The Plateau Is Not the End

i. Why progress feels irregular

Progress is something we infer, not something we watch. Nobody observes their own skill the way they observe a kettle. We notice it in snapshots: a piece that works, a compliment, a mistake that used to happen and suddenly doesn’t. Between the snapshots there is mostly fog.

The snapshots also mislead in a particular direction. Early on, nearly everything you learn is visible, because the distance between nothing and something is enormous. Later, the improvements get smaller, subtler and more specific — a cleaner rim, a steadier tempo, a function that still makes sense six weeks after you wrote it. They are real, but they are the kind of change only a trained eye notices, and the trained eye you are developing is busy raising its standards at the same time. Taste and hands improve together, and taste usually runs ahead. Hence one of learning’s stranger experiences: getting better while feeling worse.

And then there is ordinary life. Sleep, work, weather, mood and whatever else you are carrying all leak into practice. A week of bad sessions can simply be a week of bad sessions, not a referendum on your future.

ii. A learning curve is a sketch, not a law

It helps to have a picture, as long as we remember that it is a picture. The three lines below are invented. They are not drawn from data and they predict nobody’s path. Treat them like a map sketched on a napkin: useful for thinking, useless for navigation.

The first line is the one many of us secretly carry: steady effort, steady results. The second is closer to how people often describe their own learning — bursts and long stretches of apparent nothing. The third adds something we usually file under failure: a rest, taken on purpose, after which things move again.

The point is not that real learning follows any of these shapes. It is that if the straight line is your standard, ordinary unevenness will feel like evidence against you. And sometimes a flat line really is flat. The sketch cannot tell you which kind you are on. Only better questions can.

iii. What might be happening underneath

“Hidden progress” can sound like a bedtime story, so let’s be careful. I won’t pretend to know what anyone’s brain is doing during a plateau. But some kinds of improvement are plainly real and plainly hard to see.

One is consolidation: things you once did deliberately becoming things you can do while thinking about something else. A guitarist who no longer looks at her fretting hand during a chord change hasn’t gained an ability anyone would applaud. She has freed up attention, and attention is what the next skill will need.

Another is error correction, which often looks like going backwards. A self-taught programmer who starts writing tests may suddenly feel slower — more red messages, more failures — because problems that used to stay invisible are now being caught. The mistakes didn’t multiply. They became visible.

The third is feedback, or its absence. Practice without feedback is like focusing a telescope in the dark: you can work very hard and still drift. Meera’s thick-based pots may not be a talent problem at all. She may simply never have sliced one in half to see where the clay is gathering.

So some plateaus conceal real change, and some don’t. The honest position is that you usually can’t tell by feel — which is exactly why feel makes such a poor judge.

iv. Differently, not just harder

When stuck, the reflex is volume: more hours, more pots, more pages. Sometimes that’s right. Often it just rehearses the same habits with more conviction.

A writer who produces a page a day for a year will become fluent. Whether she becomes better depends on whether anything in the routine asks her to notice what isn’t working — rereading old drafts cold, rewriting one paragraph five ways, borrowing a sentence shape she’d never reach for naturally. A pianist who always starts a piece from the top gets very good at the first page. Starting at the hardest bar changes the practice without adding a minute.

Practising differently usually comes down to a few moves: narrow the target, get a clearer signal of what “right” looks like, change one variable at a time, or make the task harder in one specific way. It needn’t be clever. It needs to be different enough that the old habit can’t do the work on its own.

v. Diminishing returns, and rests that earn their keep

Most sessions have a point after which another hour mostly adds tiredness. You may know the feeling: the third hour of debugging spent reading the same line without really reading it, followed by a walk on which the answer turns up uninvited. I won’t claim to explain why that happens. The pattern is familiar enough to treat rest as part of the method rather than a confession.

A useful rest has a shape. It is chosen, not collapsed into. It has an end. And when you come back, you come back to look — at the work as it actually is, with eyes that haven’t been staring at it all week. A fortnight away from a novel draft can reveal the sagging chapter more clearly than a fortnight of rereading it.

A rest that quietly turns into months of avoidance is a different thing. It’s fine to call it that, without drama. It may even be telling you something, which brings us to the harder question.

vi. When quitting is reasonable

Persistence is not a virtue on its own. It is a tool, and tools are for something. So it’s worth separating two feelings that look alike from the inside.

Short-term boredom says: this part is tedious, and I still want what it leads to. Scales are dull; the music is not. Mismatch says: I no longer want what this leads to — or I never did; I wanted what it seemed to say about me. Boredom tends to lift when the practice changes. Mismatch tends to stay, even after you’ve fixed the method, rested properly and had a good week.

Someone who learned to code because everyone said they should, and who finds the problems leave them cold even when things are going well, isn’t failing a test of grit. They’ve gathered information. Stopping can be a conclusion rather than a collapse. It can also be partial: keeping a skill at the level where it’s useful and spending your best hours elsewhere. And stepping back from a pursuit says nothing about your worth. Output is something you make, not something you are.

vii. Progress without an audience

Applause is a terrible instrument. It arrives late, it is noisy, it depends on luck and timing, and it rewards visibility more than skill. If it is your only gauge, you will spend most plateaus convinced you’re standing still.

Here is a quieter definition: progress is a change in what you can do, notice or choose, measured against your own earlier work under similar conditions. It includes the mistakes you now catch, the choices that are now deliberate, the parts that have gone automatic. It includes knowing more precisely what you don’t yet know.

The advantage of that definition is that you can check it. Keep a few old pieces. Compare them honestly rather than hopefully. Record a passage and listen again in a month. Rerun an old exercise. You may find that nothing has changed — which is useful too, because it tells you to change the method, not your opinion of yourself.