Scheduling criticism

"Two times two is four ‘tis true But too empty and too trite I would rather seek a clue To some matters not so light."

— W. Busch, Schein und Sein (Karl Popper's translation)

Being in control of when, how, and what you get criticism on is a vital part of the growth of knowledge.

Scheduling criticism isn't being closed-minded. It's the only way to deal with the fact that there is an infinity of possible criticism, and no mechanical way to make that infinity smaller. You need some way of choosing what is useful.

The self-development idea of 'growth mindset' says: learn from criticism, don't ignore it.

There is an exaggerated version of this idea, namely: "Because criticism is not actually bad and in fact helps you learn, criticism is always good (or at least neutral)."

The grain of truth in this is: Identifying errors gives you access to the problems in your ideas. Any criticism could help. Finding conflicts between your ideas is the first step to solving them. A criticism on its own, as an idea in the ether, is only information — and this is information that can be used to make progress.

So why isn't criticism purely a good thing? How could noticing errors be bad?

Well, it's true that someone might have an irrational aversion to making mistakes (perhaps after having been penalised in school for getting things wrong), which causes them to be averse to even noticing mistakes.

But if we got over this, and didn't take criticism personally, wouldn't criticism just be a gift?

No. Criticism, when unwanted, can effectively destroy the means of error correction and the growth of knowledge. Put differently: it can structurally cripple thinking.

When I say "unwanted", I'm not referring to emotional reactions to criticism. One can feel bad about criticism by taking it personally — and it is possible to avoid this kind of bad feeling, by learning to take criticism better — but that isn't what I'm talking about here. That's a personal hangup, whereas I'm pointing at something more fundamental.

Criticism is how we learn

Criticism is part of how we learn. We have a problem we're trying to solve or a question we're trying answer, we come up with potential answers, and we criticise those answers to narrow them down or polish them. Once we have a promising solution, we criticise it further to understand it better and to test that it actually works as a solution.

We need to criticise ideas to understand them at all. (This often happens subconsciously. Criticism needn't be at the explicit level of awareness; though making criticisms explicit can help make them clearer. All thinking has a large inexplicit component, including criticism.) If we didn't criticise ideas, we wouldn't know things like what else the idea solves, where it's applicable vs we need a different concept, why that solution instead of something else, and what it even means.

Criticism also comes into play earlier, when first coming up with potential answers. If it were 'anything goes', they would be random useless nonsense. (A bit like dream logic. though even dreams must have some kind of criticism filters to make as much sense as they do.)

For instance: If you're modelling the mind of your professor, you might come up with different guesses of the solution than if you were modelling the mind of your favourite author. Those two models give different kinds of criticism. Likewise if you imagine presenting to different audiences. The ideas you come up with will be affected by what kind of criticism you're anticipating. So, even when generating ideas in the first place, the effects of criticism are always in the background.

If we're not in control of what criticism we think about, if we outsource this critical process to someone else, we can't really understand the idea. We might get the bottom line or some individual applications we've been told. But that's only useful for things like passing exams or repeating propaganda.

The criticisms you consider will depend on what your specific idea-conflicts are. Each person is really different. All individuals have unique misconceptions, assumptions, background knowledge, connections between ideas and ideas in conflict.

Someone who thinks animals have sentience might find that conflicts with eating meat, whereas someone who doesn't may not. Animals having sentience may be used in criticisms of other ideas, like how to build factories in uncultivated land, or how to treat working animals like horses. Someone who doesn't think animals have sentience may not think these are issues — or may think they are issues, but for very different reasons, and so come up with different solutions.

To learn something, we need to consider the conflicts and address misconceptions we have. Not someone else's.

This comes up in standardised education systems. Textbooks, to be accessible to a reasonably wide audience of different people, are written for an average reader. But really, there is no 'average reader' — each person's interests and problems are unique. Thick textbooks are usually written with this in mind, arranged so that the student can jump around and focus on the parts relevant to them. It's rare that reading a textbook cover-to-cover is exactly what would help a specific individual learn best.

