Stripping away scales, labels, and references for simplicity can leave readers unable to read the chart

What it is

Over-pruning chart scaffolding removes scales, labels, units, references, annotations, uncertainty, or access paths needed for the task in pursuit of simplicity. Readers may still see the marks yet be unable to establish exact values, comparison baselines, definitions, or evidential strength. Readability is not the amount of blank space. Pruning should reduce irrelevant competition while retaining the structure that makes conclusions interpretable, checkable, and accessible.

Why it happens

Ticks and gridlines provide nearby references, baselines and targets define comparison origins, labels and units connect geometry to meaning, and annotations and uncertainty constrain conclusions. Removing them transfers alignment, interpolation, memory, and definitional inference to the reader and can create false certainty. Auxiliary elements may obscure data when too strong and increase error when too weak or absent. The useful balance follows the task rather than maximising or minimising a pixel ratio.

Studying it

Hold data and encoding constant while comparing scaffolding combinations. Predefine gist, rank, exact read-off, difference, threshold decision, and uncertainty judgment as separate tasks. Measure answer error, response time, confidence calibration, missed references, and gaze distance across desktop, mobile, projection, print, and zoom. Include keyboard, screen-reader, and low-vision tasks: equivalent performance in the visible plot does not establish complete access paths.

Where it stops holding

Gist may not need dense grids, and direct labels can replace part of an axis or legend; a data table or operable detail can serve exact lookup. A zero baseline is not required for every graphic and question, but must remain visible when the intended difference depends on it. Sources, samples, missingness, and uncertainty may sit in an adjacent note rather than the plot if their association survives embedding, export, and sharing.

Applying it

• Derive the minimum evidence set from the task: variable name, unit, scale, comparison reference, critical threshold, data state, and uncertainty. Then place each item on an axis, annotation, note, or data table.
• Tune hierarchy, density, and placement instead of choosing “keep all” or “delete all.” Use local reference lines for comparisons that actually need them.
• Verify that screenshots, embeds, print, and narrow layouts retain the title, definition, and source needed to interpret the conclusion.
• Test gist and exact tasks with target users. Restore scaffolding or an alternative route when pruning raises error, time, or overconfidence.

Related

• Same group: U1.14.1 Non-data elements consume attention while carrying no information · U1.14.3 The test is whether removal affects reading
• Adjacent: U3.05.1 Tick density should match the precision readers need · U1.03.1 Length judgment requires a common starting point
• Search terms: chart scaffolding · over-pruning · reference line · uncertainty annotation