These general forms allow markup grammars, like HTML5, to be created without element counts in the thousands to address every possible need. A generalized element retains the form of greatest applicability at the expense of specifics, in other words. The HTML5 grammar, for example, solves the problem of a multitude of structural containers with only slightly differing purposes by introducing the section element.

But what help is generalized markup to a person, let alone a reading system, let alone to an assistive technology trying to use the markup to facilitate reading? Try making sense of a markup file by reading just the element names and see how far you get; a reading system isn’t going to fare any better despite a developer’s best efforts. HTML5 may now allow you to group related content in a section element, for example, but without reading the prose for clues all you know is that you’ve encountered a seemingly random group of content called section. This is structure without semantics.

You might think to make out the importance of the content by sneaking a peek ahead at the section’s heading—assuming it has one—but unless the heading contains some keyword like “part” or “chapter” you still won’t know why the section was added or how the content is important to the ebook as a whole. And cheating really isn’t fair, as making applications perform heuristic tests like looking at content can be no small challenge. This is both the power and failing of trying to process generalized markup languages and do meaningful things with what you find: you don’t have to account for a lot, but you also don’t often get a lot to work with.

Getting back to our analogy, though, it’s fair to say we’re all philosopher kings when it comes to the true nature of books; we aren’t typically interested in, and don’t typically notice, generalized forms when reading. But, whether we realize it or not, we rely on our reading systems being able to make sense of these structures to facilitate our reading, and much more so when deprived of sensory interactions with the device and content. When ebooks contain only generalized structures, reading systems are limited to presenting only the basic visual form of the book. Dumb data makes for dumb reading experiences, as reading systems cannot play the necessary role of facilitator when given little-to-nothing to work with. And that’s why not everyone can read all digital content.