Hacker News new | past | comments | ask | show | jobs | submit login
Not All Practice Makes Perfect (nautil.us)
153 points by DiabloD3 on April 23, 2016 | hide | past | favorite | 61 comments

Something I've noticed, which the article doesn't quite say (but sure hints at): you can accidentally teach yourself performance-degrading "techniques" if you keep practicing beyond a certain point.

I used to study (modern) drums 5+ hours a day. This would happen sometimes, around the time when I was learning a new technique: I'd start on the new material, practice it for a while, then I'd get it to a so-so point and I'd stall there making the same mistake over and over again. At that point I would have to stop and play something else and only come back to the new material the next day, otherwise it would take me weeks to get past the plateau.

The thing is, I think, when you practice, your body (or, I guess, your mind) doesn't know what you're learning- whether it's good or bad. Maybe for dancers or athletes there is some physical feedback, some pain from doing the wrong thing, but for many other types of performance there isn't and so it's very easy to just practice the wrong thing and learn it really, really well.

I think it is a common mistake by people starting to learn for example an instrument. You play the new piece, often a bit too fast, and make a mistake. Then just repeat in the same way, making exactly the same mistake over and over again. The hope is that if you play it enough times and put in enough effort you will cease making the mistake. That is of course wrong. Instead you learn the mistake really really well.

What you should do is to play it slow enough so you are sure not to make any mistakes. Sometimes that is ridiculously slow.

Something else that really helps is backchaining.

Split the thing you are learning into chunks... making each chunk as small/simple/easy as you can.

Learn the chunks in reverse order. Starting with chunk N, then N-1, all the way to chunk 1. This way, you always start with the hardest part and then move onto the easier (more practiced) parts.

This keeps the focus on the part you are trying to learn, and reduces learning stress.

With memorizing numbers as well, splitting them into chunks is also a good tactic, for example 392619582767 -> 392 619 582 767. It makes the whole thing a lot more tractable.

agreed, practicing slow enough to get it right is the most important thing for sure. once thats in the bag, then the problem becomes getting it up to speed, and techniques you were using at the slow pace can be non-viable past certain tempos. for me, getting it up to speed involves constantly interspersing the slow flawless practice with attempts that are right on the threshold of what i can do, tempo wise.

That seems surprising. It's not like there's a discrete threshold between "able" and "too fast". I'm not a violin player, but just about anything I can play slowly on a bass or piano, I can also play quickly, given enough practice, even if it means I can only initially play it at quarter or third speed with a metronome. Aside from some virtuouso performances, I think any beginner/intermediate player with a decent ear and headphones/sheet can play any piece at quarter speed. It seems like muscle memory is one of those things you really can achieve through mindless repitition with good technique, while theory requires a lot more focused practice.

edit: Okay, disregard that, the above comment is completely different and I guess I agree now.

An example of the parent comment's point: learning a piece slowly but with poor fingering or posture can lead to one's fingers getting tangled or becoming exhausted and stiff when trying to speed up.

I read somewhere that good practice for music specifically involves playing the piece through fast, ignoring your own errors, playing it through slowly enough that you don't make an error (and repeating this until you don't make an error), and then playing it at the intended tempo. So to someone this is a well-known idea. Is it not part of standard music instruction?

I think the point of it is just to push you out of ruts so that you can gain the motivation to get back on track in learning. A good coding analogy might be in biting the bullet and start coding when stuck in a paralysis analysis loop is often a necessity to move on.

Not a proficient musician by any stretch of metrics/imagination. But adding onto learning new pieces, what worked for me personally is hearing the parts in the original recording, and singing the melody with my voice if it's melody or vocalizing and tapping the rhythm with my whole body (foot and spine engaged) if it's a rhythm pattern.

That way, when I am playing my instrument when that problematic measure comes, I'll simultaneously sing out the melody or tap out the rhythm; and even if my technique is poor and can't play the fast series 16th notes, I'll hear the correct melody/rhythm/gestalt in my head and what I am off by and try again.