Even if we have a popular misconception, our solution won't be exactly the same as someone else's. It has different constraints, because we all have a unique set of background ideas. The things we're curious or confused about will be different.

Why you need to be in control of this

So, not any criticism will do. We need a specialised set of criticism to address the ideas we actually have.

The other problem with using non-specialised criticism is that there's practically an infinite amount of criticism available. There's more criticism in the world than you have seconds alive to hear it.

Criticism is very easy to come up with. Good criticism is harder. Good criticism that actually targets your problems and misconceptions is much harder. One needs a way of selecting what to pay attention to and narrow down.

The criticism you come up with yourself has the best bet of being relevant and broadly compatible with your background ideas.

Writers often make a few passes of a draft themselves before showing it to other people for feedback. When we have a new idea, we sometimes want to flesh it out before we share it with other people. If we share too early, criticism may cause us to take it in a different direction from what we wanted — or it might die entirely — when we could have thought about it more and explained it in a different way. It's no longer addressing our own problem, it's addressing the problem of the intervener.

After yourself, the next best is getting criticism from sources you choose. Like asking people questions, reading about the topic, comparing your solution to other people's, or submitting the specific idea/work to individuals you think will give useful feedback. Especially when you can direct this — such as asking specific questions, or reading only the parts of the book that feel most relevant.

But we are good at fooling ourselves, and sometimes we don't even know we have a misconception. It's easier to point out other people's mistakes than our own, because our own are often hidden in our blind spot. How can we correct errors in our blind spot if we are the ones directing the critical process?

Directing doesn't mean you know exactly what you're going to get beforehand. Sometimes, you want to get more general feedback, or feedback on the process you're using. Or you may read about fundamental ideas to understand one thing better, and find they reveal other misconceptions. But all of this will come out of some problem, some interest, some question that you have.

Seeking general feedback is only useful in rare situations. Usually, the more specific your question is, the more useful the feedback will be. For instance, if you posted a work of art to a general 'arts' forum, you might not get useful criticism compared to posting it to a forum dedicated to the type of art you do. Unless, of course, you had the specific problem of wanting to know what people outside your subfield think!

So to be relevant to your problems at any given time, only certain kinds of criticism are helpful — namely, criticism that is wanted. The rest is useless or actively harmful.

Why harmful, though?

The critical process needs to happen inside one's own mind to learn something at all. There needs to be some kind of process to decide which things are relevant, which things help most with the current problem you are on, and what to try next.

And it needs to address your problems and ideas — not abstract 'problems', or someone else's ideas. Otherwise, it's not really you learning it; it's not being incorporated into your mind. You could memorise some facts (or how to parrot them), but to understand them requires that they pass through this critical process.

When this process gets outsourced — such as being forced to learn things in school, or trying to learn something because you 'should' (according to someone else, who may be imaginary) — you stop judging whether a given criticism solves your problem. You stop assessing it according to your own criteria.

Criticism that you would naturally apply to the thing you're being compelled to learn thus gets suppressed. (Including meta criticism, like "This is boring", "I want to do something else", etc. Those criticisms indicate that something else might be better to learn right now.)

This suppression is harmful in two ways:

  1. It is anti-rational and anti-truth seeking.

You have criticism about either whether/when/how to learn this thing, or criticism of the thing itself, or both — which are valuable data — that you now must ignore.

Suppressing criticism of whether to learn something is siding with 'learn this thing' dogmatically, instead of rationally resolving the conflict.

Suppressing criticism of the thing itself a) assumes it is true instead of finding out whether it is true, and b) sabotages your attempts to learn it. Criticism is how we learn. The more we suppress our own criticism, the more fragile our understanding will be.

Suppose you were wondering whether light is a particle or a wave, and your high school taught that it has "particle-wave duality". You might wonder what that means, how it could be both, and what physical reality would have to be like if it were only one or the other. Then you find that some physicists say it's only a particle, but for that to make sense in physical reality it requires the existence of multiple universes — which on the face of it seems nuts to you, but also intriguing. You start wondering what reality would be like if there really were these infinite copies of you in other universes. Then you start thinking about whether that would solve the time-travel paradox of killing your grandfather, and...

