I think you overestimate how willing people are to read very long articles without a great reason to think it’ll be worth their time at the beginning. You have three meaty paragraphs before you even start what seems to be the article proper, and then that’s a summary of an abstract idea. Something this long needs to start with an awesome story.
There’s a lot of very interesting stuff here. But I only figured that out after reading the comments, copying your post to a local editor to see if there was an obvious way to cut it to 1⁄3 the length, and then reading carefully. Your writing is… unskimmable to me. Usually I can skim a paragraph, get the gist, and go back to an individual sentence or two if I feel I missed something. I could not do that with this article.
Your feedback on skimmability seems potentially really useful—that wasn’t even something I was thinking of before. I’m going to try to improve on that point, though I also suspect that I’m not going to succeed as much and as quickly as I would like. I may do some research on writing styles and tips. Thanks very much for pointing this out.
Was there an obvious way to cut it to 1⁄3 the length? If a professional editor was able to do so and you were willing to send it to me, that would probably be really helpful for me.
Was there an obvious way to cut it to 1⁄3 the length? If a professional editor was able to do so and you were willing to send it to me, that would probably be really helpful for me.
1⁄3 the length would still be far too long. Does the following leave anything out?
To improve your performance in any sphere:
Observe and learn what works.
Most goals are subgoals of higher goals. Conflicts among them can often be resolved by looking for the higher goals and asking what will really serve them.
There. 41 words instead of 4388. The remainder is unsupported folk psychology, repetition, and superfluous elaboration.
I think you need to improve your own writing, rather than using someone else to fix it up afterwards. A programmer has to fix his own code, and a writer likewise.
The role of social considerations in rationality and dysrationality
More information on how the unconscious works (and what it can do when we understand it)
A more detailed overview of the ways we can improve unconscious thinking, along with examples of actually doing so.
Information on the process of investigating this thinking
The remainder is unsupported folk psychology, repetition, and superfluous elaboration.
There should be a “looks like” in there somewhere, at least with regard to “unsupported folk psychology” (repetition and superfluous elaboration...I wouldn’t put the latter in those terms, but those may be an issue). Again, this may look similar in ways. But it is the process of multiple revisions of the ideas, looking for different ways to think of them that help me use them more productively, cutting things down to their fundamentals and removing elements from the model that didn’t buy me any bits of prediction. (Mostly) everything here is load-bearing.
I think you need to improve your own writing, rather than using someone else to fix it up afterwards. A programmer has to fix his own code, and a writer likewise.
Obviously that would be better! While I’ve received moderate compliments on my writing in the past, I obviously wish I was much better. I would love to be able to phrase an idea more clearly, simply, and accurately, while keeping the reader engaged and perhaps even entertained. These posts are my current best efforts, and I know that despite this the writing isn’t going to be that excellent, and that a more experienced writer would probably be able to put together something much better, and with less work. I would love to know how to do that!
But that doesn’t mean I’m not going to try and use whatever tools I might find available to improve that writing, such as looking at a professionally-edited version of the very thing I worked on, if I get a chance to read something like that.
One exercise that I found extremely helpful when learning how to condense my writing was:
Try to include a verb and a direct object in each of your bullet points.
Try to make sure each of your bullet points makes a falsifiable claim.
This forces you to confirm that each of your major points has substantive, useful content. “The process behind deprecation” is an excellent ‘note to self’ to remind you of what your topic is, but a person can write that note without having the faintest idea what the process behind deprecation is. My Bayesian prior for notes like that after reading thousands of social science articles is that you’re most likely going to go on for pages and pages without drawing any firm conclusions. If, instead, you make a bold claim like “Deprecation is strictly dominated by other social strategies,” then I predict that you’re going to try to say something interesting. I may disagree with your methods or your evidence, but at least I can gauge whether your effort, if successful, would be of interest to me.
