the "go left / go right" diagram is 🎯. So, if you want to actually change people, the way to do it is by first detecting where they are, before giving them the advice. OR, just teach people to figure that out so that they filter all information through that.
The problem is right now the majority are still in this thing of, if the expert says "go left", and that clearly is going in a wrong direction, they think "well, I guess they are the expert, I should override the feedback right in front of me and listen to this guy instead". Terrible!!
I think we can supercharge this whole thing with "interactive essays". Basically ask people "what do you think X means" or how do you answer this, freeform, and an LLM can say "ok you already got it, this essay isn't for you, but you can read it anyway". Those answers also themselves act as data (you, and other readers, get to see what everyone thought going into the essay, and what their reaction was afterwards)
yes! I'm really excited by that idea. Esp. to build bridges across polarisation. It's difficult to talk through complex subjects without examples, but it's hard to use examples about some subjects that don't trigger anyone. Solution: classify yourself in the beginning so that the examples are all those that *won't* trigger your defences. Then, hopefully, when you've grabbed the general logic you can toggle it to see how it looks with the other examples you'd originally reject!
My take is: Even if “predator” signaling is effective for their intended ends:
1. The shifting dynamic will always eventually turn. What was once a powerful signal becomes a weakness (having undefended takes becomes a sign of untrustworthiness). But these people probably account for that and will switch to the next maximally successful strategy as needed.
2. Joe Hudson “Managing things creates a life you have to manage”: They’re creating a community in which “outcome-maxxing” is seen as valuable. Not only will they have to keep up with managing a convoluted style of discourse; they’re also treating culture as a complicated (AKA rational) system when it’s clearly a complex one (ie meta-rational, where any action will inevitably have unintended effects).
The first obvious problem I see with the strategy is simply that they’re polluting the commons with noise.
DefenderOfBasic has the best take on this imo, where in the long-term, communication strategies that are both open source and effective will win out, BECAUSE they’re effective. Strong signaling is great at achieving its stated aim, but inevitably it will run out of underlying truth “runway” and cede to strategies that are more efficient at generating/promoting/applying useful knowledge in the long term.
That said… memetics itself is also meta-rational, so there are times when strong signaling makes sense and I’m just being a little bitch for hating the player instead of the game.
yes. i think this is what keeps it going! the arbitrage is always on the 'underbought' side, so it just keeps churning
>2
i *hope* defender is right. that's what i'm betting on in "going meta" and pointing at the dynamics themselves. but, paraphrasing what you said, this is all reflexive and it's hard to *know* which way the chips will fall i.e. the only possible rational move is to hope.
don't you see the different between Hanlon’s Razor and POSIWID? one of them better then the other. one of them saying what it saying honestly and simply, and people agree whet it mean. and other too busy on surviving and spreading, so it compromised the actual meaning. the opposite of Hanlon’s Razor is not POSIWID, the opposite of Hanlon's Razor is "Never attribute to stupidity that which is adequately explained by malice".
I don't know how to point out the difference if you can't see the difference between "The purpose of a system is what it does" and "Never attribute to stupidity that which is adequately explained by malice". but one of them was subjected to much stronger selection pressure, though you can see the same bad thing in Hanlon’s Razor, but lesser.
also, it's actually possible to create culture that punish memes fir bring false, stupid oversimplification. the Rationalsphere strive to be exactly such a place.
so one one take part in that memetic war, he also take a side on the stupidly-intellegence war, and choose the side of stupidity. I disapprove.
Again, see above. The meaning of a meme is how it is used. POSIWID *effectively* means "Never attribute to stupidity that which is adequately explained by malice", in how it is used. But "Never attribute to stupidity that which is adequately explained by malice" did not spread. It's not as fit a meme as POSIWID. Hanlon's Razor is *also* an oversimplification—it must be to spread! It just seems you prefer one side over the other of the Type 1/Type 2 error.
so, I think i learned from this discussion on reason i deeply dislike POSIWID. it contribute to the memefication of thoughts. i strive to exist in places when it's possible and encouraged to make complicated, nuanced, arguments. and POSIWID is encouraging people to throw neuanse, and to pretend that what something does in 90% of situations is what it does. but it doesn't.
