How you writeExperiment → prose
Run the experiment.Then let results enter the draft.
Topic in hand but no runnable experiment, or results scattered across folders—settle a plan you can run, then carry it into Methods, Results, and Discussion, with key numbers tied to your records.
Have results? Pick a layout and attach runs later.
- 01
State hypothesis and controls
- 02
Run and collect logs
- 03
Draft Methods and Results
- 04
Settle into the venue layout
Name it before you write
Methods fail when controls, metrics, and data provenance are vague. Settle those in the experiment first.
Rewrites usually track experiment changes—not adjectives—so experiment and manuscript should share one line.
Numbers have a source
Tables, figures, and claims should match the run log. Models can organize prose; they must not invent unrun results.
You still decide what enters the main text versus the appendix.