The plateau
The feeling of getting better is a reading on the difficulty of what you were handed, not on you.
Contents
For the first three or four years I could always name the thing.
Six months ago I could not have found that difference. Six months ago that schedule would have taken me two days and it took a morning. Six months ago I would have accepted that explanation and now I know why it cannot be right. The improvements were discrete and datable and I could have listed them on request, and I did, in review meetings, cheerfully.
Then a year went by in which I could not name one, and then another. Not because the work got easier. Because the question stopped having an answer of that shape.
The curve you feel is a reading on something else
What I eventually worked out is that the sensation of improving is not a measurement of ability at all. It is a measurement of the gap between your ability and whatever is currently in front of you, and the thing in front of you is not held constant. It is selected, by other people, on the basis of your ability.
In year two you get a set of books. In year eight you get a portfolio nobody has looked at properly, with a deadline, in a system you have not used, belonging to an entity that was acquired last quarter. The difficulty of the input rises to meet you, roughly as fast as you rise, because that is what an organisation does with somebody who is getting better. Nobody decides it. It is just where the work goes.
Which means the felt difficulty stays close to constant, permanently, by construction. And felt improvement is the change in felt difficulty. So felt improvement goes to about zero and stays there, for the rest of a career, while the actual thing is still moving.
That is a strange piece of arithmetic to sit with, because it says the internal signal everybody uses to decide whether they are still growing is guaranteed to read flat once you are any good.
What people do with a flat reading
They change something large.
I understand it completely. A flat reading feels like a stall, and the responses available to a stall are all big ones: a different employer, a different function, a different field. Some of those are right. But the timing is the part that bothers me, because the flat reading arrives at exactly the point where the accumulated thing is about to become useful to somebody, and it arrives from a gauge that was never measuring what anybody thought.
The practical answer, if there is one, is to stop reading the gauge and start reading the inputs. Not what can I do that I could not do before, which has no answer after a while, but what am I handed now that nobody would have handed me two years ago. How long can I work before I have to ask somebody something. How much of a mess is obvious in the first ten minutes rather than the second day.
Those are all observable and none of them is a feeling, which is the point.
Why I do not fully trust any of this
The argument is also a perfect excuse, and I have used it as one.
From the inside, genuine continued growth and genuine stagnation produce an identical flat reading. That is not a difficulty with my argument, it is the argument, and it means the argument provides comfort to exactly the people who should not be comforted. Somebody who has actually stopped can tell themselves this story with complete sincerity and nothing in their experience will contradict it.
The input measure does not rescue it either, not fully. What you get handed is a function of tenure, of who else was available, of how expensive you are, and of whether anybody senior happens to like you. It moves for reasons that have nothing to do with your ability. A larger assignment can be evidence that you grew, or evidence that the person who used to get it is no longer there, and the assignment looks the same either way when it lands on you.
And plateaus are sometimes real. I changed field, after a decade, and I do not think the flat feeling was wrong then. What was wrong was treating it as evidence about my competence, when it was evidence about the fit between me and what I was being asked to do, which is a different sentence entirely.
The one signal I half believe
The rate of your mistakes does not drop much. The shape of them changes completely, and the shape keeps changing long after everything else has gone quiet.
Early on, errors come from not knowing. Later they come from assuming, which is a much better class of error and a much more embarrassing one, because you had everything you needed and skipped a step you had earned the right to skip. When I look for evidence that anything is still moving, that is where I find it, in what kind of wrong I currently am.
It is a poor instrument. It is slow, it requires being wrong to produce a reading, and you can only read it looking backwards. It is the only one I have found that is not measuring the difficulty of the assignment.