A figure has a hierarchy whether or not anyone designed one, and scientific review pushes toward addition rather than priority. Decide what the figure has to establish before deciding what it contains, then rank its elements: one focal point, a quieter supporting layer, a nearly silent context layer.
The pattern
The mechanism figure has eleven labeled elements, four arrow types, two membranes, a legend and an inset. It is accurate. It was reviewed by the science team, who confirmed that nothing is missing.
Nothing being missing is the problem.
Why it breaks
A figure is a hierarchy whether or not anyone designed one. Something is seen first. Something is grouped with something else. Something is read as causal because it has an arrow.
When those decisions are not made deliberately, they are made by accident — by which element is largest, by which is most saturated, by which happened to end up in the center. The viewer’s eye is directed by the drawing’s incidental properties rather than by the argument.
Scientific review makes this worse, not better, because scientific review checks for accuracy and completeness. “Is anything wrong?” and “is anything missing?” are the questions asked, and both push toward addition. Nobody in that review is being asked whether an investor can find the point in four seconds.
So the figure grows. Each addition is individually defensible and the cumulative effect is a picture with no focal point, which the viewer resolves by not looking closely at all.
What changes
Decide what the figure has to establish before deciding what it contains.
A figure that has to show causality needs direction and sequence, and can drop anything that does not participate in the chain. A figure that has to show comparison needs a shared baseline and two things, not seven. A figure that has to show scale needs one reference the viewer already understands.
Then rank the elements. One focal point. A supporting layer that is deliberately quieter — thinner strokes, lower contrast, smaller labels. A context layer that is nearly silent. Anything that does not fit those three tiers is not context; it is a second figure.
And where the content genuinely has states, sequence or a decision point, one frame is the wrong format. That is what a sequence of figures, or an interactive one, is for.
The practical result
The figure gets a focal point, and a viewer who spends four seconds on it comes away with the right idea rather than an impression of complexity.
Accuracy does not fall. The detail that came out of the frame goes into a caption, an appendix figure, or a second panel — read by the people who want it, in the place they will look for it.
Complexity is not the enemy of clarity. Undifferentiated complexity is.
