@smach This article makes me queasy.
---
Summary:
First step: replace the actual data with an approximation.
Remaining steps: remove or modify the contextual information that gives the data meaning and add a title that states the conclusion you want people to draw from it.
Result: a picture designed to support a specific story masquerading as a graph.
---
This may improve communication of the underlying information relevant to a specific story, but it's not 'improving a line chart'. It's 'producing an illustration of a graph'.
A graph is a compact reversible representation of data, like the encoded audio data on a CD. Step one removes a reader's ability to read the original data back out or identify values associated with important elements like peaks, valleys, and line crossings.
And "compelling" as a metric feels wrong. I aim for "truthful" and "clear", not "believe what I'm telling you this graph says".
Queasy.