I think you’re a good writer, in that you form sentences well, and you understand how the language works, and your prose is not stilted or boring. The problem I personally had, mostly with the previous two entries in this series, was that the “meat”—the interesting bits telling me what you had concluded, and why, and how to apply it, and how (specifically) you have applied it - seemed very spread out among a lot of filler or elaboration. I couldn’t tell what you were eventually going to arrive at, and whether it’d be of use or interest to me. Too much generality, perhaps: compare “this made my life better” with “by doing X I caught myself thinking Y and changed this to result in the accomplishment of Z.”
I tell you this only in case you are interested in constructive criticism from yet another perspective; some undoubtedly consider the things I have mentioned virtues in an author. In any case, I have upvoted this article; it doesn’t deserve a negative score, I think—long-winded, maybe; poorly done or actively irrational, certainly not. The ideas are interesting, the methodology is reasonable, and the effort is appreciated.
I think you may have missed 3. Most people optimise for perceived status
Or was that included under ‘folk psychology’?
Possibly, or I just thought it not worth retaining. “People care what other people think of them” is the same idea, but without the LessWrong jargon, and as such a truism known to everyone.
The key thing is this: when a rationalist is investigating a bias or some irrational behavior, they may notice that there seems to be a social influence on their thinking, think to themselves “well that’s obviously silly and wrong”, and then stop there. They go on believing that rationality has to be painful, that we have to do something to overpower these instincts, and that the only way to succeed is to look for ways to trick their unconscious mind into having a belief that seems more appropriate.
An alternative to this approach is to keep going, to look deeper at what’s really going on, spend hours or days looking for something sensible that the unconscious could possibly be doing, until enough pieces come together and suddenly you say “Oh. That’s what’s going on.” And then the most important part, you can solve the problem, so that it’s not hard or painful anymore.
Or for something more direct and operable, if you notice that your aversion to something is that you don’t want to look stupid, rather than try to power through it, look for ways that you could do the same thing without looking stupid. In fact if you look at a lot of the useful rationality techniques, the way they help us out is by doing this very thing.
when a rationalist is investigating a bias or some irrational behavior, they may notice that there seems to be a social influence on their thinking, think to themselves “well that’s obviously silly and wrong”, and then stop there. They go on believing that rationality has to be painful, that we have to do something to overpower these instincts, and that the only way to succeed is to look for ways to trick their unconscious mind into having a belief that seems more appropriate.
I can’t say I’ve particularly noticed this.
An alternative to this approach is to keep going, to look deeper at what’s really going on
That’s nicely done! Clear, concise, and immediately applicable. I think Frank himself is an intelligent person with good and interesting ideas, but the “meat” of these posts seems spread out among a lot of filler/elaboration—possibly why they’re hard to skim. I wasn’t even sure, for quite a while, what the whole series was really about, beyond “general self-improvement.”
This latest article is much more “functional” than the previous two, though, so I think we’re moving in the right direction.
One thing your comment brings to mind—Frank notes something about unconscious mental processes being trainable, and the suggestion is that one can train them to be rational, or at least more accurate. (If I remember correctly.) Is this idea included your comment? Perhaps under “folk psychology”?
It seems like an interesting concept, though I was unable to find any instruction on how to actually accomplish it. (But I haven’t looked too hard yet.)
One thing your comment brings to mind—Frank notes something about unconscious mental processes being trainable, and the suggestion is that one can train them to be rational, or at least more accurate. (If I remember correctly.) Is this idea included your comment? Perhaps under “folk psychology”?
I tuned out all that stuff about “the unconscious”. How does Frank know that “our unconscious thinking is actually very powerful, very intelligent, and fairly sensible”? That it is “extremely powerful, doing massive amounts of computation very quickly”? And yet, “When they make a mistake they have no way of telling that they made a mistake”? Where does this come from? What does he mean by “the unconscious”?
When he says “When I realize this disconnect and see how the information about underlying frequency was shifted as it passed through my sources, I unconsciously come to a better estimate of how often something happens”, what is the word “unconscious” doing there? Looks like a description of something conscious.
By “the unconscious” I mean the mental operations we perform without getting internal mental feedback about the process of the operation.