Congrats, you have independently discovered solfège, a necessary tool in every musician's arsenal. :-)

This happens in learning new languages as well , often speaking quickly the wrong way is counter productive. Speak very slowly at first and focus on being correct

A coach I knew used to say "you play how you practice" i.e. if you show up and just go through the motions at practice, you will play the same way.

The phrase I've heard is "Practice makes permanent."

Followed by "Perfect practice makes perfect."

It's the same thing for sports. Even with pain as feedback something happens during sleep that helps you better absorb new lessons. The other thing is that for an athlete learning a new movement pattern they need time to build supporting muscles and perhaps improve mobility.

I was going to make a similar comment but take it a different direction. With sports, just like with music or anything else, you need to understand the context and reasoning behind skills training. A lot of people -- for example, playground ballers -- are awesome basketball players, but they're awesome at 1-on-1 and if you throw them into a structured, officiated game they'll contribute far below their raw abilities. Football is probably even more extreme.

With anything, you need to practice & study with intent, and with almost everything, having a coach/teacher/mentor/guide makes this much, much easier.

> you can accidentally teach yourself performance-degrading "techniques" if you keep practicing beyond a certain point.

See also https://news.ycombinator.com/item?id=8832006

>you can accidentally teach yourself performance-degrading "techniques" if you keep practicing beyond a certain point.

But only if your practicing is no longer "deliberate practice" as defined by the author. Practice alone isn't the problem. Deliberate practice will always result in improvements. That's what I'm getting from the article.

I thought it was way too verbose going over too many examples to make a simple point. "Okay, I get it, what used to be exceptional is now just "meh".. Get to the point"..

It didn't address the actual matter, the one in the title, that not all practice makes perfect, until very far in the article.

I then thought about an article that would have given much more information with fewer words, a seminal paper entitled "The Role of Deliberate Practice in the Acquisition of Expert Performance" by Anders Ericsson.

I was then surprised taking a look at the article's author.

Thank you for reminding me of that paper's title. Jacobian mentioned it to me at PyCaribbean, but I've since been hunting for it.

It is available for free at https://graphics8.nytimes.com/images/blogs/freakonomics/pdf/...

Oh, you're very welcome. This paper sent me down the rabbit hole: after graduating as an incompetent Electronics Engineer, I thought about the essence of competence and I stumbled upon this article. I'm always happy when I find a resource that addresses an abstract idea I'm thinking about but can't formulate for two main reasons: it always opens a door -if it has a name, it can be searched effectively- and it always amazes me how come I didn't think of it before.

For instance, after graduation, I read a book titled "How to Study in College". Why didn't I ask such an obvious question before. I learned many things, one of which is the Cornell Note-taking method.

I've always been interested in how to search for, acquire, and keep knowledge, how to apply it, how to manage it, and how to track its evolution -mental breadcrumbs, which idea lead to the other?-

So it got me interested in how memory works: how we forget, how we remember, and what to do about it. It led to Hermann Ebbinghaus and "spaced repetition". I wasn't satisfied, though, for I wanted to know the underlying brain processes. The neuroscience behind it. This led me to Hebbian Theory and the concept of Long Term Potentiation. (The Brain from Top to Bottom - http://thebrain.mcgill.ca/) which frankly blew my mind to a thousand shards: memory viewed in terms of ions. (Here's a very nice video from Carleton College "Neuroscience - Long-Term Potentiation" : https://www.youtube.com/watch?v=vso9jgfpI_c) and here's a good presentation by Carleton's Dr. Matthew Hollahan: ("How We Can Improve Memory": https://www.youtube.com/watch?v=QXX58QhNfjc)

In the meantime, I put a name on the discipline: Cognition/Cognitive psychology, etc..

