Four readings of one idea: a weighted graph. A sankey budgets it left to right, an alluvial re-divides the same quantity between categories that stay put, a chord draws it as a circle, and a network drops the weights and keeps only who is connected to whom.
These four types are read but not outlined. They sonify,
describe, braille and navigate correctly; the canvas highlight simply
clears as the cursor moves. Recovering which node the cursor is on would
mean reimplementing MAIDR's own node ordering and stage layering inside
the adapter, and a copy of that would drift silently and then outline a
confidently wrong node. See docs/amcharts.md.
One point per link, and the nodes are derived from the ends — the node list amCharts keeps separately is deliberately not read.
amCharts has no alluvial class: an alluvial is a sankey, drawn
with the nodes repeated across stages. Only the author can say which of
the two the drawing stands for, so this one carries a
userData.maidr block and nothing else changes.
A chord graph is cyclic, and MAIDR lays a flow's nodes out by their longest distance from a source. A cycle has no such layering, so every node collapses into one stage by design — navigation and sonification are correct, and the trace claims no left-to-right order it does not have.
A hierarchy series rather than a link list: the links are the tree's own
parent-child edges plus whatever cross-links each row names in
linkWith. Where the solver put each node is a fact about its
seed rather than about the data, so nothing about the layout is read.