— Except you don't. Because as soon as you ask your teacher about the idea that light is only a particle, he scoffs and tells you that you don't need to worry about that until university, and in this class all you need to know is that it has duality. You would have asked more, but your teacher seemed irritated by the question, and anyway it wasn't relevant to the assignment at hand. So you continue with the assignment, trying to build your model of reality on this thing that you have doubts about, wondering what kind of world would permit 'duality' (or not wondering, because being forced to drop your last question was unpleasant and at this point you just want to get the work over with). You never get to time travel paradoxes.

After the exam is over, you spend the next couple years forgetting most of your high school physics — which is fine, because it never came up in life again after that. Until you see a documentary, which piqued your curiosity and prompted you to re-learn some physics. This time, your model of reality is created under the direction of your own interests, and you get to use the full force of your critical faculties to understand the world.

  1. Unwanted criticism is coercive and creates hangups: lasting psychological blocks/irrationalities.

It is difficult to keep two competing versions of reality in your mind (theirs and yours), especially when you disagree with or have doubts about one of them. And when there is force or pressure to adopt a version of reality that is not your own, it's often painful, or threatens pain/punishments if you don't succumb. Boredom — especially boredom at having your attention forced onto something demanding — is famously painful, and this is why.

To suppress your own critical faculties and follow someone else's agenda, you have to find ways of directing yourself to not think about something you would naturally think about.

You have to cordon off areas of your mind, kinds of criticism, and make them off-limits. When your own critical attempts are thwarted and you find no rational way out, you must resort to coping mechanisms: you create no-go areas where you can't think, everything is foggy, your mind goes blank — and if you try, it hurts. You use your own creativity against itself.

This is how hang-ups are born. These are active irrationalities.

If the harm done from unwanted criticism were only in the moment — wasted time, wasted effort, confusions that require detangling — that might be a shame, but no big deal. Passive errors in our thinking can be corrected later. We're constantly making mistakes and correcting them; that's life.

But when faced by coercion — when we must self-suppress to get through an unpleasant situation — we wind up with errors that actively resist when we try to correct them.

Is this harm avoidable?

Someone may object, "But you can just not listen to criticism. Criticism can only hurt you if you let it."

To the extent you can do this, sure. But that's not always trivial.

  1. If you're in a situation where there's some external (or internal) pressure, you need to creatively come up with a way of overcoming that pressure.

If it's pressure from school, you may need to solve a whole host of problems around interacting with your parents. If it's social pressure, you might need to find a way to be accepted to the group without addressing their criticism. If it's pressure from yourself, you might need to deal with big unsolved personal problems like how to feel 'good enough'. Simply ignoring criticism doesn't always work.

  1. Once you hear a criticism, that destroys your ignorance.

You can't always pretend you haven't heard it, because new knowledge affects the path of your thinking. If someone tells you the twist of a movie, that spoiler means you can no longer try to make guesses about the twist yourself. It is exactly the same with all thinking. There are times when you want hints to help you solve a puzzle, and there are times when hints would ruin your problem-solving. This doesn't only happen when there's a fixed answer to discover, like in a movie or a puzzle; the same process happens in creative thinking, where the 'end' hasn't been invented yet.

  1. Conjectures that are created in the expectation of different kinds of criticism are different.

When you're in private speaking to a close friend, you say different kinds of thing from when you're speaking to a public audience. Which friend makes a difference, too — it's not just 'few vs many'. And it's not just that you say different things. You think different things. You make different conjectures around different people and sets of people. (This is part of the value of talking to people in the first place: you create different ideas than you would alone.)

These three dangers are why it's important to not share a draft before it's ready. Criticism at the wrong time can play the same role as movie or puzzle spoilers. It can kill off ideas early, or change the direction of your thought, even when you would have preferred to keep thinking about it. It does this no matter how well-meaning the person giving criticism was, no matter how non-pressured you feel, no matter how much responsibility you take for how to reply to the criticism — criticism changes the course of thinking. You cannot recover that ignorance.

