The ability to observe is probably at least 2/3rds of what separates non-artists from amateur artists. But those 2/3rds are near-useless without the ability to move your pencil the way your eyes want to it to go. And once you've transitioned into an amateur artist, around 9,000 hours of honing your technical skill is what separates you from a professional.

"Technical Skill" is a broad term - kind of a catch all for all term for various motor skills you'll need to develop, background knowledge about how particular types of lines and shapes are perceived by most humans, and how to combine those skills and knowledge to produce particular effects with your drawing.

I can't even begin to cover all of it, and most of it isn't really appropriate for Less Wrong. But I will talk about some key motor skills that tie in with the next article, and a significant bias that plays a role in them.

This article was challenging to write - distilling a kinesthetic process into written words is difficult. This article will not be a substitute for having a teacher and a model, nor will it tell you exactly what exercises to do. But it will try to lay down some concepts that I'll further expound on later.

Holding the Pencil

For many of you this may seem basic, but at least one reader commented that they went for years without understanding this, and because it seemed basic, nobody ever noticed it and corrected them.

Holding a pencil should look approximately like this:

But its a bit more complicated that that. Many skilled artists hold the pencil in different ways. The biggest things to keep in mind are:

  1. Don't grip the pencil too tightly. You'll hurt yourself, and it won't help.
  2. If you hold the pencil closer to the tip, you will have more control over it, which is useful for fine details.
  3. If you hold the pencil towards the back of the pencil, you'll have greater range of motion, and allows you to quickly draw larger lines in a single stroke. It also will be looser, which can feel hard to control but can also produce certain line qualities you may want.

(I personally tend to hold my pencil similar to the image above, but closer to the middle of the pencil)

A few examples of a pencil grip in motion:

This man's grip is similar to mine, although the technique he describes isn't something I think you should be worrying about just yet. (I'll be talking about Darrel Tank's website later on - I think he has good tutorials on technical skills, but does not prioritize them based on their low-hanging-fruit-ness.)

This cartoonist switches grips a few times, demonstrating how they can be useful at different stages of drawing. This video is particularly interesting because his "loose" grip is actually closer to the front, which I haven't seen often.

Slow Drawing and the Sunk Cost Fallacy

I've spoken a few times about "slow, small, precise lines," and implied that they are a terrible idea. They often are. You'll be drawing slowly during some initial exercises that develop observational skills. But as soon as possible, you'll want to start developing a form of hand-eye coordination that involves moving quickly using long lines. Until you achieve that, the small, meticulous lines will probably have a choppy quality, and certain compositions will be harder to capture.

Much of the "energy" of your drawing1, and the quality of the composition, will be established within the first one to two minutes. This isn't a hard and fast rule, but holds true most of the time. Yes, you can erase, and rework things. With pure observation, infinite time and brute-force-reworking, you can craft a drawing that perfectly captures reality. But every time you erase and fix a drawing, two things happen:

One is that the paper smudges, tears slightly and otherwise degrades. This might be okay if you're working on a computer tablet, but so long as you're practicing with a real pencil, it's an issue. After erasing 5-10 times, your drawing will have noticeably degraded. It's not game over, but you'll have to work harder to overcome it.2

The other, more important concern, is that the more you've drawn, the more you're attached to the existing sections of the drawing. Say you've drawn an arm bent awkwardly. You can erase it and fix it. But the arm doesn't exist in isolation. It connects to the shoulder, which connects to the torso and neck. Fix the arm, and you have fix the all those other things.

You probably won't want to fix them all, because it will feel too sad for you to have to erase large sections of your drawing. And even if you DO fix them all, the result won't be a fluid, graceful image that captures the motion and interconnected muscles of your subject - it'll be a hodge podge of Frankensteinian bodyparts, awkwardly sewn together.

(There's also anchoring involved: once a line exists, you'll have trouble evaluating new lines on their own merits, instead of how they compare to the existing ones.)

