Metrics for Reasoning About the Usability of Visualization Notations

It's a good question! I think I'd look at this through the lens of the Cognitive Dimensions of Notations, specifically the dimensions of 'consistency' and 'viscosity' (i.e. is the same thing literally called `title` in one place in the API and `name` in another? how hard is it to change from using one component to another?) and/or 'closeness of mapping' (i.e. do domain experts call it `title` but it's `name` in the API?).

Our metrics would maybe capture some aspects of consistency/viscosity, because `title` and `name` would both count towards vocabulary size, and because the distance between specs would be higher if `title` routinely needs to be changed to `name` when iterating, but that's about it. I don't believe a metrics-based approach like ours can quantify much if any of 'closeness of mapping' because that mostly happens in people's heads, not in the source code itself :)