The target audience is people who feel like their interlocutor is using Motte and Bailey, when what’s actually happening is they’re trying to talk to a group and not separating out the individuals. Each individual person who knows what Motte and Bailey is who adds Mob and Bailey to their mental toolbelt is a point scored for this post, and every frustrated argument someone would have had of the form “You all said Y, but five minutes ago you all said X! Argh!” which instead has the form “You Bob said Y, and five minutes ago you Bella said X. I’m going to respond to Bob first, then I’ll respond to Bella after” is a victory for this post.
I’m going to type up a longer version in response to Ben’s reply here, but I think it’s fine to argue with one person and have others be convinced as “splash.” Another way to put this is if you start by responding to a specific argument, then you can maybe convince everyone who holds the original argument. E.g. “Lots of people think whales are fish, because they live in the ocean. I think this is wrong because whales have warm blood and fur, making them mammals, and that’s more important.” If someone thinks whales are fish because of a different reason, you may need different arguments. Though beware the idea that just because people make the same argument that they do so for the same reasons!
Alice is a feminist activist who is pro-choice. Bob is an environmentalist who is pro-choice because he thinks the human population is too high and anything to reduce it is a good thing. Charlie is pro-choice because he is a very conservative Christian who thinks god will punish countries that allow abortions, but he hates his country and wants to see it burn.
All three agents have a mind, that could in principle be changed.
The group {Alice, Bob, Charlie} doesn’t have a mind it can change.
You send the same text to Alice, Bob and Charlie. But if the essay is written with the understanding that these are three different people with wildly different views it is likely to be more successful than if you write the essay while modelling them as some kind of combined super-agent, who is a feminist, environmentalist, conservative Christian, anti-patriot.
The difference is a matter of how you model reality while constructing your arguments.
Still, you may have a point. Screwtape was plausibly modelling some kind of “median lesswrong user” in their head while writing their post. (I am curious if this is true?). And if they were doing that it was very likely a useful tactic. So maybe the subtlety is noticing when you are clumping people into averages, and being aware of the errors that can result in if the underling group is inhomogeneous in important respects.
There’s a concept in software design called Personas. A persona is a stereotype of a user; “Bill, a veteran mechanical engineer who uses our software for work” is one persona, “Alex, a novice hobbiest who saw our software on a youtube video” is another persona. I like working with personas since they’re good intuition pumps. I can guess that Bill wants keyboard shortcuts and compact buttons, while Alex wants tooltips or even to have the buttons labeled. The way I’ve worked with personas is that the ideal persona is contemplated as a singular person, and it works best if they are actually a real person. Do a survey, find a median user, get their contact info and actually have a conversation with them. If you solve a problem for that median user—a person who, again, ideally isn’t just a hypothetical but is a human being whose hand you can shake—then it’s pretty likely you solved a problem that other people have as well.
For this essay, my first target persona is “Screwtape, but from five years ago.” I’m pretty confident if I handed this to past!me, past!me would go “oooh yeah that makes sense, I’ll do that instead.” My second target is a relative of mine who keeps arguing with “democrats” without much success. “Democrats believe X, because Y, but Y is false!” ”. . . You know that you can be a Democrat and think X is false, right?”
Lets say you agree with Mob and Bailey, but you still want to convince a group of something. Stop, think for a couple of minutes by the clock. What does that actually look like?
Some of the time for me, I want a big organization to do something or change what it does; maybe I want the U.S. Military to invade Albuquerque or something. Convincing random people that Albuquerque is full of bad vibes isn’t very productive. Because the U.S. Military is fairly solidly structured, I can talk to specific people and ask who has decision-making power, then talk to those people (or their influences in a Themistocles’ Infant Son approach) and then if I win Albuquerque gets invaded. Note that you could describe this as “convince the U.S. Military” but that other things which also get described as “convince the U.S. Military” wouldn’t work; if one by one I talk every single boot camp sergeant into thinking Albuquerque is a den of scum and villainy then Albuquerque probably doesn’t wind up invaded.
Sometimes the goal looks like more and more of the group members being convinced. There’s maybe a tipping point where the group starts reinforcing the new belief, but if every individual in the group is convinced that’s kind of a win. Like, if tomorrow every single U.S. Senator woke up convinced that polyamory was obviously correct and the importance of our found families was an applause light, that would probably be a self-reinforcing state the way that the traditional family was an applause light for a long time. Here, I think it’s important not to get so distracted by the forest that you miss the trees. Senators want different things for different reasons and have different starting conditions. Maybe you split up the senators into personas and make arguments to each of them; here’s an argument based on individual rights, here’s an argument with historical or biblical premise, here’s an argument based on sociological outcomes.
I am not proposing to go full symmetric weapon rationalist dark arts here. Tying it back to the original post, I’m noting that people can believe something for different reasons and the way that we group people is usually very lossy. If Dean believes that zoning boards and current housing policy are bad because it’s an infringement of rights, and Emma believes zoning boards and current housing policy are bad because it drives up housing prices, trying to write one argument that will convince both of them seems needlessly hard.
Looking at your example of Alice, Bob, and Charlie who are all pro-choice, I wouldn’t try and send them all the same text. That sounds needlessly hard. If I did have to send them all the same message, that message might say “Alice, I think you should change your mind because X. Bob, I think you should change your mind because Y. Charlie, I think you should change your mind because Z.” Is there a reason you have to do it the hard way?
That last bit was a mistake on my part. My comment origionally said that “If you are for some reason operating under the constraint that you have to send the same text to them all (maybe posting on a forum they all, and others, read then.” I tried shortening it and ended up with the current nonsense.
“You cannot argue with a group. You cannot convince a group of things or change a group’s mind.”
Forgive me if this comes across as trollish, but whose mind are you trying to change with this essay?
