The machine translates words of a text into patent drawings. Seven million patents — linked by over 22 million references — form the vocabulary. By using references to earlier patents, it is possible to find paths between arbitrary patents. They form a kind of subtext.
Over time a serendipitidious narrative is created.
I think it's beautiful! Reminds me of those old-school plotters that still used pens, before the days of laser plotters and inkjet.
Click here for more of the project.