Several times over the past year, I've worked on a drawing of a person for 5-10 minutes. By the 1 minute mark, I know something's off about the drawing. By the 2 minute mark, I've started erasing and reworking things. I have a nagging sense that I've done this before, and that the next 8 minutes will involve lots of erasing, and a drawing that still isn't very good.

10 minutes, and lots of erasings later, I have a disappointing drawing.

The nagging sensation that this isn't going to work well... that's what the sunk cost fallacy feels like. You've put in a few minutes of work (sometimes much longer - you can go down this path for hours). Starting over would feel like defeat, like your previous work was a waste.

The correct action is to start over anyway. It's true when you've only been working for a minute and have just noticed the feeling. It's still true 10 minutes later. It's still true if you've spent 4 hours painstakingly erasing and doublechecking against reality, adding lots of nuanced shading. It's an oddly near-universal truism, not just in drawing but in many projects, that the thing you just spent 4 hours working on, which would take another 4 hours to finish, could be done in 10-30 minutes if you started over.3

There are reasons to spend 4 hours on a drawing. Those reasons will not be relevant to you in the near future. All the most important elements of a drawing should take you no more than a few minutes. After that, you're getting distracted by details, which might make the drawing more interesting, but won't actually fix the existing problems with it.

Most importantly, it won't help you learn to avoid those problems in the first place.

Fast, Confident Lines

So, you need to capture the most important elements of a drawing, quickly:

  • You need to capture how different body parts connect together, as a seamless whole.
  • You need to establish a good composition, so that the details you work in later aren't just reinforcing a bad design.
  • I haven't elaborated on it yet, but you're going to want an energetic, interesting drawing, which is simply hard to compose without working quickly.

To do all this, you need to develop a particular kind of hand-eye coordination, which is probably different than what you're used to. You need to be able to draw large sections of the body, using a single line. That line can change direction. But it needs to be done in one fluid motion.

This artist demonstrates what I mean by "confident lines." She blocks out large chunks of body with long curves, without second guessing. She doesn't bother drawing the arms or feet, but she does end those lines AFTER they've curved in a new direction, so if/when she continues them she's set herself to continue them gracefully.

No Erasing

You need to work quickly, and fluidly. Stopping to erase will interrupt your flow. So you need to learn to draw without erasing. There are two ways of doing that:

1) Don't make mistakes ever.

This actually is not as unreasonable as it sounds. In the first 30 seconds, identify the most important lines of the drawing, and draw them. It's what the artist in the previous video did. Obviously, this is... essentially impossible for a beginner. You're going to make mistakes. But what you CAN do is draw boldly, confidently, let the mistakes happen, and then rather than trying to fix them, move on to the next drawing after 30 seconds. Over time you'll get better.

I haven't watched new students try to learn with JUST this philosophy, and I have no idea how long it'd take to develop from scratch. But if you DO have some previous drawing experience, I think this may be a good approach, at least to try out. If your goal is to produce something like the woman in the video, drawing 5 drawings in 30 seconds with simple, bold lines will probably produce at least one drawing that's better than the one you'd do in two and a half minutes.

2) Be okay with your drawing being messy.

This is what I actually recommend for beginners.

A big hurdle young artists have, when they're transitioning onto the path of a "professional", is they feel that "messy is bad." They're drawing like this:

[2017 Note: this used to be a link to a perfectly bad high school Dragonball Z fan-art, which was really useful for highlighting the sort of error modes I was pointing at, but a) it was sort of mean to use it, b) in 2017 the link is apparently dead, c) I was unable to find another example that illustrated the exact qualities I want to point to]

[Followup: Someone volunteered this old high school art, which doesn't quite hit on the same set of issues but works well enough to illustrate the basic concept]

When they should be drawing like this:

They're looking at the former, and seeing it as a fairly clean drawing that just needs to be fixed a little.

A college professor looks at Example A, cringes, and thinks "man this person is going to need to systematically broken down over the course of two semesters until they're ready to begin learning, and it's going to be painful for the both of us." They look at Example B and think "This person knows exactly what to do already, they just need to do it for another 10,000 hours."

