The Way to Break Plateaus is Not What You've Been Told
The way to break out of a plateau for whatever skill you’re trying to get better at isn’t to keep on pushing for incremental improvements given your current understanding of the fundamentals, but to break down your technique, start sucking again, and build your technique from the ground up. Not intuitive right? We’re trained to think that putting more hours into something is always the way to succeed. It’s in our culture. “Hard work pays off!”. Smart work pays off. “Quit complaining and do the work!”. Fuck you.
The Valley of Suckage
This graphic illustrates the nature of plateaus and how to break out of them. When you start learning a skill, there’s a period of time in which you put in hard work and see gains consistently. The behavior of the gains you make depends wildly on the skill - for writing you may see linear gains, for lifting weights you may see initial exponential “newbie” gains, and for playing the guitar gains may come at a slow pace after which there’s an inflection point and you get better quickly. You continue your beginner lessons, keep experiencing gains, see steady improvement. You’re happy.
And then the gains stop coming. You’ve plateaued. If you’re playing guitar, you try hammering away at metronome exercises and you’re lucky to see the metronome speed go up. If you’re learning a language, you try to memorize more words and make sure the grammar you’ve already learned is well internalized, but you still feel at a loss for words when you’re in conversation. Now you’re sad. Some people decide to quit at a plateau, thinking that they’ve got no “natural talent” in what they’re studying, or make up some other excuse. Some people let their ego get in the way and convince themselves that they’re actually really good, fearing the possibility that they have to get worse before getting better.
The right thing to do is to get worse paradoxically. The reason you’ve plateaued is because you’ve maxed out the mileage on the current technique you’ve got. There are no more gains to be made and it’s time to return to the fundamentals. Enter the Valley of Suckage.
For a period of time, you’ll be forced to break down your skill at a finer level of granularity and focus on minutiae. Your legible skill level, the skill level people assume you’re at if they watched you practice, goes down considerably even though you’re taking steps to improve your skill level in the long term. In the example of playing the guitar, this may mean relearning how you hold your pick because that’s a weak link in your ability to play fast which you’ve never thought about. At times it’ll feel that the endeavor is pointless and you should go back to performing how you were before. It’s a very real danger to get dispirited enough and cave into going back to your previous local maximum.
But as you continue to focus on granularity, you reinforce the correct habits in performing the fundamentals. Continuing with the guitar example, your brain gets used to holding the pick correctly and it doesn’t feel strange anymore. You start seeing results. Yea they haven’t brought you back to where you used to be, but you have faith in this path and can see yourself, at the very least, getting back to your old skill level. And then you get back to your old skill level. And then you pass through it. Congratulations, you’ve broken through your plateau and you see that the hard work involved in breaking your technique down and rebuilding it from the ground up has paid off. It’s a simultaneously humbling but inspiring event.
There’s an interesting parallel between this valley of suckage and the Dunning-Kruger effect. This effect, put shortly, is a cognitive bias where people with low ability at a skill overestimate their ability. Often the people exhibiting this bias are the ones who are riding the high of their noob gains and just about to hit their first plateau. With this in mind, if you find yourself getting a bit cocky about your abilities, you could take that as a sign that you’re about to hit a plateau and will need to be humbled if you want to get any better.
So what happens after you pass through the valley? Do you declare victory? Do you call yourself the grandmaster of whatever you’re trying to learn? Those are options, but they’re bad options. Fortunately (or unfortunately, depending on your perspective), you’ll hit more plateaus and need to go through more periods of rebuilding. Enter infinite suckage.
Infinite Suckage
The reality is that becoming a master at something takes a lifetime. It’s very easy to be “good enough” at a skill so that you can impress others or tell yourself that you’re talented, but to get to a point where that skill becomes an extension of who you are and provides a deep source of satisfaction is no easy feat. A sad truth is that our culture rewards settling at local maxima in this way. The external reward diminishes as one goes deeper into a skill on average.
You also will plateau multiple times in your life at a skill. The improvements weren’t guaranteed to be linear before your first plateau, and they won’t be after breaking through your first plateau. Take a look at the graph below.