Hanlon’s Razor is NOT used like you describe. it use like that in the 90% of everything that is shit, and doesn't used like that IN PLACES THAT MATTERS.
so from my point of view, you take the actually important part of thing, and claim it never matters. it's totally backwards. the thing that matters is what people who actually trying to discuss things understand, not what people who try to shout louder shout. the memes are almost worthless, it's the people who actually talk that matters.
We just have different goals. You're trying to model the dynamics of people "who actually talk". I'm trying to model (here) the dynamics of online mass communication. Just different goals.
Very well thought-out post! Am also curious if these "overly attributing good intent" and "overly attributing bad intent" perspectives can fit in a different or bigger model? From the predator-prey model you made me think of other biological/ecosystem/evolutinary models like rock-paper-scissors (with side-blotched lizards). Or 3 different reproductive strategies (like male cuttlefish). Or even an extra layer in a predator-prey system, for example a grass (or "environment") layer in a grass-rabbit-fox model.
Sorry, maybe my wording was too sparse. I meant 3 different reproductive strategies _within_ the cuttlefish population. There are 3 strategies that work in reproducing: Parental/dominant males that build nests and guard them. Sneaker males who are small and dart in and release sperm quickly. Mimic females: males that look like females and get close to spawn undetected.
So there are 3 angles/strategies that work within that system. I was just thinking that there may be a 3rd angle (or meta-angle) in the "overly attributing good intent" and "overly attributing bad intent" model as well. But no idea what though!
Got it! TBH this reminds me of second-order cybernetics lmao. Like how does this (i.e.: ours) "meta-angle" of attempting to model the whole system fit *inside* of it, inside of the system being modelled?
(I have some thoughts, not on this in particular, but on reflexivity—how modelling affects the modelled. To be shared in a future blogpost...)
the "go left / go right" diagram is 🎯. So, if you want to actually change people, the way to do it is by first detecting where they are, before giving them the advice. OR, just teach people to figure that out so that they filter all information through that.
The problem is right now the majority are still in this thing of, if the expert says "go left", and that clearly is going in a wrong direction, they think "well, I guess they are the expert, I should override the feedback right in front of me and listen to this guy instead". Terrible!!
I think we can supercharge this whole thing with "interactive essays". Basically ask people "what do you think X means" or how do you answer this, freeform, and an LLM can say "ok you already got it, this essay isn't for you, but you can read it anyway". Those answers also themselves act as data (you, and other readers, get to see what everyone thought going into the essay, and what their reaction was afterwards)
> interactive essays
yes! I'm really excited by that idea. Esp. to build bridges across polarisation. It's difficult to talk through complex subjects without examples, but it's hard to use examples about some subjects that don't trigger anyone. Solution: classify yourself in the beginning so that the examples are all those that *won't* trigger your defences. Then, hopefully, when you've grabbed the general logic you can toggle it to see how it looks with the other examples you'd originally reject!
Juicy juicy meta-rational memetic discourse 🤤
My take is: Even if “predator” signaling is effective for their intended ends:
1. The shifting dynamic will always eventually turn. What was once a powerful signal becomes a weakness (having undefended takes becomes a sign of untrustworthiness). But these people probably account for that and will switch to the next maximally successful strategy as needed.
2. Joe Hudson “Managing things creates a life you have to manage”: They’re creating a community in which “outcome-maxxing” is seen as valuable. Not only will they have to keep up with managing a convoluted style of discourse; they’re also treating culture as a complicated (AKA rational) system when it’s clearly a complex one (ie meta-rational, where any action will inevitably have unintended effects).
The first obvious problem I see with the strategy is simply that they’re polluting the commons with noise.
DefenderOfBasic has the best take on this imo, where in the long-term, communication strategies that are both open source and effective will win out, BECAUSE they’re effective. Strong signaling is great at achieving its stated aim, but inevitably it will run out of underlying truth “runway” and cede to strategies that are more efficient at generating/promoting/applying useful knowledge in the long term.
That said… memetics itself is also meta-rational, so there are times when strong signaling makes sense and I’m just being a little bitch for hating the player instead of the game.