The problem with Example A is that the artist is copying superficial elements of a particular style, without understanding the underlying principles that make good a good figure drawing. Example B has lots of overlapping lines, and vague messy shapes. But the figures there communicate a good understanding of anatomy, a grasp of weight, decent composition.

As an aspiring artist, don't ask if your drawing is better than A. Ask if it's at least as good as Example B. If you want to draw truly good Manga art, you must first learn things OTHER than the superficial characteristics of Manga. And while it may look like a mess at first, as you learn to draw that way, you'll understand that there's actually a lot of information there that Example A has missed.

(No offense to those of you out there currently drawing Example A. I've been there. It's a rite of passage. In particular, no offense to the blog I took Example A from. The blogger identifies it as one of their old, middle school works and gives other examples that show a lot of improvement. I tried to draw my own version of Example A, but it's actually really hard for me to draw that particular way now, and I can't find any older examples).

Begin Light, Emphasize with Darks

One important part of technical skill is being able to draw lines in the location you want them. Another important part is being able to adjust the lightness or darkness of those lines (as well as thick and thin-ness)

Your drawings are going to be messy. But you want a particular kind of mess. If you look at the right-most figure in Example B, you'll see that all the lines are the same thickness. This is okay - the artist has enough skill that they're all approximately correct, and the ones that are off have been repurposed - instead of being pure mess, they end up representing the volume of the figure.

The cluster of scribbles in the face suggest its roundness, and having a bunch of them devalues the importance of each individual line, so that even if none of them end *perfect*, your brain doesn't really care - it sees that they're all sort of fuzzy and accepts the average position in a sort of "Wisdom of Crowds" way.

It's okay that all the lines are the same thickness, because none of them are *completely* off. There's no giant leg that accidentally stuck out way too far and ruined the image. If it had, it'd be really hard to repair the drawing. Especially since you're trying to work quickly, without erasing.

You're going to be making significant mistakes, and you won't want to start over every single time.

The solution is to do your early work lightly, and then, once you've identified the parts you like, use dark lines to emphasize those areas. This tutorial demonstrates how to draw like this. Notice that within 30 seconds, he's established a framework, without worrying about making any "clean" shapes. Over the course of 2 minutes, he builds on that framework, filling in the mass of his subject matter, and eventually adding much darker lines to emphasize the final shape.

To do this, you need to be able to adjust the "value" of your lines (how light or dark they are). This takes some practice. A good exercise is to create a sequence of value-swatches like this:

Begin with the swatches on the far sides - make the darkest dark and lightest light you can possibly do. Then try and fill in the rest, gradually darkening.

Begin your drawing with something close to the second-lightest swatch. For now, try not to get much darker - it's easy to accidentally get too dark too quickly, and then are your lines are uniformly black and you can't emphasize the parts you want.

So... Recap:

These are only some of the skills you'll need to acquire, but they're the most important in the immediate future. So in summary:

  • Hold the pencil with three fingers, not too tightly.
  • Don't be afraid to start over.
  • Work quickly, without stopping to erase.
  • Draw strong, confident lines.
  • Be okay with your drawing being messy - let extra lines help define the form.
  • Start with light lines, make your mistakes, then emphasize the good parts with dark lines.

[0] There's something akin to anchoring bias here as well - once part of the drawing exists, even if you're trying to completely ignore it, it'll be warping your perception of what's actually going on.

[1] I promise I'll explain what I mean by "energy" soon.

[2] Each drawing tool has separate rules that need mastering. This includes pencils, charcoal, fountain pens... and computer tablets. I'll be specifically talking about the pencil here. Information here WILL still generalize to tablets, but I'll warn you that you'll experience some awkwardness transitioning to or from them.

[3] Why starting over saves time is a complex question. Part of it has to do with you already having studied the problem. Part of it is that a fresh canvas frees you from bias towards your old solutions. Part of it is that your existing work is suboptimal, and you'd need to spend extra time fixing it.

[Final 2017 note: We are now at the abrupt ending I warned you about. Sorry!]

