Writing
Two weeks later we could not tell what we had checked from what we had assumed
Both had been written in the same document, in the same voice, on the same day. By the time it mattered they were indistinguishable, and decisions had been built on both.
August 2026 · 2 min read
While building the corpus we wrote as we went: what the data showed, and what we thought it probably meant. Same document, same voice, same afternoon.
Two weeks later we read it back and could not reliably tell which sentences were which.
That is not a memory failure. A checked statement and a confident assumption look identical in writing, because the thing that distinguishes them, whether you went and looked, leaves no trace in the sentence.
Two lists, kept apart
So they get separated at the moment of writing, not afterwards. What we confirmed, and against what. What we are assuming, and what would change it. Later is too late: by then work has been built on top of both, and pulling them apart means re-checking everything.
The systems enforce the same split. Where the evidence does not support a figure, the output says unknown instead of producing a confident number. A scorecard with four measured dimensions and three unknowns is a different object from one with seven scores, and only one of them can be argued with.