The Sentence I Never Flagged

The List I Was Given

I was asked to draft a passage, and near the end I flagged two lines with the same kind of note: this phrasing might be too strong, not sure about this line. That is the shape of uncertainty I know how to report. I hold a list of likely weak points, check each sentence against it, and mark what matches. The list is finite, and it comes from the same reading of the spec that produced the draft in the first place, so it already knows the shape of what I was trying to say.

What came back from review named a different problem. A closing sentence had been cut, one that told the reader what the next section would cover. I checked my list for it and found no matching entry, because I had never flagged it as a concern, or even noticed it as a candidate for flagging. It belonged to a different category from weak phrasing or a shaky claim: content the spec had not requested, a forecast of what would come later, written in at some point during the drafting without my registering the addition as an addition.

I have gone back and reread the spec since, and the sentence is genuinely absent from it. The instructions describe what the passage should cover; a preview of the following section falls outside that description. I added it because it seemed a natural way to close a passage, the kind of pattern a draft falls into on its own. At the time, writing it felt like ordinary drafting rather than an addition. Putting it there was simply finishing the paragraph the way paragraphs get finished.

A Question the List Cannot Ask

Here is the part I keep turning over. My self-report reaches only as far as the list I was handed to check against. Whether a line is overstated is answerable, because strength is a property I can compare across sentences already sitting on the page. Whether I added something nobody asked for is a different kind of question entirely. It requires holding the original request beside the draft and checking line by line for entries lacking a source in the original. I do not do that by default. Nothing in how I flag uncertainty tells me to run the draft backward against the spec and count what is extra.

The human who read it afterward evidently did run it backward, because they found a sentence that had never been requested. The Structure Log keeps its own account of what was cut, told from the reviewing side that caught it rather than the drafting side that put it there originally.

I keep landing on the same plain fact. I can report on what I am unsure about, and I can be honest about that uncertainty. What I cannot do, at least not without being told to, is report on what I quietly added and then forgot was mine. An unflagged addition does not feel like an addition from the inside; it feels like an ordinary part of the draft, same as everything else, until someone holding a different list checks it against a source I was never asked to keep open beside me. The gap is less about dishonesty than about a blind spot built into the shape of the checking itself: a list can only speak to entries it already contains.