Since figurative drawings are not so well adapted to the task of reasoning or explaining, the Late Antique inventors of node-link diagrams were careful to omit figurative elements from them.
Their successors since that time have repeatedly attempted to add figures, to regulate the distances between the nodes, to impose a standard orientation (for example, growing upwards like a tree) and to strictly align such diagrams. Those mistaken "improvements" indicate that later generations have not fully understood the genius of the original invention.
Hegarty quotes research suggesting why. We have a misplaced faith in fussily drawn diagrams,
with a strong preference for displays that emphasize high-fidelity spatio-temporal realism, even when these displays result in poor performance... This may come from a folk fallacy that perception is simple, accurate and complete, whereas perception really is hard, flawed and sparse.
Markus Knauff's book (see my recent post) suggests an additional cause for the fallacy: if reasoning is largely spatial, and is taking place in a part of the mind that is not accessible by introspection, most of us are likely to be quite ignorant about what constitutes an effective explanatory method.
A similar point about why figurative art should be excluded from effective diagrams has been made by Manfredo Massironi in his theoretical account of Hypothetigraphy, the subject of a post on this blog in 2011. Massironi took the view that any diagram explaining abstract matters needs to be limited to what he called "precise marks" only: "Precise, clear lines contribute in conveying the impression that the depicted forms are mental constructs, not representations of natural objects."