19 A cause written on a chart is a hypothesis

What people usually ask

Plot this reactor temperature log and annotate anything unusual.

What comes back

A chart with “feed interruption at 14:20” written beside a spike in the same type as the axis title, though nothing in the log says what happened at 14:20.

Ask this instead

Attached: temp_log.csv (timestamp, reactor temperature in C, one row every 10 s) and run_notes.txt (the operator's notes for the same shift).

Plot temperature against time. Then mark features on the chart, with no causes:
- A spike, step or drift gets a label with its start time, its duration and its size in C, nothing else.
- Below the chart, list every feature as a question for the operator, for example "What happened at 14:20? Temperature rose 6 C in 40 s and came back in 3 min."
- Events may come only from run_notes.txt, quoted word for word with the time the note gives. Where a note falls within 2 minutes of a feature, show the quote beside that feature and say the times match; do not say the note explains it.
- Do not name a cause anywhere: not in a label, not in the list, not in the title.

Attach: the log CSV and the operator’s run notes.

Why it works: a label with a time and a size records what was measured, and the cause stays a question until someone who was there answers it.