sensation, and if so, what?
Back when I was a sophomore in undergrad, I had a “bro”-like roommate. I was your typical college hipster, and he had a very abrasive approach to camaraderie. (He was not unlike the competitive type you describe.) I had the hardest damn time trying to figure out what was going on in his head, and one day he called me out for my introverted personality, misreading my reclusiveness as conceit. When he did that, I remember crying in my room, because there was this fundamental misunderstanding between us. He didn’t understand my approach or emotionality, and I didn’t get his. But over time I had a sensationally gradual realization that I could actually interact with this guy as “one of his own.” It started as mimicry, and I did “toughen up” a little through this event, but it also gave me an idea.
Before this time, I didn’t codify what it actually meant to feel like someone else. You can’t walk a mile in someone else’s shoes if they don’t fit your feet, right? When I was able to (in part) embody his approach to the world (while retaining my own on a backburner), it was easier to realize that this kind of sensational transition is necessary for everyone. It isn’t that we are fundamentally different, always, full-stop. It’s just that we can’t just start imagining someone else’s experiences a priori without forgetting who we are for a little while.
Don’t mistake this for spiritual chakra mumbo-jumbo or what-have-you, all I mean is that there is a mundane, demonstrable relationship between the body and all of the factors influencing it. You might be able to consider some of these factors a priori, but take care to recognize that your evaluation of their count and intensity may not at all be accurate, and that estimate could very well be blurred by an opposing sensation (or lack thereof) that your self currently retains.
But when you do become someone else, or rather, when your identity is changing shape to accommodate some grief or other perhaps more complicated sensation, take note of the nature of that change. Even if you can’t fully comprehend someone else’s motivations as your own, taking note of how you change shape and generalizing that this can happen to wildly varying degrees within our species (thinking about a civilian’s mental adjustment to being a soldier, accruing PTSD, etc), can be a very helpful placeholder for imagining the experiences of someone else’s life, and recognizing their actions as reasonable, given the possible constraints.
Can confirm :-)
I agree that in general people can differ pretty substantially in terms of preferences and interactions in a way that makes golden rule style simulations ineffective.
e.g. I seem to prefer different topics of small talk than some people I know, so if they ask me to, say, go into details about random excerpts of my day at work I get a bit annoyed whereas if I ask them mirror questions they feel comfortable and cared for. So both of us put the other off by doing a golden rule simulation, and we’ve had to come up with an actual model of the individual to in order to effectively care for the other person.
At the same time, some of these examples you give to me feel like inside view vs outside view explanations, in particular this line stood out:
I think a fairly common failing is for people not to consider closely how the other will feel when some information is related to them (I know I personally am often less considerate in my words than is warranted). I think it’s not so uncommon to feel attacked by someone who was merely inconsiderate rather than attacking, partially because it really is hard to be sure of which the other person is until it’s too late (being considerate towards an ambiguous person who was attacking often opens up a substantially larger attack surface).
The failure of “put yourself in their shoes” seems similar to the failure of “do to others as you’d have them do to you”. You have to be hyperaware of each way that the person you’re modeling is different from you, and be willing to use these details as tools that can be applied to other things you know about them. This is where I actually find the ideas of guess/ask/tell culture to be the most helpful. They honestly seem pretty useless when not combined with modeling, precisely because it turns into “this is the one I have picked and you just have to deal with it”.
Even something simple as “hate pain” isn’t as straightforward. There are plenty of masochists out there who choose to act in a way that painful to them. Pain can make people feel alive.
I might be closer to normal people in terms of mental models compared to other people here on LW. Or the people I’m around at school may also be more interesting than average. (Or many other possibilities...)
But the point seems to be that I find that neither of these models tend to match my interpersonal experiences.
To be clear, I think there exists models of abstraction where it might be more convenient to treat people as NPC’s and/or driven by simplified incentives.
For the most part, though, my face-to-face interactions with people has me modeling them as “people with preferences that might be very different from mine, perhaps missing some explicit levels of metacognition, but overall still thinking about the world”.
This doesn’t preclude my ability to have good discussions, I don’t think.
In conversations, I’ll try to focus on shared areas of interest, cultivate a curiosity towards their preferences, or use the whole interaction as an exercise to try and see how far I can bridge the inferential gap towards things I’m interested in. And this ends up working fairly well in practice.
(EX: rationality-type material about motivation seems to be generally of interest to people, and going at it from either the psychology or procrastination angle is a good way to get people hooked.)
I’ve found that people who are “on the clock” so to speak (that is, are at that moment talking with you solely because of a job they’re doing) are almost always easier to interact with when treated as NPCs with a limited script that is traversed mostly like a flowchart.
Police officers pulling you over, wait staff at a restaurant, and phone technical support representatives (to name some examples) are sometimes literally following a script. It can be helpful for both of you to know their script and to follow it yourself.
Yeah, this is pretty much what I do, too. (Well, when I remember to do this.) But “how far I can bridge the inferential gap towards things I’m interested in” usually doesn’t get far before things get rounded to nearest cliche.
The problem is to keep remembering that I cannot say anything “weird”. Which includes almost everything I am interested at. Which includes even the way I look at things other people happen to be also interested about.
As an approximation, when I successfully suppress most of what makes me me, and channel my inner ELIZA, sometimes I even receive feedback on having good social skills. Doing this doesn’t feel emotionally satisfying to me, though.