Linear Layout for the Visualization of Networks [bcgsc.ca] promises to eradicate the typical occurrence of "network hairballs": the too densely populated network graphs that are impossible to navigate, let alone understand. Such conventional network visualizations are subordinate to the layout, as the node and edge positions plus the lengths depend on the layout algorithm, as much as on the data. The effect of such layout rules is difficult to predict, making direct comparisons of the resulting representations almost impossible.
Developed by Martin Krzywinski of the Cancer Research Center in Vancouver, Canada, the linear network layout technique constrains the nodes to linear axes, while edges are drawn as curves between these nodes. The position of a node is determined solely by network structure, node, edge annotation, or any other meaningful properties of the network. In other words, layout rules are defined by the designer, based on what properties that are meaningful and important. As a result, if the layout shows a pattern, one can be sure it is due to structure in the underlying data and not on the layout algorithm's interpretation of how the data should best be shown.
Even more, lineair networks can be used to compose stacked bar plots, ideally suitable for comparing multiple ratios. Here, edges are drawn as ribbons, with different edge lengths / nodes become data values.
As in Martin's own words: "Imagine a world where something this pretty is useful".