Method // Checklist
Reading evidence before narrative
A practical sequence for separating what is observed from what is inferred.
Narrative is what the mind does with evidence the moment you stop supervising it. This is not a flaw to be fixed — story is how humans compress — but it means the order of operations matters enormously. If the story forms first, every subsequent piece of evidence gets sorted by whether it fits. If the evidence is pinned down first, the story has to earn its shape. This checklist is the sequence we run before we let ourselves say what we think is going on.
First pass: capture without interpreting
- Write down what was actually observed, in the flattest language you can manage. "The invoice was paid twice" — not "the process failed."
- Quote exactly where wording matters. A paraphrase is already an interpretation.
- Timestamp every observation. Sequence is evidence, and memory reorders events to fit stories.
- Note what you expected to see and did not. Absences are observations too, and they are the first thing narrative erases.
Second pass: separate the columns
Go back through the capture and mark every line: observed or inferred. The test is simple. Observed means a camera in the room would agree with you. Inferred means you are supplying a mechanism, a motive, or a cause. Most working notes fail this test in the first sentence, and the failure is invisible until you look for it — inference wearing the grammar of observation. "They ignored the warning" feels like a fact. What the camera saw is that a warning was sent and no reply arrived.
- Rewrite each inferred line so the inference is explicit: hypothesis: the warning was ignored — to confirm: check whether it was read.
- For every inference, write the observation that would distinguish it from its nearest rival. If no observation could, the inference is decoration.
- Keep the columns physically separate. The moment they share a paragraph, they will share a truth-status in the reader's mind.
Third pass: hunt the convenient hypothesis
Every investigation produces one explanation that would be cheapest to accept — the one that lets you stop looking, blames nobody in the room, or confirms what you already decided last quarter. That hypothesis deserves the most hostile treatment precisely because it will get the least resistance from everyone else. Ask what the strongest alternative is, and check the inconvenient one first. If the convenient story survives an honest attempt to break it, you have earned it.
Only then: narrative
Now write the story — with its seams showing. Every causal claim in the narrative should trace back to a line in the observed column or be marked as a hypothesis with its confirming test named. The narrative you produce this way is slower to write and less tidy to read. It is also the only kind that survives the arrival of the next piece of evidence, which is the entire point of writing anything down.
- Every observation is flat, quoted where it matters, and timestamped.
- Every line is marked observed or inferred, and the inferences carry their confirming tests.
- The convenient hypothesis has been named and attacked before being accepted.
- The final narrative cites its own evidence, claim by claim.