Scale Utilities

A checkpoint of current thinking on position scales and statistical data types.

Position scales can be described roughly by statistical data type:

• nominal and ordinal positions use labels (eg bar chart). unlike current frameworks that typically treat it like nominal/ordinal cast/coerced to a ratio scale
• interval positions use ticks that emphasize size, not position (eg map scales, but also for use in streamgraph etc)
• ratio position use ticks with absolute positions (and possibly a zero point). conventional scales

During inference/resolution, scales may start out as interval and be elevated to ratio. For example, the height of a rectangle is an interval quantity, but it can be elevated to a ratio quantity by alignment, eg. (TODO: As I'm writing this I'm wondering if the size is actually some kind of ratio thing (maybe not a ratio position yet because the position is undefined?) so maybe it is not having any sway over the conversion to an interval scale and that has to do with shape embedding...)