It sounds like this project could benefit from using machine/deep learning/genetic solutions.
Right now from what I am reading, humans try various solutions and modify the font based on the results observed.
If one were to define various properties of the font (different features of each letter that can be set at various values), one could devise a test (either a straight up test system where people volunteer to read a given version/set of those parameters, and give a note on how well that version/set worked for them, or even a game) in which different values for those features are tested and given a notation, this data could then be used (possibly live, with adjustments/next generation testing) on the go to improve and define the next generation of parameters.
I feel like this might possibly be powerful in its ability to improve the font.
If somebody else is interested in exploring this, I'd love to give a hand and help create a proof of concept for what I'm describing here. I'd need a bit of help to figure out what parameters it would make sense to adjust/explore/optimize, but once I've been helped with that, creating the proof of concept game/test platform is something I feel I could take a crack at.
Asking this just in case somebody thinks this is interesting, sorry if this is out of the bounds of what is of interest to the project.
Sorry if this is already something you are working on, I browsed the available documentation and did not find this, but I might have missed it.
It sounds like this project could benefit from using machine/deep learning/genetic solutions.
Right now from what I am reading, humans try various solutions and modify the font based on the results observed.
If one were to define various properties of the font (different features of each letter that can be set at various values), one could devise a test (either a straight up test system where people volunteer to read a given version/set of those parameters, and give a note on how well that version/set worked for them, or even a game) in which different values for those features are tested and given a notation, this data could then be used (possibly live, with adjustments/next generation testing) on the go to improve and define the next generation of parameters.
I feel like this might possibly be powerful in its ability to improve the font.
If somebody else is interested in exploring this, I'd love to give a hand and help create a proof of concept for what I'm describing here. I'd need a bit of help to figure out what parameters it would make sense to adjust/explore/optimize, but once I've been helped with that, creating the proof of concept game/test platform is something I feel I could take a crack at.
Asking this just in case somebody thinks this is interesting, sorry if this is out of the bounds of what is of interest to the project. Sorry if this is already something you are working on, I browsed the available documentation and did not find this, but I might have missed it.
Cheers!