Privacy is criticism scheduling

"When a person is learning successfully, the trail of wilful ignorance, as it were, that they leave behind them – the sequence of rejected opportunities to acquire information – characterises their creative process just as faithfully as the sequence of problems that they faced and solved."

— David Deutsch, In Praise of Ignorance – Taking Children Seriously journal issue 31

Choosing when to get movie spoilers, puzzle hints and creative feedback are all types of criticism scheduling: directing the critical process, so that you're using the best ideas you have on which criticism would be good to introduce next.

When most people think "violating privacy is bad", they think of things like:

  1. People not reading their private diary, or their browser history, or letters to friends. Not anything those people might do or say, but the act of reading those things itself being violating.

  2. Entering a bedroom unannounced, or letting oneself in to a house of an acquaintance. Again, this is considered separate from what they might do or say while there.

The sense of violation is in fact intimately connected to what they do or say to you — even if they're not directly commenting on what they've seen. It affects your expectations of them and what kinds of things they could conceivably do or say to you.

It also affects your internal model of their mind, which affects the criticism environment of your ideas.

So, there are two broad reasons violating privacy is bad:

First, it creates uncertainty about what kind of information (including criticism) is going to be brought up to you that you then have to deal with. In other words, it can directly interfere with criticism scheduling.

Second, it changes your model of them. Your mental models can shape the criticism you think about in the first place. If you can't do something without thinking how they would think about it, that forces you to confront or deal with certain criticisms.

Certain mental models are easier to engage with than others. Modelling someone hostile takes criticism in a very different direction compared to modelling someone cooperative. But even modelling someone friendly can change the way you think in ways that you may not necessarily want, such as if they know things about you that you're not ready for them to know.

If you're vulnerable to your parents reading your diary and discovering your secret thoughts, then even if they say nothing — even if you think they just may have seen it — your model of them becomes more complicated. You have to interact with them on the basis that they may or may not know those things.

You'd have to think something like, "Either they did or did not read that page, and as a result may think such-and-such or not, and that might mean they do such-and-such or not." — It causes a whole load of things that complicate your relationship.

Whereas, the way should be is that you know — you are certain — that your parents do not look at those things. They don't come in to your room because there is a lock on the door, or because there's a strong family institution that they do not do this, and if they happen to be in your room and see your open diary then they still won't read it.

And even if they were amenable to rational argument, they may still have misunderstandings based on the private information they discovered about you, because it's never possible to share the information 100% accurately. Some things take a lot of background knowledge to understand.

A piece of private information might make sense in context, but another person may not have that context — especially if it's a personal thing that comes out of the nitty-gritty details of your own life.

Sharing non-personal ideas can be fraught too, for the same reason. You're faced with the problem: try to explain the context fully (impossible), dumb down the idea to make it comprehensible to the other person, or keep it private. Simplifying the idea for another person to digest risks you losing sight of your idea's depth and potential. If you understand an idea well already, this may not be such a danger, but having this early on in an idea's development can be devastating.

Controlling the privacy of your mind means you can control what gets exposed to criticism and when. This is one of the most important uses of privacy. Without this control, our thinking would be impaired or disabled.

We see invasion of privacy sometimes used intentionally to create obedience and disable criticism. In Soviet Russia, people would spy on each other — if you were suspected of having disloyal thoughts, you risk being thrown in the Gulag. In some religions, God is omniscient and can read your mind, including every sinful thought, which you must repent for. In coercive parenting, children are snooped on, and then are given punishments if there's something the parent doesn't like.

But this kind of disabling can happen when everyone is well-meaning. You might reveal something, and your well-meaning friend gives a comment intended to help but actually brings up a problem you weren't conscious of, which you now can't ignore, and it throws you off the fruitful thinking you were doing before it came up. This can happen even when no one is being irrational or sensitive. If those things are at play, it's worse.

When to have privacy

"One cannot be an individual—a person separate from family and society—without having secrets."

— Thomas Szasz

What feels fun and enlivening to share, and what makes you inwardly cringe?