That was good, but I still wanted to think about it in a "meta" way. Some overarching theme or discipline that englobes other things I might be missing. After some time, though, I stumbled on "metacognition", and here's where it's truly amazing how the mind works: How didn't I make the connection before between "cognition" and "meta-cognition" even though "meta" was the very word I used to think about it. It's the very thinking about thinking and was humbled because I missed such an obvious link. I have already met that word before and read about it, but it didn't "click" until that very moment.

This has also driven me to think about "learning" in a larger context: learning organization and learning countries. Why are some countries more evolved than others, etc. (Which led me to a book titled "Why Nations Fail" (again, how on Earth did I miss such an obvious question). How can we make poor countries better, or asked from another perspective: in a post-apocalyptic world where all countries are the same, what would we do to rebuild civilization? How to bootstrap civilization? Can we take those steps and apply them on difficult countries. (Bill Gates was on Jimmy Fallon and the latter asked a marvelous, in its form, question: How do you solve a poor country?).

It got me thinking about the necessity of having some conditions A and B present to have X and those conditions being absent in poor countries, you can't have X. It then got me thinking about it backwards: Given C and D, can we get Y that's either a substitute for X or something at least better than nothing. First thing that popped in my mind was the Apollo 13 carbon dioxide filter hacking. After some time, I stumbled across an article titled "The Genius of the Tinkerer" [http://www.wsj.com/articles/SB100014240527487039893045755037...] and it appears there is a company called "Design that Matters" that does just that. It also was interesting to see the Apollo 13 feat mentioned. Steven Johnson has also some talks on Youtube and a book titled "Where Good Ideas Come From".

Anyway, the point is you never know where a piece of content can lead to and it can be frustrating to forget some of those references or the title of an article.

Here are some useful resources, too:

[Book]: The Complete Problem Solver - John R. Hayes. [Book]: Make it Stick, The Science of Successful Learning - Peter Brown et al. [Video]: Study Less, Study Smart - Marty Lobdell (https://www.youtube.com/watch?v=IlU-zDU6aQ0).

Sometimes when teaching, repetition of examples to drive the point is key so it makes sense that for a vulgarization article, the same author takes time to get to the point while for an academic paper, he gets directly to the point...

And this is why I was bored out of my mind in school as a kid.

Thanks for the name of the academic paper, I'll spend some time reading it.

That's a valid point. I didn't hold it against him, though: as a reader, I can skim through but someone who's not familiar with it could appreciate the verbosity.

This was an excerpt from his book PEAK. It's verbose because he has written this for the layman.

I wasn't aware of that book, thanks for the title! Sorry for the late reply.

I really liked the read but I kinda agree, he seemed to make the same or similar point paragraph after paragraph.

> The approach that he took, which we will call "purposeful practice," turned out to be incredibly successful for him. It isn't always so successful, as we shall see, but it is more effective than the usual just-enough method — and it is a step toward deliberate practice, which is our ultimate goal.

The article then concludes with a description of purposeful practice. Where is the promised description of deliberate practice?

I too was left disappointed and wondering the same thing when I came to the end of an otherwise great article.

My best guess is that anyone can do purposeful practicing simply by changing things up and staying out of one's "comfort zone", whereas deliberate practice is more or less the same thing, with the difference being that you have an experienced mentor guiding you through the process in a more structured manner with a tight feedback loop.

In other articles I've read about deliberate practice, the other elements include:

1. Maintaining a difficulty level that is just barely out of your comfort zone: not so hard that you flail wildly, but not so easy that it's trivial. This is touched upon in the article, but presumably "deliberate" practice also deliberately adjusts the difficulty level in response to performance.

2. Mindfulness while you're practicing. The description of "purposeful" practice in the article mentions seeking out errors and correcting them after the fact. The descriptions of "deliberate" practice I've heard mentioned paying careful attention to your performance while you're practicing, so you can instantly correct or rehearse trouble spots. (And this requires a corresponding attention to the difficulty level so that you can spare this attention to avoiding & fixing mistakes.)

