Communications we read or see produce four kinds of changes in our beliefs: Random – uncorrelated with much else of interest, Info – more correlated with the world as it is, vs. as it might be, Persuasion
I am enjoying this exchange. It recaps debates I'm always having with myself as a writer.
I think both of you would enjoy the book "Clear and Simple as the Truth", by Francis-Noel Thomas and Mark Turner. It is one of the best books I've read on writing. It is about standard English prose style. However, unlike most writing manuals, which just vaguely exhort you to be clear, Clear and Simple observes that standard style is a style of "disguised assertion" and that the appearance of clarity is always kind of illusion. It discusses the history of this style its unspoken assumptions about the structure of the subject mater and the nature of argument, the settings in which it is appropriate, etc.
It is also a short book. To me at least, this signals quality -- I suppose because it shows someone is not trying to signal quality by writing an unnecessarily long book!
I read an interview with a teenager who made about a million dollars by buying lousy penny stocks, and then sending out spam emails saying something like
BUY FOOBAR STOCK NOW!!! INCREDIBLE VALUE!!! GOING UP UP UP!!!
He experimented with it a lot, and said that though heinitially thought that using grammar and punctuationwould be more persuasive, he got better results using all caps.
This indicates that most people don't divide domains intoemotional ones in which persuasion is appropriate, andinformational ones in which reason is appropriate.Rather, those domains Robin singles out as "persuasion"domains are simply those domains in which more people operate,and so they are tailored to a style that is persuasive tomost people - a style Robin calls propaganda.
P.S.
Eliezer wrote: "The idea that our actual, evolutionary motives are lurking in our subconscious and actively steering things is not conventional evolutionary psychology."
AFAIK, this is not just conventional evolutionary psychology; it is the whole of conventional evolutionary psychology,and of psychology in general. Psychology usuallyproceeds as if there were no such thing as consciousness.
Eliezer, surely how friendly I act toward anyone one person at any one time is the result of far more than just some stable "is he my friend?" label stored somewhere in my brain. And surely whatever are the relevant internal states, many of the processes that change those states are influenced by indicators of how much it is in my genetic interest to act friendly to this sort of person in these sort of situations. If you agree with these claims, I don't see what you think you disagree with me about.
Robin, my understanding of the conventional view in evolutionary psychology is that we have evolved to signal certain attributes by actually having them - for example, to signal strong friendship by actually being strong friends with someone. The idea that our actual, evolutionary motives are lurking in our subconscious and actively steering things is not conventional evolutionary psychology.
The idea that our strong friendship might, for some odd reason, start to fade when our friend can no longer be of service to us, would be more conventional evolutionary psychology - but that's a new hidden circuit activating that the thinker doesn't know about; it's not a subconscious motivation that was at work all along.
Yes, we've evolved a general faculty to (a) in general, be influenced by 'unvirtuous' desires such as desire for high status; and (b) in general, put the best possible rosy cast on our actions. But to suppose that every trace of virtue is just status-seeking by trying to appear virtuous, is simply too cynical; it does not seem psychologically realistic.
I note that the rules may be different in economics than in evolutionary psychology. I can see why, in economics, you would try to put the burden of proof on someone claiming that a behavior is genuinely noble. But in evolutionary psychology, the burden of proof is on whoever claims the more complex adaptation. If there's a selection pressure to signal quality X, then one simple adaptation that does the trick will be to actually have quality X. If this seems on its face to be psychologically realistic for humans, then the default is to say, "This would seem to explain why we have quality X!", and maybe try to invent an experimental demonstration by reasoning about the ancestral conditions that should activate or maintain quality X. A more complicated conscious-plus-subconscious hypothesis that enabled even more sophisticated behavior, would bear the burden of further demonstration in a way that distinguished it from the simpler hypothesis. It would also prompt the question of why attention to the signal hadn't faded over evolutionary time.
Economic actors routinely invent quite complicated and intelligent plans on the spur of the moment; evolution tends to be relatively more parsimonious.