If it makes you inwardly cringe, that's a sign sharing it would interfering with your thinking.

It's not possible to understand our mind fully and have an explicit idea of all the ways that sharing something private can cause problems. A lot of the things I've been talking about happen at the inexplicit or subconscious level.

Discomforts and fears are part of the intuitions you can use to form judgements about what to share and when.


Harry glanced away uncomfortably, then, with an effort, forced himself to look back at Draco. "Why are you telling me that? It seems sort of... private..."

Draco gave Harry a serious look. "One of my tutors once said that people form close friendships by knowing private things about each other, and the reason most people don't make close friends is because they're too embarrassed to share anything really important about themselves." Draco turned his palms out invitingly. "Your turn?"

— Eliezer Yudkowsky, Harry Potter and the Methods of Rationality, Chapter 7

We live in a culture that broadly undervalues privacy. Even aside from modern concerns like your employer finding embarrassing photos on your Instagram feed, or people posting their entire lives to Facebook, there is pressure to share private details in order to bond with others.

And Eliezer Yudkowsky's Draco does have a point. People often keep things private out of fear about what people would think of them. For some people, stopping worrying about what people think of them and revealing more would allow them to have closer and more meaningful relationships, and make their lives better.

But revealing things before you've resolved that fear — before you know whether that fear has useful information — can be dangerous. It could be that the fear has a valid concern, such as not wanting to be defined. As long as you have that fear, that's an indication there's something you're not okay with. Bypassing that fear through force of will is raw material for a hangup.

So when should you reveal things?

As a guideline: When doing so is necessary for your joint problems (or projects or interests).

People-pleasers and social folk tend to offer up far more than makes sense for their shared problems.

Loner types or asocial folk may do the opposite. People who block out social information also try to limit their own information from going out. (Explained more depth in my Beware Social Coping Strategies post.) They may maintain so much privacy that it sabotages them. How to deal with too much privacy is beyond the scope of this article; but keep in mind that even there, the solution is not to force yourself, but instead to make updates on what might be nice for your life.

Because society so far has a lot of coercive control aspects (such as religion and coercive education), we live in a culture that tends to undervalue privacy. If people had more privacy, individuals would have much more opportunity to disobey authorities and avoid peer pressure. It would be a more individualistic and less tribalistic society.

Privacy means you can think freer. You have more slack.

Whether it's details about your personal life, or what matters to you, or new ideas you're developing, or a budding project you're working on, privacy is important for managing the flow of information that affects your thinking.

You can try ideas that other people can't even follow without a lot of context, let alone agree with. You have the option to make no thought unthinkable. You can go anywhere.

When you're not beholden to make yourself agreeable or comprehensible to someone else, you can freely explore the unknown, the unthought.

This unknown is the only place that progress happens. It's where discovery lies.

New to LessWrong?

New Comment
8 comments, sorted by Click to highlight new comments since: Today at 7:35 AM

I try to avoid criticism and use observation whenever possible. For example, let's say you're making a video game. You could post it to a forum, but that will give you a self-selected pool of critics (the most vocal ones) who will also mix in their own suggestions (like "the player should be able to talk to the monsters"). A better way is to hire random playtesters, watch them play, and draw your own conclusions.

People we like are often those we feel free to share, as Herbert Simon put it, "half baked, quarter baked, and no baked ideas." Intelligent communities often wind up with criticism based cultures as a means of showing off cleverness, which is death for generative processes. I think of this as search and prioritization having different ideal conditions in which to operate. Search/generation/ideation doesn't work well in the presence of criticism, which is part of the prioritization process. New things are also often stupid without a bunch of other things being different also. But sometimes changing all those other things is more tractable than it first appears.

I've never thought about it this way before, but crab buckets are a stable equilibrium in punishing search spaces. i.e. the young people trying new stuff are going to get their fool selves killed.

Bravo! This essay is very well put together, and it make my mind go "bling" a couple times.