The article keeps hinting at deliberate practice including,

> The approach that he took, which we will call “purposeful practice,” turned out to be incredibly successful for him. It isn’t always so successful, as we shall see, but it is more effective than the usual just-enough method—and it is a step toward deliberate practice, which is our ultimate goal.

but then cuts off before it actually describes deliberate practice... guess they want you to buy the book.

I've bought the book and would highly recommend it to others wanting a glimpse into what it takes to become an expert performer.

There are two main differences between deliberate practice and purposeful practice.

Deliberate practice requires a well established field where there is a clear gap between the top performers and those entering the field.

The second difference is having a teacher who can design practice activities to help you improve on very specific skills.

Taken from the book "Deliberate practice is purposeful practice that knows where it is going and how to get there."

I literally said the same thing to my wife after reading this article out loud to her.

"I guess I will have to buy the book!" Laughing as I assumed there was a book.

I will likely research this topic further whether there is a book or not.

Slightly related, I'm trying to learn how to read faster. My first goal is to become an average reader in terms of speed. My wife is an elementary teacher with a master's in remedial reading. I'm going to see if she can help me.

One issue that she noticed with my verbal recitation is that I stumble when reading sentences with conjunctions.

Now I have something to work on and fix which I hypothesize will increase my speed, and likely comprehension, and should reduce my chances of stopping and rereading sentences.

The article did not do a thorough job pointing to the new book: http://www.amazon.com/Peak-Secrets-New-Science-Expertise-ebo...

Also, I would recommend Moonwalking with Einstein which has fair bit of information on how to learn.

Really interesting. I wonder how you could use a similar technique to improve your code? Things like code reviews would give you the necessary feedback, but I can't think of a rigorous way to make sure you'd always be on the limit of what you're capable of, pushing a little further with each project.

Most of the challenge in writing code is in understanding the problem you're trying to solve. Not the act of entering statements into editors.

Yes it's important to know the capabilites of your language and deployment stack -- but that just requires study, which in my mind is different from practice. I've never found that coding well required the sort of intense dedication that learning to play an instrument well, or play a sport well, requires.

> Yes it's important to know the capabilites of your language and deployment stack -- but that just requires study, which in my mind is different from practice.

I agree. Which is maybe why Stackoverflow solves so many problems for a developer. You know what you want to do, you know how you want to do it. But you're stuck because you can't make sense of an API or don't know where to look for the functions you need.

I wondered the same thing. I was a classically trained musician and learned to implement some of these techniques in practice sessions. They're really effective.

But how could we use "deliberate practice" to improve coding? If we assume coding to stand for solving problems with code, then it seems we'd need to deliberately practice problem solving.

It's possible to measure problem solving skills, and push those skills outside of a comfort zone. But it's not as straightforward as measurement, "I'll play this passage without mistakes 3 times in a row", or "I'll place this golf ball on the fairway 5 times" and adjusting performance based on feedback.

It depends on what you're trying to learn with coding. I've recently had good success with this technique when applying it specifically to OpenGL. I've used a few specific parts of OpenGL extensively, and not used the rest of it very much at all. So I took some time to go through the Red Book examples very carefully and made sure I truly understood what they were doing. Then I wrote my own simple project using the techniques to make sure I understood them outside of the book. Then I started applying parts of them in my work at my job. Then I extended what we were doing in my work at my job by going a little further. And then I did it again going a little further still. It's been pretty effective.

You can certainly practice improving your mechanics, by which I mean your IDE/development environment whether it's keyboard shortcuts or workflow imrpovements.

You can also do katas whether it's practicing TDD or setting up a new instance of a project in whatever framework you're using. I also find katas useful for keeping familiar with syntax of languages that I don't use a lot, which is useful if you do a lot of language switching.

>You can certainly practice improving your mechanics, by which I mean your IDE/development environment whether it's keyboard shortcuts or workflow improvements.