Robin, informative authors at the very least want audience to conclude the info they are giving is true. How does whether they can guess audience members' next steps of inference from it change the analysis?
Robin, yes of course it is wise not to insult high-status audiences. That wasn't my point; I was wondering if we should be paying attention to readers' signaling about themselves, and to authors' signaling about their beliefs about their readers, as we try to understand writing conventions. In particular, I was wondering if the desire to signal OB readers' (intelligence/seriousness/analytic skill/peerhood) is playing some role in the dislike of Eliezer's fiction.
Eliezer, humans evolved tendencies to take the actions that signal high status and when possible to consciously believe that such actions were done for some other more noble purposes.
Anna, yes one is wise not to insult high status audiences.
Katja, informative authors often have little idea what conclusions readers will draw from info provided.
Douglas, my topic is not what I can infer about Eliezer from his total corpus of writings.
All, I'd think it more important to understand typical author and reader behavior than to focus so on a few outliers like Smullyan or Feynman.
Robin makes two explicit (and correct) points:(1) people (ie EY) should signal(2) people (ie RH) should read signalsbut there is a subtext (pretty explicit in the comments, on other posts and here) that RH is penalizing EY for failing to signal. Since RH has a lot more information about EY than the signal, it is irrational to make much use of it. So it looks a lot less like the explicit theory of the local game and a lot more like cartel enforcement.
The way that signals start out as rational equilibria and calcify into hazing and arbitrary ways of limiting the size of the in-group is a very important fact about the world that I don't understand well. One theory is that people are unwilling to deal with multiple sources of information, particularly incomplete information, so simple tests that apply to everyone take over.
Robin,The variable persuasiveness is unnecessary (plus causing boring confusion). Presumably all communication is aimed at moving our beliefs toward what the author would have them be, to as much of a degree as the author can produce. Otherwise the author wouldn't bother. The question is whether the author would like us to believe true things or other things. I agree that communication style signals this.
A communicator can signal that they are persuading us of truth by actions like writing clearly partly because doing so makes critique easier. This doesn't mean that listeners are much more likely to reason for themselves, or should. It signals to listeners that the communicator is willing to put more at stake for their claims. This is same as other situations where information transparency allows common trust without everyone having to read the information.
Communication not aimed at truth *is* often more persuasive, but that's not due to differences in how much the author would like to persuade. Persuasive techniques undermine audiences' abilities to reason, as well as not including clear information. Thus to use them is to make critique harder, so what's at stake lower. Also persuasive techniques are more desirable to audiences who don't want to reason, so using them says you aren't expecting an audience interested in truth. My point is that persuasiveness is part of the signaling, not the feature beneath it.
Part of Smullyan's brilliance in "The Tao is Silent" is that he makes clear that he is not taking himself too seriously, though he thinks his conclusions rigorously correct. He succintly states his own inclinations, and then engages the reader to discover for himself or offer up a counterargument. Feynman's writings share that quality. Some academics are more pedagogically gifted than others.
It's also interesting to consider what a medium (e.g., complex academic text vs. fiction) signals about the the author's apparent beliefs about readers.
In my first research internship some years ago, when I was asked to present my progress to the research group, I tried to present the material tutoring-style, with questions (ten-second exercises) so listeners could check their comprehension and could actually get fluent with the formalism I was using. I was (correctly) told that I shouldn't use such formats as they are insulting to the listeners. A tutorial style exposition implies that the speaker is in charge of the "correct" answers and is competent to structure the listeners' learning, while standard academic presentations present research to be looked over and commented on by listeners, with a presumption that the listeners will understand.
Perhaps some of the annoyance at Eliezer's use of fiction stems from readers' insult at the view such fiction signals of their abilities or of their non-peerhood with Eliezer (and not from concern about fiction's effects on readers). If readers can instead claim to be members of a blog filled with math and complex, passive-voice sentences, they can signal that they can read such material.
Douglas - I think it was in Wired magazine a few years ago.