¯\_(ツ)_/¯
> meta-rational
would meaningness approve? one hopes
>1
yes. i think this is what keeps it going! the arbitrage is always on the 'underbought' side, so it just keeps churning
>2
i *hope* defender is right. that's what i'm betting on in "going meta" and pointing at the dynamics themselves. but, paraphrasing what you said, this is all reflexive and it's hard to *know* which way the chips will fall i.e. the only possible rational move is to hope.
:3
it is healthy to develop the right kind of proper predator energy, that can be made available and called upon when & where necessary.
have you seen the Dan Davies pieces?
http://webseitz.fluxent.com/wiki/2024-05-08-DaviesSeeingLikeAScrewdriver
http://webseitz.fluxent.com/wiki/2024-07-10-DaviesThePurposeOfASystemIsYouCantAlwaysGetWhatYouWant
had not, thanks for linking!
don't you see the different between Hanlon’s Razor and POSIWID? one of them better then the other. one of them saying what it saying honestly and simply, and people agree whet it mean. and other too busy on surviving and spreading, so it compromised the actual meaning. the opposite of Hanlon’s Razor is not POSIWID, the opposite of Hanlon's Razor is "Never attribute to stupidity that which is adequately explained by malice".
I don't know how to point out the difference if you can't see the difference between "The purpose of a system is what it does" and "Never attribute to stupidity that which is adequately explained by malice". but one of them was subjected to much stronger selection pressure, though you can see the same bad thing in Hanlon’s Razor, but lesser.
also, it's actually possible to create culture that punish memes fir bring false, stupid oversimplification. the Rationalsphere strive to be exactly such a place.
so one one take part in that memetic war, he also take a side on the stupidly-intellegence war, and choose the side of stupidity. I disapprove.
Again, see above. The meaning of a meme is how it is used. POSIWID *effectively* means "Never attribute to stupidity that which is adequately explained by malice", in how it is used. But "Never attribute to stupidity that which is adequately explained by malice" did not spread. It's not as fit a meme as POSIWID. Hanlon's Razor is *also* an oversimplification—it must be to spread! It just seems you prefer one side over the other of the Type 1/Type 2 error.
so, I think i learned from this discussion on reason i deeply dislike POSIWID. it contribute to the memefication of thoughts. i strive to exist in places when it's possible and encouraged to make complicated, nuanced, arguments. and POSIWID is encouraging people to throw neuanse, and to pretend that what something does in 90% of situations is what it does. but it doesn't.
Hanlon’s Razor is NOT used like you describe. it use like that in the 90% of everything that is shit, and doesn't used like that IN PLACES THAT MATTERS.
so from my point of view, you take the actually important part of thing, and claim it never matters. it's totally backwards. the thing that matters is what people who actually trying to discuss things understand, not what people who try to shout louder shout. the memes are almost worthless, it's the people who actually talk that matters.
We just have different goals. You're trying to model the dynamics of people "who actually talk". I'm trying to model (here) the dynamics of online mass communication. Just different goals.
Very well thought-out post! Am also curious if these "overly attributing good intent" and "overly attributing bad intent" perspectives can fit in a different or bigger model? From the predator-prey model you made me think of other biological/ecosystem/evolutinary models like rock-paper-scissors (with side-blotched lizards). Or 3 different reproductive strategies (like male cuttlefish). Or even an extra layer in a predator-prey system, for example a grass (or "environment") layer in a grass-rabbit-fox model.
thank you ser
what "third option" where you thinking of?
Sorry, maybe my wording was too sparse. I meant 3 different reproductive strategies _within_ the cuttlefish population. There are 3 strategies that work in reproducing: Parental/dominant males that build nests and guard them. Sneaker males who are small and dart in and release sperm quickly. Mimic females: males that look like females and get close to spawn undetected.
So there are 3 angles/strategies that work within that system. I was just thinking that there may be a 3rd angle (or meta-angle) in the "overly attributing good intent" and "overly attributing bad intent" model as well. But no idea what though!
Got it! TBH this reminds me of second-order cybernetics lmao. Like how does this (i.e.: ours) "meta-angle" of attempting to model the whole system fit *inside* of it, inside of the system being modelled?
(I have some thoughts, not on this in particular, but on reflexivity—how modelling affects the modelled. To be shared in a future blogpost...)