Maybe practicing this allows you to get into the zone more easily since you won't be thinking about the mechanics. Getting into the zone more easily can lead to getting better and deeper understanding of the material.

A guitar player practices repetitive drills that aren't creative to help improve.

A good lead developer should help you grow, partially by selecting problems and partly by providing more/less structure. E.g. I might tell a junior developer to implement the design on the whiteboard and provide multiple rounds of code review, because just getting a complex implementation right is plenty challenging; for a more-senior colleague, I'm much more likely to throw him/her at a complex problem and I'll provide a few "you could have improved <X> by <Y>", possibly afterwards.

Force yourself to use patterns you dont like or dont suite the task at hand. Such as: today write everything using lambda callbacks, everything mvc, no classes allowed, no malloc allowed, etc.

That doesn't seem to me to be a directed, deliberate approach as the article implies is necessary. There's no way to know what patterns would be on the boundary of my skillset, so I might spend time using something that's unfamiliar but actually very basic which wouldn't challenge me, or something that's unfamiliar and incredibly difficult which would frustrate me because I couldn't make any progress. Isn't the reason that the system outlined in the article worked because the next level was always within reach but also something that provided a challenge? How do we get that in an arbitrary subject?

Essentially I wonder if the problem is that in some things we learn (eg coding) there isn't really a defined "ladder" that allows someone to practise harder and harder things in a deliberate way.

Yes, especially if you don't have a senior colleague where you work and there are no other developers who could review your code like in my case (sole Frontend-Dev amongst Java-EE-Devs).

Argh, I wish I could find a citation right away but having trouble in search - while I get the example (memorize X amount of numbers) and discovering there's an upper bounds is understandable. But what I had in mind was a musical instrument, in that there are a finite number of notes and how those notes can be played in a harmonious fashion. The study I saw, or think I did, showed that when performing a music solo some musicians were able to turn off the 'active' part of thinking about the notes to play. It was some different mental channel. That would, to me, explain how a guy like Freddie King could play a song like Hideaway for 45 minutes with his band and people would enjoy it.

Quite a cool article. Lots to consider. Phrase around my 'woodshed' was always "Practice makes better" though because perfect and guitars don't really ever align.

One of my trainers always said 'Practise does not make perfect - practise makes permanence. Only perfect practise makes perfect'.

I'd wager most of us who have had training in something had a trainer that said that. I can't imagine one that wouldn't; like "use the best tool for the job", it's one of those trite sayings that makes one sound wise.

A learning phenoma that I find fascinating is learning things that seem impossible. I experienced most in music, learning rhythmic independence. Most instruments require this skill, but where I really noticed it is trying to play and sing songs where the rhythms are disimilar.

Upon first attempts, there is simply no ability to perform the task at all. No purchase on which to begin efforts. But somehow the trying to do it eventually does work, and will manifest as a breakthrough.

This is in contrast to practice if of skills where a path of incremental improvements is followed. (As described in the article when learning to memorize strings of numbers).

What's strange is that the beginning attempts feel like you're not learning anything, because you actually aren't doing anything that seems similar to the action you are trying to learn.

For those who might have missed it, one of the authors (Ericsson) is the same person who originally came up with the "expertise requires 10,000 hours of practice" idea that Malcolm Gladwell later popularized. Unfortunately, Gladwell mangled it as badly as he mangles everything else. As I had to point out on another story here recently, Ericsson was talking about 10,000 hours of highly focused and disciplined practice, carefully analyzing and correcting errors etc. Thanks largely to Gladwell, many people have dumbed that down to 10,000 hours of any old thing, and it's no wonder that they find the idea highly questionable. The original makes much more sense.

Actually Gladwell presented the idea pretty well in "Outliers". However, it quickly became a watered-down meme. People latched on the 10,000 hours and forgot about the deliberate practice part.

Ericsson clears up this misconception in PEAK.

My father always said "Practice makes permanent" and I think he was right about that.