I have experienced guilt for not taking well to criticism, and I feel this piece helps to explain why: the criticism didn't address my own unsatisfaction with the work, nor highlight what I thought was an important shortcoming. Looking forward, it required things of me without actually helping me making something better. But as you mentioned, feedback (just an alias for criticism) is almost sacred is certain circles nowadays.

This was fantastic. Even short as it is, it's the most clear examination of criticism I've seen, and thorough enough an explanation to lead to truly deep understanding of how it relates to creativity and learning of all kind. Thanks for writing this!

Great article. This covers more of morality than most lesswrongers will notice, I guess. If:

Criticism = an explanation of a problem with an idea

Idea = a recipe (https://arxiv.org/pdf/1407.0681.pdf) that programs a vehicle to execute it

Problem = a conflict between two ideas

Explanation (of a problem with an idea) = a virtual reality environment in which the idea dies when executed by the environment (i.e. executed recursively by a vehicle that bottoms out in the environment.)

, then this idea that it's wrong to criticise people unless they ask for it can explain:

  • why it's wrong to kill an animal - e.g. by putting an animal in a slaughtering machine, which is basically a virtual reality environment that contains an explanation of problems with the animal's survival strategies.
  • why it's wrong to hurt people by physically injuring them - if someone executes a program with their body (e.g. executing a decision to punch someone with their arm), then their body is a virtual reality environment that someone near it is in. It contains the explanation of problems with the ideas for survival that the person near it has. Ideas (programs/recipes) that the body executes conflict with the ideas the person near it who gets injured has.

>So to be relevant to your problems at any given time, only certain kinds of criticism are helpful — namely, criticism that is wanted. The rest is useless or actively harmful.

What about a technologically poor, isolated tribe of people who migrated away from a bigger tribe early on in life - and none of them have died yet so they haven't discovered aging - receiving unsolicited criticism from members of an advanced future country that they are going to die if they don't use specific anti-aging technologies?

That is wanted in some sense, but it does seem to be an invasion of privacy and disruption of their schedule of criticism.

Maybe this only applies to people who have immortality - safety from risks from unintelligent sources. (Because risks of harm or death from other intelligences would be covered by this - prohibited by the rule against invading another intelligence's privacy)

>And it needs to address your problems and ideas — not abstract 'problems', or someone else's ideas.

Interesting - what are the abstract 'problems' you're talking about (do they include problems with our understandings of abstractions like problems in math)?

Actually I have a correction of the below:

>this idea that it's wrong to criticise people unless they ask for it can explain:

  • why it's wrong to kill an animal - e.g. by putting an animal in a slaughtering machine, which is basically a virtual reality environment that contains an explanation of problems with the animal's survival strategies.
  • why it's wrong to hurt people by physically injuring them - if someone executes a program with their body (e.g. executing a decision to punch someone with their arm), then their body is a virtual reality environment that someone near it is in. It contains the explanation of problems with the ideas for survival that the person near it has. Ideas (programs/recipes) that the body executes conflict with the ideas the person near it who gets injured has.

My correction is that the environments that cause killing and violence are not virtual reality environments, so they are not explanations and therefore not criticism, if my definition of explanations as kinds of virtual reality environments is used (one could alternatively call those VR environments "models of the world" and only call specific kinds of natural language expressions explanations).

Virtual reality environments are usually simplifications of non-virtual environments. Virtual reality environments (i.e. simulations of environments) have different hardware than the environments they simulate.

I'm not sure where I heard this first, but I think about ideas like a good bottle of wine: once opened, it needs a little breathing time.

I've been enjoying this sequence on privacy, in no small part because I disagree with some of its fundamental premises so strongly. (Hopefully, someday soon I'll make the time to pull together a sequence laying out the grounds of my disagreement.)

But without backing up that teaser (sorry), I'll say that the brief mention of privacy as a guard against falling into inferential-distance chasms seems very straightforwardly true in hindsight (even to one skeptical of the guard-against-coercive-control angle), though I hadn't been able to put it nearly so cleanly in my own thoughts. If you've got even a short post's worth of content on related ideas in deploying privacy in deeply cooperative/aligned contexts, I'd be especially interested to hear more thoughts in that direction.