It's 90% boxes-and-arrows, 8% Scratch clones, and 2% of unique stuff

This strikes me as an important insight which leads to the question, “Why is the distribution so heavily skewed towards boxes-and-arrows?” If we’d be completely in the dark, couldn’t we expect a lot more variety? Maybe one possible answer has to do with the challenge that building a generic visual programming system is by definition domain-agnostic and therefore can’t take advantage of established metaphors in a target domain? I’m thinking of (as usual) music or video production tools, where a lot of the visual tools rely on metaphors that represent time by mapping it to (horizontal) space, which rests on the importance of time in these domains. If you can’t define certain specifics because you want to create something universal, the visualizations available to you will be limited to the most abstract of domains, often math — and then you end up representing minimal structure as an abstract graph, with generic arrows, their positions and lengths without meaning, pointing from and to equally as dimensionless boxes, all floating in a space that has no meaningful coordinates either — you can move things around and it doesn’t change anything other than the visualization.

The value of visualization and of visual organization (moving things around) are dependent on each other.