To me it seems like your point is either self-refuting (in form, if not meaning) or, at best, incomplete.
The target audience is people who feel like their interlocutor is using Motte and Bailey, when what’s actually happening is they’re trying to talk to a group and not separating out the individuals. Each individual person who knows what Motte and Bailey is who adds Mob and Bailey to their mental toolbelt is a point scored for this post, and every frustrated argument someone would have had of the form “You all said Y, but five minutes ago you all said X! Argh!” which instead has the form “You Bob said Y, and five minutes ago you Bella said X. I’m going to respond to Bob first, then I’ll respond to Bella after” is a victory for this post.
I’m going to type up a longer version in response to Ben’s reply here, but I think it’s fine to argue with one person and have others be convinced as “splash.” Another way to put this is if you start by responding to a specific argument, then you can maybe convince everyone who holds the original argument. E.g. “Lots of people think whales are fish, because they live in the ocean. I think this is wrong because whales have warm blood and fur, making them mammals, and that’s more important.” If someone thinks whales are fish because of a different reason, you may need different arguments. Though beware the idea that just because people make the same argument that they do so for the same reasons!
Alice is a feminist activist who is pro-choice. Bob is an environmentalist who is pro-choice because he thinks the human population is too high and anything to reduce it is a good thing. Charlie is pro-choice because he is a very conservative Christian who thinks god will punish countries that allow abortions, but he hates his country and wants to see it burn.
All three agents have a mind, that could in principle be changed.
The group {Alice, Bob, Charlie} doesn’t have a mind it can change.
You send the same text to Alice, Bob and Charlie. But if the essay is written with the understanding that these are three different people with wildly different views it is likely to be more successful than if you write the essay while modelling them as some kind of combined super-agent, who is a feminist, environmentalist, conservative Christian, anti-patriot.
The difference is a matter of how you model reality while constructing your arguments.
Still, you may have a point. Screwtape was plausibly modelling some kind of “median lesswrong user” in their head while writing their post. (I am curious if this is true?). And if they were doing that it was very likely a useful tactic. So maybe the subtlety is noticing when you are clumping people into averages, and being aware of the errors that can result in if the underling group is inhomogeneous in important respects.
Mu.
There’s a concept in software design called Personas. A persona is a stereotype of a user; “Bill, a veteran mechanical engineer who uses our software for work” is one persona, “Alex, a novice hobbiest who saw our software on a youtube video” is another persona. I like working with personas since they’re good intuition pumps. I can guess that Bill wants keyboard shortcuts and compact buttons, while Alex wants tooltips or even to have the buttons labeled. The way I’ve worked with personas is that the ideal persona is contemplated as a singular person, and it works best if they are actually a real person. Do a survey, find a median user, get their contact info and actually have a conversation with them. If you solve a problem for that median user—a person who, again, ideally isn’t just a hypothetical but is a human being whose hand you can shake—then it’s pretty likely you solved a problem that other people have as well.
For this essay, my first target persona is “Screwtape, but from five years ago.” I’m pretty confident if I handed this to past!me, past!me would go “oooh yeah that makes sense, I’ll do that instead.” My second target is a relative of mine who keeps arguing with “democrats” without much success. “Democrats believe X, because Y, but Y is false!” ”. . . You know that you can be a Democrat and think X is false, right?”
Lets say you agree with Mob and Bailey, but you still want to convince a group of something. Stop, think for a couple of minutes by the clock. What does that actually look like?
Some of the time for me, I want a big organization to do something or change what it does; maybe I want the U.S. Military to invade Albuquerque or something. Convincing random people that Albuquerque is full of bad vibes isn’t very productive. Because the U.S. Military is fairly solidly structured, I can talk to specific people and ask who has decision-making power, then talk to those people (or their influences in a Themistocles’ Infant Son approach) and then if I win Albuquerque gets invaded. Note that you could describe this as “convince the U.S. Military” but that other things which also get described as “convince the U.S. Military” wouldn’t work; if one by one I talk every single boot camp sergeant into thinking Albuquerque is a den of scum and villainy then Albuquerque probably doesn’t wind up invaded.
Sometimes the goal looks like more and more of the group members being convinced. There’s maybe a tipping point where the group starts reinforcing the new belief, but if every individual in the group is convinced that’s kind of a win. Like, if tomorrow every single U.S. Senator woke up convinced that polyamory was obviously correct and the importance of our found families was an applause light, that would probably be a self-reinforcing state the way that the traditional family was an applause light for a long time. Here, I think it’s important not to get so distracted by the forest that you miss the trees. Senators want different things for different reasons and have different starting conditions. Maybe you split up the senators into personas and make arguments to each of them; here’s an argument based on individual rights, here’s an argument with historical or biblical premise, here’s an argument based on sociological outcomes.
I am not proposing to go full symmetric weapon rationalist dark arts here. Tying it back to the original post, I’m noting that people can believe something for different reasons and the way that we group people is usually very lossy. If Dean believes that zoning boards and current housing policy are bad because it’s an infringement of rights, and Emma believes zoning boards and current housing policy are bad because it drives up housing prices, trying to write one argument that will convince both of them seems needlessly hard.
Looking at your example of Alice, Bob, and Charlie who are all pro-choice, I wouldn’t try and send them all the same text. That sounds needlessly hard. If I did have to send them all the same message, that message might say “Alice, I think you should change your mind because X. Bob, I think you should change your mind because Y. Charlie, I think you should change your mind because Z.” Is there a reason you have to do it the hard way?
That last bit was a mistake on my part. My comment origionally said that “If you are for some reason operating under the constraint that you have to send the same text to them all (maybe posting on a forum they all, and others, read then.” I tried shortening it and ended up with the current nonsense.
Gotcha, that makes sense. Thanks for clarifying!