@Robin, Eliezer:
I am enjoying this exchange. It recaps debates I'm always having with myself as a writer.
I think both of you would enjoy the book "Clear and Simple as the Truth", by Francis-Noel Thomas and Mark Turner. It is one of the best books I've read on writing. It is about standard English prose style. However, unlike most writing manuals, which just vaguely exhort you to be clear, Clear and Simple observes that standard style is a style of "disguised assertion" and that the appearance of clarity is always kind of illusion. It discusses the history of this style its unspoken assumptions about the structure of the subject mater and the nature of argument, the settings in which it is appropriate, etc.
It is also a short book. To me at least, this signals quality -- I suppose because it shows someone is not trying to signal quality by writing an unnecessarily long book!
Phil Goetz,could you link to (or otherwise cite) that interview?
I read an interview with a teenager who made about a million dollars by buying lousy penny stocks, and then sending out spam emails saying something like
BUY FOOBAR STOCK NOW!!! INCREDIBLE VALUE!!! GOING UP UP UP!!!
He experimented with it a lot, and said that though heinitially thought that using grammar and punctuationwould be more persuasive, he got better results using all caps.
This indicates that most people don't divide domains intoemotional ones in which persuasion is appropriate, andinformational ones in which reason is appropriate.Rather, those domains Robin singles out as "persuasion"domains are simply those domains in which more people operate,and so they are tailored to a style that is persuasive tomost people - a style Robin calls propaganda.
P.S.
Eliezer wrote: "The idea that our actual, evolutionary motives are lurking in our subconscious and actively steering things is not conventional evolutionary psychology."
AFAIK, this is not just conventional evolutionary psychology; it is the whole of conventional evolutionary psychology,and of psychology in general. Psychology usuallyproceeds as if there were no such thing as consciousness.
Eliezer, surely how friendly I act toward anyone one person at any one time is the result of far more than just some stable "is he my friend?" label stored somewhere in my brain. And surely whatever are the relevant internal states, many of the processes that change those states are influenced by indicators of how much it is in my genetic interest to act friendly to this sort of person in these sort of situations. If you agree with these claims, I don't see what you think you disagree with me about.
I'm expanding my comment into a post.
Robin, my understanding of the conventional view in evolutionary psychology is that we have evolved to signal certain attributes by actually having them - for example, to signal strong friendship by actually being strong friends with someone. The idea that our actual, evolutionary motives are lurking in our subconscious and actively steering things is not conventional evolutionary psychology.
The idea that our strong friendship might, for some odd reason, start to fade when our friend can no longer be of service to us, would be more conventional evolutionary psychology - but that's a new hidden circuit activating that the thinker doesn't know about; it's not a subconscious motivation that was at work all along.
Yes, we've evolved a general faculty to (a) in general, be influenced by 'unvirtuous' desires such as desire for high status; and (b) in general, put the best possible rosy cast on our actions. But to suppose that every trace of virtue is just status-seeking by trying to appear virtuous, is simply too cynical; it does not seem psychologically realistic.
I note that the rules may be different in economics than in evolutionary psychology. I can see why, in economics, you would try to put the burden of proof on someone claiming that a behavior is genuinely noble. But in evolutionary psychology, the burden of proof is on whoever claims the more complex adaptation. If there's a selection pressure to signal quality X, then one simple adaptation that does the trick will be to actually have quality X. If this seems on its face to be psychologically realistic for humans, then the default is to say, "This would seem to explain why we have quality X!", and maybe try to invent an experimental demonstration by reasoning about the ancestral conditions that should activate or maintain quality X. A more complicated conscious-plus-subconscious hypothesis that enabled even more sophisticated behavior, would bear the burden of further demonstration in a way that distinguished it from the simpler hypothesis. It would also prompt the question of why attention to the signal hadn't faded over evolutionary time.
Economic actors routinely invent quite complicated and intelligent plans on the spur of the moment; evolution tends to be relatively more parsimonious.
