You follow the tutorial, every step makes sense, and then you open your own editor and nothing comes. That gap is not a memory problem, and it is not a sign that you are slow. Understanding and doing are built by two different kinds of practice, and watching more explanations only builds one of them.
You watch the lecture on pointers. Every line makes sense. The examples compile, the diagrams are clear, and by the end you would say, honestly, that you understand pointers.
Two days later you sit down to write a linked list and cannot start.
Nothing was lost in those two days. What the lecture gave you and what the blank file asks for are not the same thing.
Following an explanation and producing an answer are two different abilities. Only one of them is built by watching.
The two kinds of knowing
The first kind you are conscious of. It comes from explanation, it is quick to acquire and slower to use, and you can adapt it to a situation you have not met before. You can list the flags of ls. You can define a race condition. Learning researchers call this declarative knowledge.
The second kind you cannot easily put into words. It comes from doing rather than from being told. It is slow to build and then fast to use, and it does not carry over as well to situations that differ from the ones that built it. It is what makes you reach for the right command when a directory looks wrong, or sense that a piece of code has a race in it before you can say why. The usual name for it is procedural knowledge.
One caution. The distinction was worked out to describe how people learn languages, not how they learn C. Applying it to your assignment is my extension of it, and it is better treated as a way to organise practice than as a claim about engineers.
The kind you can explain is quick to get and slow to use. The kind you cannot explain is slow to get and fast to use.
Why this does not go away with experience
An engineer with eight years of work can stand at a whiteboard and describe the locking rules of a kernel subsystem accurately, from memory. Three days later the same engineer adds a call inside a section holding a spinlock and does not notice that the call can sleep. Same gap, later in the career, with higher stakes.
This is a different question from how deeply you know something, which is the subject of Four Levels of Knowing. Most Engineers Misjudge Which One They Are On. That piece is about the level you have reached; this one is about which of the two kinds your last month of study actually built.
Why watching it again does not help
This is where students lose the most time, and it is not laziness. It is a correct diagnosis followed by the wrong remedy.
You freeze on the problem, conclude that you must not understand the topic well enough, and go back to the video. That is more practice for the part that was already working. What that feeling of smooth understanding is worth on its own is the subject of Rereading Your Notes Feels Like Learning. It Mostly Isn’t.
An AI assistant makes this easier to do without noticing. It supplies a clear explanation the moment you want one, so the explainable half stays fully stocked and the discomfort that used to tell you the other half was thin never arrives.
Watching the lecture again is practice for the half that already works.
The practice that builds each kind
The two are fed by different activities. Explanation, testing yourself from memory, and spreading your review across several days feed the kind you can state. Varied practice, and mixing related problems together instead of grouping them, feed the kind you cannot.
Two conditions on that. Mixing helps when the problems are close enough that you have to decide which approach applies; mixing pointer problems with Git commands is task switching under a better name. And repetition is not automatically practice, which is the same reason Ten Years of Work Is Not Ten Years of Experience.
It would also be neat to say that training one kind suppresses the other. That is not established. Treat this as a way to plan your week rather than as a claim about your brain, because the practical part does not depend on it.
Varied practice is not the same exercise with the labels changed. The variation has to sit in the part that decides the answer.
What to do this week
- Before you watch another explanation of something you can already describe, close everything and spend twenty minutes on a problem you have not seen. When that order is the right one is worked out in Try First or Read First Is a Decision, Not a Temperament.
- Change what decides the answer. A doubly linked list, then one with a sentinel node, then one where you must free everything correctly, is three repetitions. The same list three times is one.
- Mix related problems in one sitting and keep unrelated ones out.
- Spread the reading out and put the practice where the distance is. Re-reading next week is easy; meeting the idea next week in a form you have not seen is the scarce part.
- After each session write one line: did I explain something, or did I produce something? A month of each looks identical in your notes and leaves you in very different places.
One test for this week. Pick something you are confident about — pointers, grep, why a driver’s probe runs late. Set fifteen minutes, close every tab, and solve one instance you have not seen before. Note the moment you stop. That is the only honest measurement in this article, and you can take it yourself.
This is the part that is hard to arrange alone. Varied practice needs someone to supply the variation, and the variation has to be chosen well — the next problem must differ in the part that decides the answer, not in a part that merely looks different. That is why our sessions are live and mentor-led with attendance enforced, and why they are built around problem scenarios: so that the second problem is not the first one again.
Explanations produce the knowledge you can state. Only doing, in conditions that keep changing, produces the knowledge that arrives in time.
None of this makes explanations a waste. You cannot practise something nobody has explained to you. The trap is that the explanation is the easy half, it feels like progress while you consume it, and it is now available the moment you ask — which is the older point in Information Is Free. Skill Is Not. Where it gets decided is the two halves of your evening.
— Raghu Bharadwaj