New Comment
36 comments, sorted by Click to highlight new comments since:

No offense to those of you out there currently drawing Example A.

I don't actually understand what is wrong with Example A, and you don't seem to have explained it explicitly in the article. I look at Example A and think "wow, impressive lifelike drawing beyond my ability to reproduce". Example B looks to me like a different style or genre of art, rather than something which is directly comparable and strictly superior to Example A.

Upvoted for truth. I think I can articulate my guesses regarding what's wrong with Example A, but it would be good to receive an official explanation.

In fact, this could be a subject of an entirely separate article: "How to tell good drawings from bad drawings".

I think it was more or less conveyed in the text following. Example B shows a better grasp of anatomy, weight distribution, etc.


This was the intended point, but I'll edit it for clarity.

All art is subjective, and obviously Example A is more successful at a particular thing than Example B is. But if you're trying to improve, Example A just imitates the superficial characteristics of a particular style, whereas Example B is teaching you important things that'll apply to whatever figure drawing you do later (including things like Example A)

Example A just imitates the superficial characteristics of a particular style

This implies that there are deep characteristics of that style which example A is missing. What are they?

This is hard to explain, but I might be able to approximate it. Our brains tend to decompose objects we see into salient features: in Example A, consider muscles, clothing, and hair. Naive artists often try to reproduce those features without giving enough thought to the underlying volume or dynamics, which leads to errors in anatomy, proportion, and shading, and severely limits poses, representations of action, and interactions between objects in the field. That all adds up to a flat, disconnected look in the finished product: think Trogdor the Burninator.

In this case there's more going on, too: the artist isn't trying to draw from life but rather to draw in a certain style (pretty clearly that of Akira Toriyama). That adds another abstraction layer: the artist is drawing a representation of another artist's visual vocabulary, without any apparent understanding of how that style works to represent objects. Doing so tends to compound all the errors above, and also carries the stigma of being common among pubescent fanchildren.

Example B is a messier piece of work than Example A and was almost certainly drawn more quickly, but it directly attacks volume, pose, and weight distribution, and doesn't try to translate through an affected style. Although it subjectively might look less appealing, that grounding allows it to be foundational to work that's much better by almost any standards, and therefore I'd be much more impressed if I saw it in someone's sketchbook.

(Source: am not a professional-quality artist, but am good enough to occasionally fool people into thinking I'm talented.)

Thanks, that's useful.

I thought this was clear but a few people have been confused. I don't think I can explain it better than the followup paragraph already tried though:

"But the figures there communicate a good understanding of anatomy, a grasp of weight, decent composition. You can see from the third example, where the body turns and the shoulders overlap, that they're drawing what they see. You can tell that one foot is pointing towards the audience and the other is pointed to the right, even though both feet are a buzzy blob of lines."

Example A is floating in the air, not showing any realistic weight to how a body would fall, and each limb is sort of presented "flatly", instead of foreshortened.

Oh, okay. Your original wording just seemed to imply that there was some deeper skill involved with drawing pictures like example A, which was specific to that style and independent from the stuff in the paragraph that you just quoted.

Part of the confusion came from the fact that I interpret the style of example A as being one that doesn't even try to look realistic, so a paragraph explaining the greater realism of example B came off as unrelated.

This is too abstract by a few parsecs. (I'm tearing my hair out in frustration and raising my fist at the computer screen.) I think you should assign homework - detailed step-by-step exercises with a fixed number of repetitions, not "here's a concept, try to do that a lot".

All I can make out is: "Draw big, sloppy, messy lines roughly where you see your subject has lines. Start over. After 10000 hours of this, the lines will somehow magically start falling where they should be.". Is that anything like what you're saying?

I'm frustrated with this series as well. Here, try the exercises from Betty Edwards instead:

  1. Pick any good line drawing, turn it upside down and carefully redraw it line by line, viewing it as an abstract bunch of scribbles rather than a depiction of something real. Be careful not to turn your drawing right side up until you finish.

