draw-the-system-not-your-studylisted

# Draw the system, not your study If the source's own figures are on disk, follow `draw-the-source-figure-panel-for-panel` first and use this skill only for the slots it leaves unassigned. This one is the index to use when they are not: the target paper is absent from the supplied related work, the task is specified in prose, or the field has no standard figure for this object. Then the system itself is the only index available, and a plan built by walking your own hypothesis list will pass every coverage check it owns while leaving a reader unable to see the thing that was studied. ## Three rules that decide this before the slot list does **A count belongs in the caption; the array belongs in the panel.** If your caption can state how many entities, edges, samples, atoms or pixels there are, then you are holding an object with that many of something, and you can draw its matrix, its distribution or its map. A bar whose height *is* a count carries only the number you already wrote beside it. The usual form of this failure is a "data overview" figure whose source is an inventory or a file-hash manifest -- a document about the data -- rather than the arrays the run loaded to do the work. **A panel whose caption is a verdict on your own preregistration is not the panel carrying the result.** If a slot's plan already contains an *if refuted* branch, that slot has been spent on your study's bookkeeping. Put the per-entity result in the panel and the shortfall in one sentence u