Just going after the title, "folk aphorisms" are such pseudoscience, it's not even funny.

Not that they don't apply, they do. But "many" of the times, not "most", let alone "all of the times".

And you have aphorisms that are exact opposites of each other, like "birds of a feather flock together" and "opposites attract".

The end result? people use them more to support their opinions or gut feelings rather than uttering them after a careful analysis of the situation.

In terms of rigor, they stand at the same level as anecdotal evidence, meaning something that adds practically zero information to an ongoing situation.

To sum up: "it's pretty easy to become average at something just by following instructions. Getting any better requires a different approach - well defined baby steps beyond what you're capable of at the moment." From me: Theory is easy, but altogether it still takes time and those baby steps on top don't make it easier to excel at something, so few people do. The reward flattens with level (big rewards early on, very little visible improvement later)

The stuff about Cortot is bizarre -- he belonged to a completely different era, when technical perfection was simply not the goal.

Wasn't there an article on motor skill recently, how practising variations of a motion (or two different motions) can be wildly more effective than simply focusing on one single movement? I can't seem to find a hn posting, but this one is along the lines of what I'm thinking of:

"The Fastest Way To Learn" http://www.jonahlehrer.com/blog/2016/2/15/the-fastest-way-to...

While looking, I also came across:

"Why Slow Movement Builds Coordination": http://www.bettermovement.org/blog/2010/why-practice-slow-mo...

"There are several excellent reasons to use slow and gentle movement as a means to develop coordination. Probably the most interesting reason (I'll start with that one) is based on an obscure principle called the Weber Fechner rule. The Weber Fechner rule describes the relationship between the magnitude of a particular stimulus and the brain's ability to sense differences in the amount of the stimulus. The basic rule is that as you increase the stimulus, the ability to tell a difference in the amount of the stimulus decreases. This is a very common sense idea. Imagine you are in a dark room with only one candle lit. It will be very easy to sense the difference when one additional candle is lit. But if you are in a room with two hundred candles, you will have no idea when an extra candle comes on.

This rule works for all varieties of sensory perception, including sensations of muscular effort. So, imagine you are holding a one pound potato in your hand while blindfolded. If a fly landed on the weight you would not know the difference, but if a little bird landed you would know. Now imagine holding a fifty pound potato. You wouldn't be able to feel the little bird landing. It would have to be an eagle. The point is that when you increase the weight from one pound to fifty pounds, you become about fifty times less sensitive to changes in the amount of muscular force you are using to lift the weight."

I'm not sure how much it is possible to generalize from pure mental practice like memorizing numbers to physical practice like playing the piano (like this article does, without much substance beyond hand-waving about "brains will be brains", so to speak).

But as mentioned in the comments here, and touched on in the article, it is probably true that it is important to be able to understand your own ability and progress (what you do right, what you get wrong), to continue to challenge one selves (making no mistakes makes for bad practice, but making the same mistakes over and over also makes for bad practice).

I think that for complex skills, one of the easiest ways to get direct feedback, is to work with someone who is a master already: spar one-on-one with a world champion, or play with a maestro, pair-program with an experienced veteran of several successful projects -- and you'll get much more from your sessions than if you practice alone, or with others that don't know what they're doing.

We often don't have the luxury of practising one-on-one with a master, and then the challenge becomes to a) still get correct feedback on what we're doing, and b) find out what we should be trying to achieve. The latter can sometimes be helped by looking at videos of people that preform at an excellent level, reading great code etc.

Yes, I am reminded of having read the same. Was it not the first link in your post? The skill subjects had to learn was based on a mouse-like control operated by squeezing.

Quite sure that the link I found was not the one I was thinking of. But I believe it probably was a link concerning the same paper.

Came across the link by accident, apparently I had posted it to my facebook feed, here's the HN link:

"Researchers have discovered a much faster way to learn new skills" https://news.ycombinator.com/item?id=11168515

practice makes permanent

Applications are open for YC Winter 2022

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact