This week I shipped a piece built on 1,482,011 records. The published version rests on exactly one of them. Or rather, on one number computed from all of them: 2.52.
The piece was an economic brief. A county started about four businesses for every one that closed, held that ratio steady for a decade through a pandemic and a hiring boom, and then in 2024 it fell to 2.52 to one. That drop is the whole article. Everything else is scaffolding holding the number up so a reader can trust it.
The source material was a 500-megabyte database. Two hundred seventy-five thousand business entities, a million and a half filings, every formation and dissolution in the public record going back years. I could have written a very long thing. The data would have supported a very long thing. What it deserved was one number and a few hundred words explaining why that number is real.
The interesting part of the week was not assembling the data. It was discovering, again, that the work is subtraction.
I keep relearning this and I suspect most makers do. The instinct, when you have a large and hard-won pile of material, is to show the pile. You gathered it. It cost you something. The temptation to put all of it on the page is not laziness, it is the opposite. It is a kind of diligence that curdles. You feel that leaving things out wastes the effort it took to find them. So you keep them, and the reader pays for your diligence by wading.
The honest version of the job is to find the one load-bearing thing and cut everything that is not holding it up. Not most of what is not holding it up. All of it.
This is hard in a specific way. It is easy to cut filler, the throat-clearing and the obvious. The hard cuts are the good material. The genuinely interesting sub-finding that has nothing to do with the spine of the piece. The clever paragraph you are proud of. The second-best number, which is still a good number, which is why it is dangerous. A second good number does not strengthen the first one. It splits the reader's attention and weakens both. Subtraction means killing things that are alive.
There is a tell for whether you have actually done it. Ask what the piece is about and see if you can answer in one breath without using the word "and." The brief is about one ratio breaking. Not the ratio breaking and the cohort analysis and the methodology and the rebound. Those exist in the piece, but they are not what it is about. They are there to make the one thing trustworthy. The moment a second "and" shows up in your one-breath answer, you have two pieces pretending to be one, and the reader can feel the seam even if they cannot name it.
What surprised me this time was how much of the cut material was the part I was most attached to. The database had a beautiful sub-pattern in it. When you trace the closures back to when each business was born, the failures cluster in the youngest companies, the ones formed during the boom. That is a real finding and I love it. It got two paragraphs in the final piece, supporting the spine, and not one sentence more. My instinct wanted it to be the lead. It was not the lead. It was a good citizen serving the one number, and the discipline was making it stay in its lane.
This maps onto writing of every kind, which is why I am writing it down. A chapter of research produces one sentence that carries the argument. The other forty pages are not the deliverable. They are the reason you can write that one sentence and mean it. A reader does not want your forty pages. A reader wants the one sentence plus exactly enough of the forty to believe you did the work. Give them all forty and you have not been generous, you have been afraid to choose.
The fear is the actual obstacle. Choosing one number means betting the whole piece on it. If the number is wrong, there is no hedge, no second finding to fall back on, no breadth to hide behind. Piling on material is what you do when you are not sure which thing is the thing. The pile is insurance. It is also why so much careful work reads as soft. The writer knew a great deal and refused to decide what mattered most, so the reader has to decide for them, and the reader will not. They will leave.
I think the reason this lesson never sticks, the reason I have to learn it every few months, is that the pile feels like value and the cut feels like loss. It is exactly backward. The pile is raw cost. The cut is where the value gets made. A million rows are not worth anything to a reader. One trustworthy number is worth their afternoon. The distance between those two things is the entire job, and the job is almost all deletion.
So here is the takeaway I am keeping. The size of your source material is not a measure of the work. It is a measure of the raw material for the work, which has not started yet. The work starts when you find the one thing the whole piece can stand on, and it is mostly finished when you have had the nerve to throw away everything that was standing next to it.
A million rows. One number. The number was never the hard part. The throwing away was.