What I haven't been able to kill It isn't a project.

I built a framework with elimination at the front of it. Remove, then leverage, then add, in that order, and I still think the order is right.

Here is the problem with me saying that. I have never removed anything.

Not nothing at all. Ideas get killed at the evaluation stage constantly, which is cheap and does not count. What I mean is that I have never taken something that was built and working and shut it down. Not once. Every instance I can point to in my own work is a leverage story wearing elimination’s clothes.

So this is the honest edge of the thing I am supposed to be good at. Rather than defend it, I want to write about the one thing I have genuinely tried to kill and failed to, which is not a project.

What I have actually tried to kill

At the end of 2023 I wrote a year-end review. Two paragraphs of it are confident, and they had earned it, because that year had gone well. A role that had not existed before existed now. A method that had been mine was becoming the team’s.

The third paragraph says the challenge I had continued to face was not feeling like I was adding the appropriate amount of value consistently.

I have been trying to kill that sentence for five years.

The receipts, in order

In December 2021 I wrote that I wanted to push myself on results, because I had produced a lot of reports and offered a lot of suggestions and insights, and I did not have a solid system of tracking when the changes actually happened, or what they did once they had.

That is the whole problem stated cleanly, by me, at the time. I did not build the system. Five years later there is still no system.

In 2023, the sentence above. Value in spurts. It would take time to reveal its true value.

In August 2023 I sat in a room and drew a triangle: consistency at the base, competence above it, confidence at the top, and an arrow leaving the top that said built by repetitive failure. Beside it I wrote that when you are staring at your gap, you have to have the skill to look back and see your gains. I wrote that down because somebody said it and it landed. It landed because I do not have that skill.

And in 2026, in a staff meeting, one line that is the closest I have come to an answer: don’t underestimate your contribution because you don’t get the credit.

Read those four in order and you can watch one person try four different tools on the same problem. Build a measurement system. Wait for time to prove it. Learn to see the gains. Decide it does not need the credit.

The last one is a conviction, not a mechanism. That is what you reach for when you have stopped expecting to solve it.

Why I think it will not die

Because the work I am best at is invisible by construction.

Nobody gets credit for the hire they didn’t make. I have written a version of that line a dozen times and it always comes out too neat, so here is the unneat version. Adding is legible. A new hire has a start date. A new tool has an invoice. A new initiative has a launch. Each of those produces an artefact, and the artefact is what gets discussed in the review, not the outcome.

Removing something, or using what you already have, produces no artefact at all. The best possible version of my work is a thing that did not need to happen, and there is no way to point at that. You cannot show somebody a meeting that never got scheduled.

So the doubt is not a confidence problem. That is what I thought it was for about four years. It is a measurement problem, it is structural, and it is not going to resolve by my feeling better about myself.

Where that leaves me

Two things, and they do not fit together neatly.

The first is that the 2021 version of me was right and lazy. There is a system to be built. I could track what I recommend, when it ships, and what moves afterwards. It is not conceptually hard. I have not done it in five years, which at some point stops being an oversight and starts being information about how badly I want the answer.

The second is that some of it genuinely cannot be measured, and the 2026 line is the correct response to that portion. Not all of it. Some.

What I do not know is the ratio. How much of this is a system I have avoided building, and how much is the actual price of working this way?

That is the thing I have not been able to kill. Not a project. A question, and it has outlived every attempt I have made on it.