Reposting my Inkhaven post on ontology of advice here.
Are you interested in learning a new field, whether it’s programming, writing, or how to win Paper Mario games? Have you searched for lots of advice and couldn’t choose which advice to follow? Worse, have you tried to follow other people’s Wise Sounding Advice and ended up worse than where you started?
Alternatively, have you tried to give useful advice distilling your hard-earning learnings and only to realize it fell on deaf ears? Or perhaps you’ve given advice that filled a much-needed hole that you now regret giving?
If so, this post is for you!
While this post is far from exhaustive, I hope reading it can help you a) identify the type of advice you want to give and receive and b) recognize and try to avoid common failure modes!
Here are 7 semi-distinct categories of good advice. Some good advice mixes and matches between the categories, whereas others are more “purist” and just tries to do one style well.
I. The Master Key
This is where someone who deeply understands a field tries to impart the central tenets/frames of a field so newbies can decide whether the model is a good fit for what they want to do. And the rest of the article/book/lecture will be a combination of explaining the model and why they believe it’s true, and examples to get the learner to deeply understand the model. Eg “Focus on the user” as the central dictum in tech startups, or understanding and groking the Popperian framework for science.2
My previous post, How to Win New Board Games, is an unusually pure example, where I spend 2000 words hammering different variations and instantiations of a single idea (“Understand the win condition, and play to win”).
In writing advice, Clear and Simple as the Truth (by Thomas and Turner, review + extension here) works in this way as well, doing their best to model how to write well in the Classic Style.
II. The Toolkit
“When art critics get together, they talk about form and structure and meaning. When artists get together, they talk about where you can buy cheap turpentine.” - Pablo Picasso, supposedly3
The motivating theology is something like “reality has a surprising amount of detail“, and you want to impart these details onto novices. In tech startups, this could be a list of 20 tips. In videogames, this could be a youtube video that goes through a bunch of different tips.
My first post this month, How to Write Fast, Weird, and Well, is mostly in this vein, with a collection of loosely structured tips that I personally have found to be the most helpful in improving myself as an online writer.
III.The War Stories
Teaching through stories and examples rather than principles or tips. Business schools love this approach (case studies), as do many mentors who say “let me tell you about the time I...” The idea is that patterns emerge from concrete situations better than from abstract rules. While not stereotypically considered an “advice book,” the nonfiction book I’m currently the most engrossed in, Skunk Works, is written almost entirely as a collection of war stories.
For games, this could be videos of professional streams. In writing, this would be books like Stephen King’s “On Writing”, which weaves memoir with advice.
IV. The Mirror(s)
Can you think of good questions to guide your students, rather than declarative statements?
The minimal viable product for this type of advice is just being a “pure” mirror. Have you tried just asking your advisee what they’re thinking of doing and what they think the biggest flaws with their plans are? Otherwise known as “Socratic ducking,” this is where your questions essentially just mirror your advisee’s thoughts and you don’t try to interject any of your own opinions or tastes in the manner. Surprisingly useful!
In more advanced “mirror strategies,” the advisor’s questions might serve more as prisms, lenses, or funhouse mirrors. Can you do better than a pure mirror? Can you think of some common failure modes in your field and ask your advisees pointed questions so they can address those? Can you reflect your subtle judgments in taste and prioritization and reframe your tips into central questions of interest?
Coaching and therapy often works this way. Instead of “focus on the user,” it’s “who do you think will use this and what do they need?”
There’s a spectrum of mirror purity vs detail. In the most detailed end, maybe you should just give your normal advice but say it with an upwards inflection so it sounds like a question?
V. The Permission Slip
This is advice like “be yourself” or “chase your dreams.” This might initially seem to be semantically useless, but there’s real value in giving license for people to do (socially positive) things they kind of want to do anyway. In Effective Altruism, this could be something like “when in doubt, just apply (to socially positive jobs)” or “you don’t need permission to do the right thing.”
In writing advice, this could be seemingly trivial advice like telling aspiring writers to just start writing, or telling writers worried about style that your style ought to be an expression of your personality.
VI. The Diagnosis
Advice that helps you figure out what specific bottleneck is, or the specific thing (among a suite of options) that would help you the most. Some product reviews might look like this.
