Is it a tough problem, though? Fonts have hinting built in, executing those bytecode instructions should be something even low-powered chips should do just fine. Text flow algorithms have been solved well since Knuth's TeX work in the 1970s, as I understand it.
Obviously, publishers also need to make a better efforts in producing ebooks, but I'm not super familiar with EPUB and don't know where the technology is lacking.
- typeset the document, look through each paragraph for bad breaks/poor spacing, and if concerned about a 2-up option, check the page breaks to make sure pages are even, and adjust as necessary
The problem is it's _hard_ to detect bad breaks, and problems such as "stacks"
the instance of the same word appearing at the left or right edge of a paragraph,
the paragraph shown here has a four word stack forced to occur at the left of
the text as currently written when shown at a reasonable width on a display
are very, very hard to address while maintaining even spacing, esp. if the text
Obviously, publishers also need to make a better efforts in producing ebooks, but I'm not super familiar with EPUB and don't know where the technology is lacking.