That’s not very concrete. The most widely recognized extension of this part of reality is emotions we don’t understand the reasons for, along with other mysterious-by-default things like why we spend a long time mentally reviewing our stated positions. We can simply ask “Why do I feel that way, and why do I spend my time that way?” This question doesn’t require any mention of unconscious thinking, or thinking at all. At this state of knowledge, the answer could conceivably involve any number of mechanisms, and those mechanisms may not be mental.
But in my own attempts to answer these questions, the most efficient way to model the source of those things is actually to model them as a mind. (I say “efficient” to emphasize a goal of using the concept, but the model appears to be accurate as well.) By a “mind”, I mean something that has a model of the world, that takes in evidence it receives about that world, that performs a very great amount of inference on that model and evidence, and even undertakes strategic thinking in the attempt to reach goals. In other words, the answer to why we feel and do those things is that there’s a genuine optimization process there, and the feelings and actions are its output.
We receive the conclusions of this part of our thinking, but we don’t have feedback about the thoughts taking place there. The system-of-us does not receive as input the process of these thoughts, and this is what distinguishes conscious and UNconscious thinking.
It also really needs to start with some more concrete content. I’m not sure how interested I am that you’ve theorised about your experiences until I’m convinced there’s something of value in those experiences; and having your “core content” would really help with that.
If you’re writing a textbook about an established field, then you can afford to start with the theory; it may even be helpful, but in this case I think you should start at a lower level of abstraction.
(Upvoted) [Is it poor etiquette to say so? I recall seeing it in the past but I’m not familiar with online LW etiquette.]
Yeah, I think that’s a very good point. The things a model should be built on include actual uses of that model, some weight that it’s lifting. In this case I’m not actually sure that starting with the overview was not the way to go; it may well not have been, but many of the particular points draw from a larger model that might differ from some common beliefs, such as that people are intrinsically incoherent kludges or that our unconscious instincts can’t and don’t respond to subtle and genuinely important details in the world around us. If I included ideas like “Just about everything people do makes a fair deal of sense” in a concrete model without providing more information about that general claim—and that fact that it’s actually present throughout all of the ideas I’ve been using—I think it may be taken as an arbitrary and unjustified assumption, rather than something that’s come up time and time again in my attempts to understand what people are doing.
I think you overestimate how willing people are to read very long articles without a great reason to think it’ll be worth their time at the beginning. You have three meaty paragraphs before you even start what seems to be the article proper, and then that’s a summary of an abstract idea. Something this long needs to start with an awesome story.
That...would be why I posted the first two posts :)
But I think you mean that it would be good to have something short at the beginning of this post. That would probably be a good idea.
So you’re hoping people will read two other long articles so they know that this long article will be worth their time?
There’s a lot of very interesting stuff here. But I only figured that out after reading the comments, copying your post to a local editor to see if there was an obvious way to cut it to 1⁄3 the length, and then reading carefully. Your writing is… unskimmable to me. Usually I can skim a paragraph, get the gist, and go back to an individual sentence or two if I feel I missed something. I could not do that with this article.
Your feedback on skimmability seems potentially really useful—that wasn’t even something I was thinking of before. I’m going to try to improve on that point, though I also suspect that I’m not going to succeed as much and as quickly as I would like. I may do some research on writing styles and tips. Thanks very much for pointing this out.
Was there an obvious way to cut it to 1⁄3 the length? If a professional editor was able to do so and you were willing to send it to me, that would probably be really helpful for me.
1⁄3 the length would still be far too long. Does the following leave anything out?
To improve your performance in any sphere:
Observe and learn what works.
Most goals are subgoals of higher goals. Conflicts among them can often be resolved by looking for the higher goals and asking what will really serve them.
There. 41 words instead of 4388. The remainder is unsupported folk psychology, repetition, and superfluous elaboration.
I think you need to improve your own writing, rather than using someone else to fix it up afterwards. A programmer has to fix his own code, and a writer likewise.
