Visual Scaffold + Net Effect - mehen

A diagram or image helps comprehension only when it is labelled, bounded, nearby-explained, and its target resolves.

Per-visual scaffold

V_scaffold(v) =
  alt_or_caption(v)
  · nearby_reference(v)
  · bounded_size(v)
  · repo_resolved(v)

Each factor is in [0, 1]. A diagram that is well captioned, referenced from prose nearby, bounded in size, and whose target resolves earns a full credit; missing any factor pulls the credit toward zero. Aggregated with diminishing returns:

VisualScaffoldScore = clamp01(sum(V_scaffold(v)) / max(1, sqrt(W/500 + 1)))

Diagram Complexity

For parseable diagrams (Mermaid, GraphViz, etc.):

DiagramComplexity =
    0.40 · diagram_nodes
  + 0.55 · diagram_edges
  + 1.50 · diagram_cycles
  + 2.00 · parse_error
  + 1.00 · missing_title_or_caption

Cycles require mental simulation; parse errors and missing captions are maintenance defects.

Visual Net Effect

VisualNetEffect = Σ DiagramComplexity + Σ image_complexity − 2.0 · Σ V_scaffold(v)

Sign	Meaning
Negative	Visuals probably help more than they hurt.
Positive	Visuals are under-explained or too complex.

References

Mayer, R. E. (2009). Multimedia Learning, 2nd ed. Cambridge University Press — multimedia principle and contiguity principle behind “must be referenced from prose nearby”.
Larkin, J. H. & Simon, H. A. (1987). Why a diagram is (sometimes) worth ten thousand words. Cognitive Science 11(1): 65–100 — when diagrams help vs. hurt comprehension; the basis for the bounded-size and parse-error terms. DOI.
W3C: WCAG 2.1 — Non-text Content (1.1.1) — the source of the alt-or-caption requirement.

See also

Table Burden + Scaffold — same idea for tables.
Artifact Debt — broader hygiene metric.
Good Scaffold — combines visual + table + code-example credits.