What expert knowledge is

If an expert cannot tell you how they do it, the first explanation that comes to mind is that they simply know more: they hold more information than fits into the time of an answer.

I do not think that is it. Or rather, it is true and irrelevant — because the difference is not one of quantity, it is one of form. Expert knowledge is not the beginner's knowledge multiplied. It is that knowledge rewritten into another format, and the new format is not made of propositions.

Rules versus cases

The beginner has rules, and applies them. No parallel fifths, the seventh resolves down, the leading note rises. They are sentences, they can be repeated, they can be used to check what you write. And they get applied slowly, one at a time, while looking at them.

The expert no longer applies those rules. They have seen a few thousand concrete situations and they recognize which one they are in. They do not derive the solution: they find it already there, together with the recognition. Which is a completely different mental operation, not a faster version of the same one.

The simplest proof is that the expert also recognizes the cases in which the rule must be broken, and in doing so is not applying a meta-rule about when to break rules. They are recognizing a situation, the way you recognize a face. Ask them why breaking it is right there, and they will construct a justification — and the justification is nearly always plausible, orderly and false: it is not what decided, it is what they invented afterwards in order to answer.

As a composer I have learned to distrust my own explanations for exactly this reason. When I cut a passage, the reason I then give for it is a reconstruction, not a record.

Why it stops being sayable

The move from rules to cases has a price, and the price is precisely access.

A rule is already made of words: it travels by being said. A recognized case is not made of words — it is a configuration, a resemblance to hundreds of other situations that nobody ever labelled. To say it, you would have to list the resemblances, and it is not that the expert is hiding them: they have never had them in the form of a list.

There is then a second effect, which is what makes the phenomenon inevitable rather than accidental. Whatever is used constantly gets compressed: the brain does not keep in explicit form what it has to run fast and often. So the most practised part of a competence is, by construction, the least accessible. It is not that the expert forgot to tell you the important part. It is that the important part became the part that does not get told.

The sign that tells you an expert is real

There is a practical clue I have used for years to tell whether somebody actually knows a thing or has only studied it.

Someone who has rules and no cases answers questions completely and in order. Someone who has cases answers badly: they start, they break off, they say "it depends", and then — this is the point — they give an example. Almost always a specific one, with useless detail, of the "once it happened to me that" kind.

That example is not an anecdote and it is not a waste of time. It is the native format of that knowledge. They are passing you a case because that is the form in which they hold it. The people who answer best in the abstract are often the ones who can do least.

It also works in reverse, and this is the uncomfortable part: I come across as more convincing on the things I master least. On composing, which is what I have been doing for forty years, I speak worse than on any other subject.

What it means for anyone who wants to learn

If expert knowledge has this form, two practical consequences follow on their own.

Asking for explanations yields little. Asking for cases yields a lot. The right question to a master is not "how is it done" but "tell me about a time it came out wrong, and what you did". The first produces a rationalization, the second produces real material.

Exposure beats comprehension, at the start. If competence is made of recognized cases, then what you need is to go through many of them, not to understand a few deeply. It has an unromantic implication: in the early stages it is worth more to play twenty pieces badly than one impeccably, because you are building the warehouse of recognitions, not the quality of a performance.

But the question that makes all this awkward remains. If what matters does not travel through words, and if the expert has no access to what they know, what is a teacher for?

Which is the next piece, and the answer is not the usual one.