The post you’re reading right now is also somewhat in this vein! Hopefully after reading the post, you’d have a better sense of what types of advice you’d find most useful to give or receive.
VII. The Landmines
Advice for what not to do, scary things newbies should avoid, etc. In most fields, learning the Landmines is supplementary advice. But in certain high-stakes domains where beginners have enough agency to seriously hurt themselves or others like firearms practice, outdoor climbing, or lab chemistry, it’s the prerequisite advice.
Integration
Of course, many advice posts/essays/books might integrate two or more of the above categories. For example, my Field Guide to Writing Styles post mixes a meta-framework (master Master Key) for choosing which framework/frameworks to write in (Diagnostic), with specific frameworks/writing styles you might want to write in. While in some sense this is more sophisticated and interesting to write (and hopefully to read!) than advice in a single “pure” category, it also likely suffers from being more scattered and confusing. So there are real tradeoffs.
Is the above ontology complete? What am I missing? Tell me in the comments!4
Core Ways Advice Can Be Bad
There are an endless plethora of ways advice can be bad, and fail to deliver value to the intended audience (eg the advice is ignored), or deliver anti-value to the intended audience (the advice is taken, and taking the advice is worse than not taking it).
In this article, I will just focus on the biggest ones.
The biggest three reasons are that the advisor can fail at metacognition, the advice can fail to center the advisee, or the advice can otherwise fail to be persuasive
Failures of Metacognition
The advisor can fail at metacognition, and do not know the limits of their knowledge
Most simply, the advice could straightforwardly be wrong
They might confuse survivorship bias/luck with skill
They might confuse innate talent with specific choices they made, or specific choices that work well given a specific skillset/talents they have
The advice can be in the wrong category for the topic of interest
Back to the ontology above, an advice essay could try to fit a “master key” ontology for a field that essentially does not have a (known) master key. Ie, they might try to force a “Central Dogma” when there isn’t one, or there isn’t a known one.
I was at one point a fairly good self-taught programmer, and worked professionally in FAANG for a couple of years.
As far as I know5, programming does not have a central dogma, nor have any of the attempts I’ve seen other people try to claim a central dogma for programming be plausible
Instead, good advice for programming looks like a collection of tips, war stories, or a permission slip.
Advice for getting good at programming mostly looks like advice for getting good at things in general, plus extremely specific advice for things like which IDE to use, (in Current Year) which AI assistant to employ, etc.
Advice as identity: Some people become “the person who gives X advice” or “person who’s good advising people” and then keep giving advice even when circumstances change or they get contrary evidence. They’re too attached to their signature advice, and insufficiently attuned to reality.
Many of these failure modes can be alleviated through clearer thinking and better rationality.
The advisor might not realize that good advice should is about the interaction of information with the (intended) advisee, than some objective fact about the world
The advisor might realize that they should center the advisee, but not understand the advisee’s problems well enough to be useful
In the context of de-centered internet advice, this could also come from insufficiently accurate audience segmentation, or bad luck
The advice can be overly academic and descriptive, and insufficiently practical.
The advice can actually be more about the advisor’s neuroses than about the topic in question
For example the advisor could spend too much of their time proving their expertise or other desiderata (intelligence, interestingness, sexual appeal, or other traits not very relevant to the advice at hand)
For example, writing advice proving how the writer is smart, or business advice justifying the advisor’s past choices.
Nietzsche’s great at self-promotion, but not the best at meta-cognition or audience awareness.
The advisor could be assuaging their own emotional problems with the topic in question
The advice can be in the wrong category for the audience of interest
For example, first time advice for (most) novices should not look like Landmines
Telling new chess players how not to play chess will just confuse them, since they don’t know how to play chess to begin with.
(The main exception, as previously mentioned, are fields where safety is critical)
Incentive blindness
The advisor might be blind to the ways in which they, individually or collectively, are incentivized to give specific advice in ways that are not always in the advisees’ interests
eg professors incentivized to tell young people to go to graduate school, Paul Graham advising young/ambitious/technically talented people to found a startup
Many of these failure modes can be alleviated through greater empathy and revealed desire to help others.
The advice can fail to be persuasive
This category is somewhat less bad than the previous two categories, as the damage is limited.
