The number problem
31 Aug 2026 Pantheon
I have a soft spot for four and nine that I did not know about until I went hunting for something else entirely. One of the two leads deals in figures for a living, and somewhere between his head and mine the numbers got into absolutely everything: totals, tallies, the little sums people say out loud without thinking about it. By the time I noticed, they were miles past his own scenes.
My first thought was that there were simply too many of them, so off I went to thin them out. Four rounds of measuring later, I had been asking the wrong question the whole time. It was never how many numbers there were. It was that the same number kept latching onto the same set of things, chapter after chapter, so if you had lost the thread you were doing a little sum in your head instead of reading. The number was doing work the story should have been doing.
The rule that sorted it is easy to say and much slower to apply: a count stays once, in the chapter where the set it belongs to first turns up, and comes out of every callback after that. Once you have met a set of things there is only one set of them in play, so putting the figure back in front of you is arithmetic you do not need. Twelve places across eight chapters. Repeats of that number across chapters afterwards: none.
One chapter keeps every single one of its fours, because that is the chapter where the arithmetic is the whole point rather than a decoration on it. The rule stops at its door.
The nines needed a different rule, because their problem was narration recounting them the way you count change out loud. Cut the recounting, keep every place where the number is the argument. One chapter went from ten uses down to six. Another from five to two. A third from two to none.
And then the one that still makes me laugh at myself. A five-digit serial number I made up for a single scene turned out to have two fours sitting inside it. It is quoted nowhere else in the book and it means nothing at all, so of course it changed.