Bad ePub? Blame Automation!

By David Alan Rech of Scribe Inc.


In the posting of December 16, 2010, we discussed reliance on the PDF as a source for e-book development. We recommended using the typesetter (or mechanics) files as the source for your e-book development. But mere reliance on automation to produce ePub files can still result in bad ePubs!

Many naively believe that the ePub generator from InDesign will automatically create e-book files. The promise is that you can go from InDesign directly into ePub format, not to mention a variety of XML formats and XHTML. There is no doubt the Export for Digital Editions plug-in works, but using the plug-in does not solve your ePub problems.

Reliance on automation to produce ePubs results in books that do not function in every device, are internally inconsistent, or vary from your other publications. Generating ePubs automatically can also result in missing content. And since there are no real standards for ePub files, an ePub read within a Sony device will behave differently from one read using an iPad or Nook. The Adobe plug-in is built to create valid ePub files that will work using Adobe Digital Editions, and if you want to restrict your customers to a single piece of software that is fine; but the files may not behave correctly in other readers. Certain elements for which there are ePub specifications (e.g., tables) are not supported within some of the readers. Plus, many characters will not display in all ePub readers. Because it is imperative to make adjustments to the ePub output from InDesign, relying solely on automation is not feasible.

Another issue involves the use of InDesign. The plug-in is designed to read styles and to convert them into a usable markup schema (e.g., HTML for ePub). Like any computer program, the plug-in uses an algorithmic function to convert, or parse, the style information. To get correct results consistently, the publication must be developed in a consistent and correct fashion. If you want all of your a-heads automatically to be treated the same in your ePub, then you would need to treat them in a consistent fashion within InDesign. This means properly and consistently using styles and avoiding ad hoc formatting. Even if the plug-in were to work universally, no computer algorithm can account for inconsistent human intervention.

In addition, the plug-in does not have a consistent way of handling elements that fall outside of the normal flow of your publication. Thus elements are often presented inconsistently or in ways that break up the flow of other elements. In some cases, they are excluded entirely. To have the correct flow of elements, elements within books must be properly created and linked. Because the plug-in relies on the visual relationships among elements on the page, even the so-called proper methodology will derive poor results.

Some elements do not function well, or at all, within ePub format. Tables, many foreign characters, numerous symbols, indentation, and spacing do not display in some readers, or function poorly when used in readers where they are supported. Since the plug-in has not been designed to deal with these issues, it blindly converts tables to HTML tables and recognized characters into an entity code. Other characters are either substituted with a character occupying the same ASCII position or dropped altogether. The result is valid but nonfunctional. In previous blog postings, we have suggested methods to plan for this in the editorial and typesetting phases of book development. For the foreseeable future, there is no automated way to handle these issues effectively.

E-book automation has a long way to go. It is, and will be, incumbent upon publishers to produce consistent files that can be interpreted algorithmically, should e-book automation ever be perfected. This, after all, is foundational to XML. No matter what, consistency is required in the editorial and production process. We must plan well and work in a consistent fashion to produce print and electronic versions of our books. Automation cannot undo poor work.

Experiencing bad ePubs? Perhaps automation, or reliance upon it, is to blame.