The advice can be told in a boring/uninteresting way despite being correct
The advice can be correct and exciting without being sufficiently justified to be persuasive
The advice can be correct and persuasive if it’s ever read, but buried in ways that never gets accessed by the intended audience
Many of these failure modes can be alleviated by improvements in writing qualityin general. They can also be reduced via learning about other (ethical/semi-ethical) forms of self-promotion, which I have not yet cracked but hope to do so one day (and will gladly share on the blog).
What do you think? Are there specific categories of advice you're particularly drawn to? Are there (good) categories of advice that I missed? Tell us in the comments!
I realize this is a bit of a trap/local optima to be stuck in, so starting tomorrow, I’m calling for a personal moratorium on publishing advice posts until at least the end of this week!
(For simplicity I’m ignoring high-context/specific/very situational advice, like specific suggested edits on a blog post, or an experienced programmer providing code review). I’m also excluding exercise “books” like Leetcode or writing prompts.
I was a decent but not great (by Silicon Valley standards) programmer, so it’s possible there is a central dogma I was not aware of. But I also read dozens of books at programming and worked at Google for almost 2 years, so at the very least the memescape did not try very hard to impress upon me a central dogma
Reposting my Inkhaven post on ontology of advice here.
Are you interested in learning a new field, whether it’s programming, writing, or how to win Paper Mario games? Have you searched for lots of advice and couldn’t choose which advice to follow? Worse, have you tried to follow other people’s Wise Sounding Advice and ended up worse than where you started?
Alternatively, have you tried to give useful advice distilling your hard-earning learnings and only to realize it fell on deaf ears? Or perhaps you’ve given advice that filled a much-needed hole that you now regret giving?
If so, this post is for you!
While this post is far from exhaustive, I hope reading it can help you a) identify the type of advice you want to give and receive and b) recognize and try to avoid common failure modes!
7 Categories of Good Advice
Source: https://englishlive.ef.com/en/blog/english-in-the-real-world/5-simple-ways-give-advice-english/
Here are 7 semi-distinct categories of good advice. Some good advice mixes and matches between the categories, whereas others are more “purist” and just tries to do one style well.
I. The Master Key
This is where someone who deeply understands a field tries to impart the central tenets/frames of a field so newbies can decide whether the model is a good fit for what they want to do. And the rest of the article/book/lecture will be a combination of explaining the model and why they believe it’s true, and examples to get the learner to deeply understand the model. Eg “Focus on the user” as the central dictum in tech startups, or understanding and groking the Popperian framework for science.2
My previous post, How to Win New Board Games, is an unusually pure example, where I spend 2000 words hammering different variations and instantiations of a single idea (“Understand the win condition, and play to win”).
In writing advice, Clear and Simple as the Truth (by Thomas and Turner, review + extension here) works in this way as well, doing their best to model how to write well in the Classic Style.
II. The Toolkit
“When art critics get together, they talk about form and structure and meaning. When artists get together, they talk about where you can buy cheap turpentine.” - Pablo Picasso, supposedly3
The motivating theology is something like “reality has a surprising amount of detail“, and you want to impart these details onto novices. In tech startups, this could be a list of 20 tips. In videogames, this could be a youtube video that goes through a bunch of different tips.
My first post this month, How to Write Fast, Weird, and Well, is mostly in this vein, with a collection of loosely structured tips that I personally have found to be the most helpful in improving myself as an online writer.
III.The War Stories
Teaching through stories and examples rather than principles or tips. Business schools love this approach (case studies), as do many mentors who say “let me tell you about the time I...” The idea is that patterns emerge from concrete situations better than from abstract rules. While not stereotypically considered an “advice book,” the nonfiction book I’m currently the most engrossed in, Skunk Works, is written almost entirely as a collection of war stories.
For games, this could be videos of professional streams. In writing, this would be books like Stephen King’s “On Writing”, which weaves memoir with advice.
IV. The Mirror(s)
Can you think of good questions to guide your students, rather than declarative statements?
The minimal viable product for this type of advice is just being a “pure” mirror. Have you tried just asking your advisee what they’re thinking of doing and what they think the biggest flaws with their plans are? Otherwise known as “Socratic ducking,” this is where your questions essentially just mirror your advisee’s thoughts and you don’t try to interject any of your own opinions or tastes in the manner. Surprisingly useful!
