The first year you store everything

The first stretch of my career was spent moving between small Ottawa practices and my own self-employed work, and every one of them ran on different software and different conventions. CaseWare in one place, ProFile somewhere else, Sage and QuickBooks and Adagio depending on the client. Different file naming, different review habits, different idea of what a complete working paper looked like.

I remember it as relentlessly tiring, and for a long time I assumed that was because I was not very good yet.

Not knowing what to ignore

The tiredness has a specific cause and it is not effort. It is that a beginner has no basis for discarding anything.

You cannot tell which details will matter, so you keep all of them. How the file was named. What the client said on the phone before getting to the point. Which line the previous preparer adjusted and the two words they wrote next to it. The order somebody did things in, which you copy without knowing whether the order means anything. All of it goes in, undifferentiated, because sorting requires a criterion and the criterion is exactly what you do not have.

That is enormously expensive to run. It is also the only thing available.

What makes an experienced person fast is not that they store less. It is that they store compressed. They have seen a shape enough times that a new instance costs almost nothing: one stored exemplar plus the difference. But the exemplar is not issued to anybody. It is what several hundred undifferentiated cases turn into after enough of them have been through you, and there is no route to the compressed form that does not pass through the raw one.

The person who skipped it

This is why the shortcut produces a recognisable kind of professional. Someone who knows the rules and misses the cases.

They can state the treatment accurately. What they cannot do is notice that a file feels wrong. That sensation is not intuition in the mystical sense, it is a comparison running against a large stored set, reporting a mismatch before the mismatch has a name. It is the single most useful thing I have, and it is completely unavailable to someone who has not done the storing, and it cannot be taught in the afternoon because it is not information.

Two things I have to concede

Most of what I stored was worthless.

I retained an enormous amount about menus, keyboard paths, and one firm’s conventions for naming a folder, and all of that expired within two years and compressed into nothing. The material that turned out to be raw material was almost entirely about how things go wrong: the specific way a set of numbers looks when somebody has been rounding to make it tie, the sound of a client who has not told you something yet. I could not have known that at the time. Nobody could, because knowing which class of detail pays off is the judgement the whole process is supposed to produce.

Which means the phase is both necessary and genuinely wasteful, and you cannot fix the waste by being selective earlier. Selection requires the thing you are trying to build.

The second concession is that the compression is not reliably better. When an experienced person meets a case that resembles a stored one and is not the same case, they will be fast and confident and wrong, and the beginner storing everything indiscriminately is occasionally the only person in the room who notices, precisely because they have no pattern to fall into. I have been on both sides of that. Being the confident one is worse.

What I cannot locate

I do not remember most of what I stored. I could not reconstruct a single one of those files now, and I have no memory of the point at which the storing turned into something else.

There was no year where I noticed it happening. I only know it did because at some stage the work stopped being tiring in that particular way, and I could not tell you what I lost when it did.