  2. Cup your hand and spend five minutes drawing the wrinkles in it, one by one, without looking at the paper. Just turn away from it, slowly follow every small wrinkle with your eye and let your pencil do whatever. Don't worry about the result, only the process matters.

  3. Make an outline drawing of something moderately simple, like a chair, while forcing yourself to look at the outlines of the space around the chair rather than the thing itself. Again, only the process matters here.

  4. Draw something simple, like a table or a door, taking extreme care to consciously measure all the relative sizes and angles that you're seeing, e.g. by holding your pencil in an outstretched hand. "Okay, the visible length of this line is about 5/6 of that other line."

Right now I'm getting nice results from #4. After doing several of these simple, slow, measured drawings, I find myself able to do a fast outline of any complex thing I see, automatically noting the proportions in my head without any effort. It was pretty surprising at first.

When I actually get to the exercises section, I'll be recommending you do pretty much this before you get to the gesture drawing. But I do think it's important to transition to the gesture drawing sooner rather than later.

Actually, the solution is suddenly pretty obvious - go back and add the observation exercises to the end of the observation article, rather than waiting for a single monolith exercise post. (That was actually my original intention but the article got really long. I'm not sure if it's better to stick them all in one article or to have a separate article that's mostly exercises).

Don't know about the other readers, but here's what I had hoped for from your series:

  1. New exercises, and amendments to Betty Edwards's exercises. It has to be very detailed and specific: explaining even one exercise with Betty Edwards level of detail, speaking only about the "how" without touching the "why", should take you a whole post or more. This is the important part. Half-ass it, and everything else is useless.

  2. Explanations why the exercises work. Might be interesting in a just-so-story kind of way, though people like me would just skip them.

  3. Tie-ins with general rationality skills. Useless filler that actually hurts the overall message. Throw it out.

Okay. I definitely think the detailed-observation-exercise-post needed to come before this one.

Something that concerns me is that I'm not sure I actually have anything to say about Betty Edwards' material that's any different from what's originally in the book. They're important exercises, and I think there are other exercises that she DOESN'T go into that need to be emphasized, but an article about them specifically would essentially be the same content, just rewritten in my own words. Which I think is legal, but still has me a little worried.

An eventual article I want to do would be a write-up of the actual workshops I've been doing, and what results have been occurring, which would give it a context similar to this article.


Most of the sections that do that are talking about something I think is genuinely important. Is it just the extra-verbage of "oh, by the way, insert less wrong buzz word" that you're concerned about, or do you feel that the accompanying paragraphs feel like filler?

Yeah, it's the extra verbiage, like the reference to map and territory at the start of "Observing Reality".

I actually liked the introduction to Observing Reality, but I did independently worry that the references to Sunk Cost, Anchoring and Wisdom of Crowds in this article were a bit ham-fisted.

I confess to a secret ulterior motive, which was to be able to link smart aspiring artists (but not yet aspiring rationalists) to this series and lay the seeds for greater interest in rational thinking. This wasn't my primary motive, just an additional goal I was trying to work in if possible. But I realize this is an incredibly dangerous secondary motive - if the result is a bunch of shoehorned in buzzwords that just detract from the result, that's bad.

I was fairly happy with how the earlier articles dealt with the issue, but if multiple people had the same reaction I'll recalibrate.

Only the wisdom of the crowds reference stuck out to me - I thought sunk cost was a perfectly reasonable thing to bring up, and anchoring wasn't just reasonable but actually useful, in that I hadn't really thought of drawing in that way before.

First off, it's not my intention for you to be ready to start practicing yet (given the information I've provided). I'm still in the process of bridging inferential gaps. There will be an article with exercises later. I added a disclaimer in the beginning to make this more clear. I also may be revisiting this article, because it WAS a challenge to write and I think it could be better.

If you want to get started now, I think the best choice is to purchase the "Drawing on the Right Side of the Brain" workbook (not the textbook, a separate workbook that includes exercises and explanations of why you're doing them).