In more advanced “mirror strategies,” the advisor’s questions might serve more as prisms, lenses, or funhouse mirrors. Can you do better than a pure mirror? Can you think of some common failure modes in your field and ask your advisees pointed questions so they can address those? Can you reflect your subtle judgments in taste and prioritization and reframe your tips into central questions of interest?
Coaching and therapy often works this way. Instead of “focus on the user,” it’s “who do you think will use this and what do they need?”
There’s a spectrum of mirror purity vs detail. In the most detailed end, maybe you should just give your normal advice but say it with an upwards inflection so it sounds like a question?
V. The Permission Slip
This is advice like “be yourself” or “chase your dreams.” This might initially seem to be semantically useless, but there’s real value in giving license for people to do (socially positive) things they kind of want to do anyway. In Effective Altruism, this could be something like “when in doubt, just apply (to socially positive jobs)” or “you don’t need permission to do the right thing.”
In writing advice, this could be seemingly trivial advice like telling aspiring writers to just start writing, or telling writers worried about style that your style ought to be an expression of your personality.
VI. The Diagnosis
Advice that helps you figure out what specific bottleneck is, or the specific thing (among a suite of options) that would help you the most. Some product reviews might look like this.
The post you’re reading right now is also somewhat in this vein! Hopefully after reading the post, you’d have a better sense of what types of advice you’d find most useful to give or receive.
VII. The Landmines
Advice for what not to do, scary things newbies should avoid, etc. In most fields, learning the Landmines is supplementary advice. But in certain high-stakes domains where beginners have enough agency to seriously hurt themselves or others like firearms practice, outdoor climbing, or lab chemistry, it’s the prerequisite advice.
Integration
Of course, many advice posts/essays/books might integrate two or more of the above categories. For example, my Field Guide to Writing Styles post mixes a meta-framework (master Master Key) for choosing which framework/frameworks to write in (Diagnostic), with specific frameworks/writing styles you might want to write in. While in some sense this is more sophisticated and interesting to write (and hopefully to read!) than advice in a single “pure” category, it also likely suffers from being more scattered and confusing. So there are real tradeoffs.
Is the above ontology complete? What am I missing? Tell me in the comments!4
Core Ways Advice Can Be Bad
There are an endless plethora of ways advice can be bad, and fail to deliver value to the intended audience (eg the advice is ignored), or deliver anti-value to the intended audience (the advice is taken, and taking the advice is worse than not taking it).
In this article, I will just focus on the biggest ones.
The biggest three reasons are that the advisor can fail at metacognition, the advice can fail to center the advisee, or the advice can otherwise fail to be persuasive
Failures of Metacognition
The advisor can fail at metacognition, and do not know the limits of their knowledge
Many of these failure modes can be alleviated through clearer thinking and better rationality.
Failures of audience centering
The advice can fail to center the advisee.
Nietzsche’s great at self-promotion, but not the best at meta-cognition or audience awareness.
The advice can be in the wrong category for the audience of interest
Many of these failure modes can be alleviated through greater empathy and revealed desire to help others.
The advice can fail to be persuasive
This category is somewhat less bad than the previous two categories, as the damage is limited.
Many of these failure modes can be alleviated by improvements in writing quality in general. They can also be reduced via learning about other (ethical/semi-ethical) forms of self-promotion, which I have not yet cracked but hope to do so one day (and will gladly share on the blog).
What do you think? Are there specific categories of advice you're particularly drawn to? Are there (good) categories of advice that I missed? Tell us in the comments!
1
I realize this is a bit of a trap/local optima to be stuck in, so starting tomorrow, I’m calling for a personal moratorium on publishing advice posts until at least the end of this week!
2
Sometimes the central tenet can be explained “well enough” in one sentence, like “focus on the user.” Often, it cannot be.
3
Apocryphal
4
(For simplicity I’m ignoring high-context/specific/very situational advice, like specific suggested edits on a blog post, or an experienced programmer providing code review). I’m also excluding exercise “books” like Leetcode or writing prompts.
5
I was a decent but not great (by Silicon Valley standards) programmer, so it’s possible there is a central dogma I was not aware of. But I also read dozens of books at programming and worked at Google for almost 2 years, so at the very least the memescape did not try very hard to impress upon me a central dogma