Some people play guitar for fifteen years and end the fifteenth year playing exactly what they played in the second.
They practised. Genuinely, regularly, for hours. And they got very good at the eight things they already knew, because that is what they spent fifteen years doing. Meanwhile someone three years in is playing things they can't, having put in a fraction of the total time.
The difference isn't talent and it usually isn't hours. It's that one of them spent the time doing what they could already do, and the other spent it on what they couldn't.
Why repetition on its own doesn't work
Practice feels productive in direct proportion to how smoothly it goes, which is precisely backwards.
Running through the piece you can play, redoing the exercises you can already solve, rewriting the code you've written before — all of it is pleasant, all of it produces a sense of having worked, and almost none of it changes what you're capable of. Your body and brain adapt to demands they can't currently meet. Meet a demand comfortably and there's nothing to adapt to. You are maintaining.
The reason it's such a common trap is that the alternative feels bad. Working at the edge of what you can do means failing repeatedly, in a session where progress isn't visible, and coming away with the distinct impression you're worse than you thought. That impression is the signal you were after. It's just an unpleasant one, which is why so few people go looking for it, and why "just put the hours in" quietly fails so many people who did exactly that.
The same mechanism sits under the Input-Only Trap in reading — consuming feels like progress because it's smooth. Practice has its own version, and smoothness is the tell in both.
Three things worth doing
Practise the Broken Bit.
Isolate the specific part you can't do and work on that, rather than running the whole thing and hoping the difficult section improves by association. Musicians know this and mostly ignore it: playing the piece top to bottom eight times means eight repetitions of the bars you already own and eight fumbles through the bar you don't. Take the bar. Slow it down until you can play it correctly, then speed it up. In anything else it's the same move — the one function you don't understand, the one type of question you keep getting wrong, the one phrase that always comes out wrong. Twenty minutes on the broken bit beats two hours of the whole. It's also considerably less fun, which is the entire reason nobody does it.
Work at the Edge of Awkward.
There's a level of difficulty where you succeed most of the time but fail often enough to notice — roughly one in five, as a rough feel rather than a rule. Below that you're maintaining. Far above it you're just failing, and constant failure with no traction teaches nothing except that you're bad at this. The skill is finding that band and staying in it, which means adjusting constantly, because the band moves as you improve. Practically: if today's session felt entirely comfortable, next session is too easy. If you got nothing right at all, it was too hard and needs breaking into a smaller piece.
Build in an Immediate Check.
Feedback matters most when it's fast. Knowing you were wrong within seconds lets you correct; finding out three weeks later means you've been practising the error the whole time. This is the real reason self-teaching goes wrong — not lack of discipline, but a long gap between doing it wrong and discovering it. So engineer the check into the task: answers you cover and reveal, code that either runs or doesn't, a recording of yourself played back, saying the answer out loud before you look. Anything that closes the loop in seconds. Where a person can watch you and correct in real time, that beats all of it, which is what a decent lesson is actually buying.
Things that look like practice and aren't
Watching someone else do it. Useful for learning what to attempt, worth nothing as practice. The tell is that you feel you understand and can't produce it.
Rereading your notes. Recognition again, not production.
Preparing to practise. Reorganising the folder, buying the kit, finding a better course, building a schedule. All of it feels like the work and delays the moment of doing something badly, which is the moment that actually matters.
Doing the thing you're good at because it's satisfying. Legitimate — enjoyment is why anyone continues, and a practice regime with no pleasure in it doesn't survive. Just don't count it as the session.
The rough split people find workable is most of the time on the broken bits, some of the time on playing properly. Both are needed. Only one improves you.
How long any of this takes
Slower than you want and faster than you fear.
The awkward truth about the edge of awkward is that improvement there is almost invisible session to session — you're failing at roughly the same rate every time, because you keep moving the difficulty up as you get better. The only way to see it is to look back at something you couldn't do six weeks ago and notice it's now the easy part. Which is a strong argument for keeping some record, even a scrappy one.
And there's no honest way to say how many weeks anything takes. It depends on the thing, how often you're at it, and how much of your time is spent where it's uncomfortable. What's safe to say is that the person practising twenty focused minutes on the hard part will pass the person doing an hour of the easy part, and won't take long about it.
This is Pillar 03: Knowledge
Find the bit you can't do. Work where you fail sometimes. Close the feedback loop fast.
The reason this belongs in Knowledge rather than filed under self-improvement is that it's the same mechanism underneath everything in this pillar. Memory is built by retrieval, not rereading. Skill is built by difficulty, not repetition. In both cases the pleasant version of the activity is the one that does nothing, and almost nobody is told this, so a great many capable people conclude they've hit their ceiling when what they've actually hit is a method problem.
Being taught badly is not the same as being incapable. Neither is practising badly for fifteen years.
More at https://thequietreset.uk