DotRSotB emphasizes observation (slower exercises, designed to turn off the verbal center in your brain so you stop being pre-occupied with names, symbols and existing [wrong] knowledge). It won't actually tie in with THIS article, because this article is prepping you for gesture drawing, which requires you to already have an understanding of observation, but which uses different mechanical skills)


"Draw big, sloppy, messy lines roughly where you see your subject has lines. Start over. After 10000 hours of this, the lines will somehow magically start falling where they should be" isn't not TOO far off (if you have an instructor helping you, and/or more information that I haven't gotten to yet, it will take 4-8 hours, not 10,000." This was literally the response I got from the student who was most "trusting" (i.e. went along with the process, trusting me with his brain). At hour 6 he was like "I.... I dunno if this is working and I don't really understand it" and then at hour 8 he was like "oh wow holy crap look at that"

(which is not to say he was drawing perfectly then, but his messy lines were appearing in the right place, and he understood when they were in the wrong place and why)

In order for that to happen, you WILL need to understand things better than you do now. Yes, if I were to abruptly cut off now, you'd be right to feel frustrated.

Oh, okay then. Thanks. I was expecting that after one or two introductory posts you'd start interlacing theory and exercises.

A fair assumption. Sorry about the confusion.


That sounds reasonable, doesn't it?

Would it be possible to see some of your art? Do you have a website, perhaps?

I don't have a website for drawings. I do have a website for computer animation.

I probably should be posting some examples though.

As I get into some more complicated sections, it's becoming clear that the piece-meal way I'm posting this may not be the best for readers. I'm considering holding off on posting more until I've completed the tutorial section, so that I can edit it as a whole and better optimize the order than information is presented. If I go that route it'll probably be a month before new content goes up.

Edit: I would still wait a few days between sections, so people aren't bombarded with a lot of information at once.

Anyone have strong opinions on that, one way or another?

I like the piece-meal way because:

  • It gives me longer to digest each idea.
  • Since I tend to go back and skim earlier entries in sequences like this when a new one comes out, more frequent updates indirectly tend to improve my retention.
  • If I stop reading in the middle of a huge long article, I'm less likely to pick it back up again later. With short articles, I'm more likely to get it properly finished before I get interrupted or distracted by something else.

In this scenario, I would still be posting them as individual articles, I'd just wait until I had complete more of them and looked at how they worked together. (Probably post them a day apart or so)

I think this might be a time to follow your own advice. As much as I'd like to be able to read your articles now, you're probably better off waiting and organizing everything properly so you can present everything in a clear and useful manner the first time.


I am inclined toward the piece-meal delivery.

I like the content drip fed rather than coming in bursts. I also tend to skim over older sequence posts again as each new one comes in, cementing concepts better than attempting to digest big runs of posts over a few days and then going without for a long period.

Sorry to revive a seemingly dead thread, but I was curious as to whether you are planning on finishing this article series. As a beginner artist, I'm finding these tips refreshing, and I would love reading more about it.

How does Ctrl-z fit into starting over vs. erasing? On one hand, it can get you stuck on something for to long and fail to hit the real problem, on the other hand it has less of the specific drawbacks like disrupting flow and lines.

Ctrl-Z is definitely better, since you're catching the mistake as it happens, rather than letting it influence future lines. Working with a computer gives you a lot of tools to avoid this particular problem (for example, if you do a sketch and the lines get too thick, dim the opacity on that layer and and then use it as a vague reference for another). You can also easily hide multiple versions of the drawing and then see how they compare, afterwards.

The basic lesson still applies, which is to not get too attached to individual executions. But the computer makes it easier to apply the lesson without feeling bad about "losing" work.

[...] using a single a line.

[...] being able to just the lightness or darkness of those lines

Minor typo here: you don't want that "a" near the end of the first quote, looks like it wandered over from the second quote where it belongs.

Thanks, fixed.

About half of the images are no longer there

Sorry, the website I was hosting them on disappeared a while ago (it was hosted on a friend's server, who stopped maintaining it. I won't have time in the foreseeable future to find new suitable drawings.