There are multiple plateaus, multiple valleys, multiple perspectives etc. all along the axes. Each time you plateau, you’re going to be forced to return back to the basics of your skill, the ground principles, and take a magnifying glass to them to see what the weak links are. The resolution at which you look at a skill is the granularity, and the more granularity at which you work on the underlying techniques, the higher the local maximum you can achieve. The flipside is that the more granularity at which you look at a skill, the longer it will take to plateau and make your gains. In the case of guitar, it’s very easy to take a shortcut and pick with your wrist, and you’ll get to playing music faster than if you’d gone through the correct mechanics of picking from your forearm, but you also max out the mileage on that technique sooner. That’s why the graph above shows more time spent in the valley at each visit.
The closeness you are to asymptotic mastery is directly dependent on the level of granularity at which you’re looking at your technique. Since there are infinite levels of granularity at which you can look at technique, the time to mastery of a skill is infinite. It becomes important to decide what your goals are for this reason, because you may only want a certain level of granularity or want just enough skill level to achieve a certain goal. If you do want absolute mastery, you’ll want to set milestones along the way which you can reach. And after reaching them, you can break down the technique again, increase the granularity, and build up to get to a higher max.
Life in the Infinite Rat Race
If mastery is an infinite process, what do you do? Do you give up? Do you send me rude emails saying I’m an idiot and that I’ve gotten this all wrong? Both are valid options, but not as good as embracing mediocrity.
The truth is that you can’t achieve perfect mastery of a skill. You can be better than you were before, but you will not reach the asymptotic limit. Even people who appear to be masters at something will tell you that their skills hit a limit, these limits often requiring a superhuman level of granularity and attention to detail. In some cases, it’ll be impossible to even get to the far right end of the distribution in terms of mastery - it may be the case that you’ll be “good” or even “alright” at whatever you’re trying to learn for the rest of your life. And this is despite attempts at going down the ladder of granularity and rebuilding your skills.
The solution to this problem is to carefully examine the reasons for which you’re learning a skill. If the purpose is for some external validation, trying to get to the far right end of the distribution, then you’ve set yourself up for anguish. You’ll repeatedly hit plateaus, repeatedly break down your technique with a higher amount of attention to detail, and repeatedly think to yourself “when am I going to be done with this?”. None of it will be enjoyable. When you finally do achieve the goal that you set out for, you may find that it doesn’t provide you the satisfaction that you’d been searching for.
Learn the skill because you find it inherently meaningful. And if that means you’re happy at where you plateau because your current skill level is bringing you joy, there’s no point in forcing yourself to break everything down unless you feel that the joy of improving outweighs what you’ve currently got.
If You Want to Go Up, Try Taking a Right Turn
At times, getting better at your skill can involve not thinking about your skill at all and trying to learn something else. This counter-intuitive reality is due to the network effects of learning and building skills, where one is able to make bizarre connections, equivalences, and metaphors between different skills to learn better. In the same vein, taking things you already know and manipulating them in interesting ways can help you create new frameworks for thinking about a skill you’re trying to learn.
One can look at humans as possessing a set of attributes and underlying values to these attributes. These attributes are predetermined values that dictate an individual’s ability to excel at certain skills. For example, taking memory as an attribute, a person with a higher memory would have more predisposition to excelling in language, certain fields of science that are more memory dependent (biology?), and history. Using this model, it becomes clear that boosting the value of underlying attributes increases proficiency in skills, but it also works the other way around in that improving at certain skills boosts the underlying attributes. In a way, learning a skill is a form of attribute training, and an attribute improves in accordance to how much variation there is in training regiments that employ it. What this suggests is that skills are inherently “coupled” and can grow together.
What does this mean for you? It means that it’s possible that you benefit more from learning something that stresses similar underlying parameters as what you’re trying to learn. Some examples include:
Mathematics → computer programming: improves ability to chain together logic
Running → cycling: another aerobic exercise but stresses different leg muscles
Writing fiction → writing non-fiction: it’s still writing, but writing non-fiction may put more of an emphasis on clarity, which could have positive carry over effects when going back to writing fiction
Playing guitar → improv comedy: improv stresses your ability to think of things on the spot. Also stresses your ability to perform in front of crowds. The benefit to being able to play guitar in front of crowds is obvious, but perhaps there’s a link between the ability to think of jokes on the spot and your ability to improvise on the instrument?
You can get quite creative with this. The benefits of “transfer learning” in this manner are dependent on how you model the different skills in your mind and create links.
On a side note, many great scientific discoveries have been made by taking techniques and approaches from one field and applying them to another field.