GraphScape: Modeling Similarity and Sequence Among Charts (2017)

Associated paper (and talk video): https://dl.acm.org/doi/10.1145/3025453.3025866

Associated code: https://github.com/uwdata/graphscape

I get a real "Zen and the art of motorcycle maintenance" vibe out of this.

It's a challenge to quantify the 'Quality' of a chart, but trying to measure it by the production steps seems to completely miss the point. Feels to me like deciding that a painter can be copied by using their brushes, pigments, and studio. Yeah, those are a part of the process. But if you focus on them, you'll find you've completely missed the process.

Not that I'm going to try and design a Tufte evaluator for finished products, which is the sort of thing I guess I'm advocating. :P That's harder still.