I would summarize the main points as:
The process behind deprecation
The role of social considerations in rationality and dysrationality
More information on how the unconscious works (and what it can do when we understand it)
A more detailed overview of the ways we can improve unconscious thinking, along with examples of actually doing so.
Information on the process of investigating this thinking
There should be a “looks like” in there somewhere, at least with regard to “unsupported folk psychology” (repetition and superfluous elaboration...I wouldn’t put the latter in those terms, but those may be an issue). Again, this may look similar in ways. But it is the process of multiple revisions of the ideas, looking for different ways to think of them that help me use them more productively, cutting things down to their fundamentals and removing elements from the model that didn’t buy me any bits of prediction. (Mostly) everything here is load-bearing.
Obviously that would be better! While I’ve received moderate compliments on my writing in the past, I obviously wish I was much better. I would love to be able to phrase an idea more clearly, simply, and accurately, while keeping the reader engaged and perhaps even entertained. These posts are my current best efforts, and I know that despite this the writing isn’t going to be that excellent, and that a more experienced writer would probably be able to put together something much better, and with less work. I would love to know how to do that!
But that doesn’t mean I’m not going to try and use whatever tools I might find available to improve that writing, such as looking at a professionally-edited version of the very thing I worked on, if I get a chance to read something like that.
One exercise that I found extremely helpful when learning how to condense my writing was:
Try to include a verb and a direct object in each of your bullet points.
Try to make sure each of your bullet points makes a falsifiable claim.
This forces you to confirm that each of your major points has substantive, useful content. “The process behind deprecation” is an excellent ‘note to self’ to remind you of what your topic is, but a person can write that note without having the faintest idea what the process behind deprecation is. My Bayesian prior for notes like that after reading thousands of social science articles is that you’re most likely going to go on for pages and pages without drawing any firm conclusions. If, instead, you make a bold claim like “Deprecation is strictly dominated by other social strategies,” then I predict that you’re going to try to say something interesting. I may disagree with your methods or your evidence, but at least I can gauge whether your effort, if successful, would be of interest to me.
I think you’re a good writer, in that you form sentences well, and you understand how the language works, and your prose is not stilted or boring. The problem I personally had, mostly with the previous two entries in this series, was that the “meat”—the interesting bits telling me what you had concluded, and why, and how to apply it, and how (specifically) you have applied it - seemed very spread out among a lot of filler or elaboration. I couldn’t tell what you were eventually going to arrive at, and whether it’d be of use or interest to me. Too much generality, perhaps: compare “this made my life better” with “by doing X I caught myself thinking Y and changed this to result in the accomplishment of Z.”
I tell you this only in case you are interested in constructive criticism from yet another perspective; some undoubtedly consider the things I have mentioned virtues in an author. In any case, I have upvoted this article; it doesn’t deserve a negative score, I think—long-winded, maybe; poorly done or actively irrational, certainly not. The ideas are interesting, the methodology is reasonable, and the effort is appreciated.
I think you may have missed 3. Most people optimise for perceived status
Or was that included under ‘folk psychology’?
Possibly, or I just thought it not worth retaining. “People care what other people think of them” is the same idea, but without the LessWrong jargon, and as such a truism known to everyone.
The key thing is this: when a rationalist is investigating a bias or some irrational behavior, they may notice that there seems to be a social influence on their thinking, think to themselves “well that’s obviously silly and wrong”, and then stop there. They go on believing that rationality has to be painful, that we have to do something to overpower these instincts, and that the only way to succeed is to look for ways to trick their unconscious mind into having a belief that seems more appropriate.
An alternative to this approach is to keep going, to look deeper at what’s really going on, spend hours or days looking for something sensible that the unconscious could possibly be doing, until enough pieces come together and suddenly you say “Oh. That’s what’s going on.” And then the most important part, you can solve the problem, so that it’s not hard or painful anymore.
Or for something more direct and operable, if you notice that your aversion to something is that you don’t want to look stupid, rather than try to power through it, look for ways that you could do the same thing without looking stupid. In fact if you look at a lot of the useful rationality techniques, the way they help us out is by doing this very thing.
