.notyetdef (2019)

This collaboration with Fred Benenson trains a generative adversarial network on Unicode-enabled font indexes to create hybrid glyphs from extant writing systems. These new glyphs serve as a vantage from which we are able to survey font internationalization and global script representation in computing. The Unicode standard unifies the world's writing systems with a peculiar shortcut: while it maintains address locations, or "code points," for some 137,993 characters, it does not specify what those characters should look like when a user's computer displays them onscreen. More, there is no way to download "all of Unicode": no usable archive of its characters exists. In response, we have pieced together our own archive of font indexes and generated new glyphs therefrom, using these hybrids to reflect on the standard's still-outstanding absences and the thousands of code points it has yet to fill. Our future plans include investigating how machine learning methods can indicate the influence of typographic font conventions in digital renditions of script systems like Bengali, Hanzi, and so on.

A version of ".notyetdef" is currently hosted at a2p.