That's a very interesting point and suggests that, counterintuitively, it would actually be bad for (for example) vowels to look similar to each other. Another

That's a very interesting point and suggests that, counterintuitively, it would actually be bad for (for example) vowels to look similar to each other. Another way to use language statistics would be to make the font variable-width (which may help with legibility) but make commonly-occurring letters thin.