A blog that takes a rational approach to writing improvement is Disputed Issues
I can’t say I’ve particularly noticed this.
This is what I am more familiar with myself.
Cool!
That’s nicely done! Clear, concise, and immediately applicable. I think Frank himself is an intelligent person with good and interesting ideas, but the “meat” of these posts seems spread out among a lot of filler/elaboration—possibly why they’re hard to skim. I wasn’t even sure, for quite a while, what the whole series was really about, beyond “general self-improvement.”
This latest article is much more “functional” than the previous two, though, so I think we’re moving in the right direction.
One thing your comment brings to mind—Frank notes something about unconscious mental processes being trainable, and the suggestion is that one can train them to be rational, or at least more accurate. (If I remember correctly.) Is this idea included your comment? Perhaps under “folk psychology”?
It seems like an interesting concept, though I was unable to find any instruction on how to actually accomplish it. (But I haven’t looked too hard yet.)
I tuned out all that stuff about “the unconscious”. How does Frank know that “our unconscious thinking is actually very powerful, very intelligent, and fairly sensible”? That it is “extremely powerful, doing massive amounts of computation very quickly”? And yet, “When they make a mistake they have no way of telling that they made a mistake”? Where does this come from? What does he mean by “the unconscious”?
When he says “When I realize this disconnect and see how the information about underlying frequency was shifted as it passed through my sources, I unconsciously come to a better estimate of how often something happens”, what is the word “unconscious” doing there? Looks like a description of something conscious.
By “the unconscious” I mean the mental operations we perform without getting internal mental feedback about the process of the operation.
That’s not very concrete. The most widely recognized extension of this part of reality is emotions we don’t understand the reasons for, along with other mysterious-by-default things like why we spend a long time mentally reviewing our stated positions. We can simply ask “Why do I feel that way, and why do I spend my time that way?” This question doesn’t require any mention of unconscious thinking, or thinking at all. At this state of knowledge, the answer could conceivably involve any number of mechanisms, and those mechanisms may not be mental.
But in my own attempts to answer these questions, the most efficient way to model the source of those things is actually to model them as a mind. (I say “efficient” to emphasize a goal of using the concept, but the model appears to be accurate as well.) By a “mind”, I mean something that has a model of the world, that takes in evidence it receives about that world, that performs a very great amount of inference on that model and evidence, and even undertakes strategic thinking in the attempt to reach goals. In other words, the answer to why we feel and do those things is that there’s a genuine optimization process there, and the feelings and actions are its output.
We receive the conclusions of this part of our thinking, but we don’t have feedback about the thoughts taking place there. The system-of-us does not receive as input the process of these thoughts, and this is what distinguishes conscious and UNconscious thinking.
It also really needs to start with some more concrete content. I’m not sure how interested I am that you’ve theorised about your experiences until I’m convinced there’s something of value in those experiences; and having your “core content” would really help with that.
If you’re writing a textbook about an established field, then you can afford to start with the theory; it may even be helpful, but in this case I think you should start at a lower level of abstraction.
(Upvoted) [Is it poor etiquette to say so? I recall seeing it in the past but I’m not familiar with online LW etiquette.]
Yeah, I think that’s a very good point. The things a model should be built on include actual uses of that model, some weight that it’s lifting. In this case I’m not actually sure that starting with the overview was not the way to go; it may well not have been, but many of the particular points draw from a larger model that might differ from some common beliefs, such as that people are intrinsically incoherent kludges or that our unconscious instincts can’t and don’t respond to subtle and genuinely important details in the world around us. If I included ideas like “Just about everything people do makes a fair deal of sense” in a concrete model without providing more information about that general claim—and that fact that it’s actually present throughout all of the ideas I’ve been using—I think it may be taken as an arbitrary and unjustified assumption, rather than something that’s come up time and time again in my attempts to understand what people are doing.
It would be good to have something short, period.
Hmm, a good point :)