Robin, informative authors at the very least want audience to conclude the info they are giving is true. How does whether they can guess audience members' next steps of inference from it change the analysis?
Robin, yes of course it is wise not to insult high-status audiences. That wasn't my point; I was wondering if we should be paying attention to readers' signaling about themselves, and to authors' signaling about their beliefs about their readers, as we try to understand writing conventions. In particular, I was wondering if the desire to signal OB readers' (intelligence/seriousness/analytic skill/peerhood) is playing some role in the dislike of Eliezer's fiction.
Eliezer, humans evolved tendencies to take the actions that signal high status and when possible to consciously believe that such actions were done for some other more noble purposes.
Anna, yes one is wise not to insult high status audiences.
Katja, informative authors often have little idea what conclusions readers will draw from info provided.
Douglas, my topic is not what I can infer about Eliezer from his total corpus of writings.
All, I'd think it more important to understand typical author and reader behavior than to focus so on a few outliers like Smullyan or Feynman.
Best...Comments thread...Ever!
Robin makes two explicit (and correct) points:(1) people (ie EY) should signal(2) people (ie RH) should read signalsbut there is a subtext (pretty explicit in the comments, on other posts and here) that RH is penalizing EY for failing to signal. Since RH has a lot more information about EY than the signal, it is irrational to make much use of it. So it looks a lot less like the explicit theory of the local game and a lot more like cartel enforcement.
The way that signals start out as rational equilibria and calcify into hazing and arbitrary ways of limiting the size of the in-group is a very important fact about the world that I don't understand well. One theory is that people are unwilling to deal with multiple sources of information, particularly incomplete information, so simple tests that apply to everyone take over.
Robin,The variable persuasiveness is unnecessary (plus causing boring confusion). Presumably all communication is aimed at moving our beliefs toward what the author would have them be, to as much of a degree as the author can produce. Otherwise the author wouldn't bother. The question is whether the author would like us to believe true things or other things. I agree that communication style signals this.
A communicator can signal that they are persuading us of truth by actions like writing clearly partly because doing so makes critique easier. This doesn't mean that listeners are much more likely to reason for themselves, or should. It signals to listeners that the communicator is willing to put more at stake for their claims. This is same as other situations where information transparency allows common trust without everyone having to read the information.
Communication not aimed at truth *is* often more persuasive, but that's not due to differences in how much the author would like to persuade. Persuasive techniques undermine audiences' abilities to reason, as well as not including clear information. Thus to use them is to make critique harder, so what's at stake lower. Also persuasive techniques are more desirable to audiences who don't want to reason, so using them says you aren't expecting an audience interested in truth. My point is that persuasiveness is part of the signaling, not the feature beneath it.
Part of Smullyan's brilliance in "The Tao is Silent" is that he makes clear that he is not taking himself too seriously, though he thinks his conclusions rigorously correct. He succintly states his own inclinations, and then engages the reader to discover for himself or offer up a counterargument. Feynman's writings share that quality. Some academics are more pedagogically gifted than others.
It's also interesting to consider what a medium (e.g., complex academic text vs. fiction) signals about the the author's apparent beliefs about readers.
In my first research internship some years ago, when I was asked to present my progress to the research group, I tried to present the material tutoring-style, with questions (ten-second exercises) so listeners could check their comprehension and could actually get fluent with the formalism I was using. I was (correctly) told that I shouldn't use such formats as they are insulting to the listeners. A tutorial style exposition implies that the speaker is in charge of the "correct" answers and is competent to structure the listeners' learning, while standard academic presentations present research to be looked over and commented on by listeners, with a presumption that the listeners will understand.
Perhaps some of the annoyance at Eliezer's use of fiction stems from readers' insult at the view such fiction signals of their abilities or of their non-peerhood with Eliezer (and not from concern about fiction's effects on readers). If readers can instead claim to be members of a blog filled with math and complex, passive-voice sentences, they can signal that they can read such material.
"Manipulation" seems closer to the idea in certain key ways. Also, "beguilement" and "seduction".