EPUB has become so much a part of the global publishing ecosystem that one could be forgiven for taking it for granted, not realizing that its specifications, which were first issued by the IDPF in 2007 and updated since 2017 by the Worldwide Web Consortium (the W3C), had not, in fact, been formal W3C standards.
Until now. On May 25, the W3C—the global standards body for such fundamental web technologies as XML, HTML, CSS, and scores of other standards that power today’s digital world—published EPUB 3.3 as a formal W3C Recommendation. And in what I’d characterize as deft sleight of hand, EPUB 3.3 manages to be both dramatically different from its predecessors while being “backward compatible,” ensuring that all of the millions of EPUB 3.2s already published are still valid.
Three specs in one
The most striking difference between versions is immediately apparent. EPUB 3 is now not one specification, like all of the previous versions, but three.
• EPUB 3.3 specifies how all EPUB 3 documents are structured and encoded. It’s referred to as the “authoring” specification: the spec that creators of EPUBs should follow.
• EPUB Reading Systems 3.3 specifies how the systems and software that deliver and render EPUBs should interpret the spec.
• EPUB Accessibility 1.1 specifies the tagging and metadata that will ensure than an EPUB 3 is fully accessible.
This is a very useful restructuring. The previous EPUB specs didn’t clearly distinguish between these three important contexts. They mostly focused on authoring, with only a little information about accessibility and almost none on how reading systems should behave. The new EPUB 3.3 spec was significantly rewritten to make things much clearer.
Testing, testing, testing
Another really important aspect of the development of the new spec is that every feature of it was thoroughly tested. That’s one of the requirements of a W3C Recommendation: there must be at least two independent implementations of every feature of the spec or the feature has to be removed.
This accomplishes two fundamental things: obviously, that the given feature actually works in the real world—but not incidentally, that there were at least two independent organizations that thought each feature was worth the trouble of implementing.
This is not trivial. One of the weaknesses of some of the earlier versions of EPUB was that they included features that nobody ever implemented.
To become a recommendation, the spec also needs to pass the W3C’s “horizontal review,” which affirms that it conforms to the requirements of accessibility, internationalization, security, privacy, and alignment with the fundamental web architecture. This ensures that the spec can work for everybody, regardless of ability; that it can be used in languages throughout the world; and that it works and plays well with web technologies in general. All this is true of the new EPUB 3.3.
Meeting E.U. standards
Key European members of the W3C’s EPUB Working Group conferred with the European Commission as the European Accessibility Act was being developed. Now finalized, the EAA requires, among many other things, that a book cannot be sold in the E.U. as of 2025 if it is not available in an accessible format. But it does not provide much specific guidance as to what will constitute sufficient accessibility.
One way to know: the W3C has confirmed that a book available in an EPUB that conforms to EPUB 3.3 (and specifically the EPUB Accessibility 1.1 Recommendation) will in fact be considered accessible—and thus saleable in the E.U.
The ecosystem is all set
End users often hold back implementing a new specification, waiting to make sure downstream technologies that are going to use it have caught up. The long, deliberate, and very public development process for EPUB 3.3 has ensured that the ecosystem is ready for it on day one. Almost all of the many thousands of producers and disseminators of EPUBs, from publishers and their pre-press and data conversion suppliers to the aggregators and retailers in the e-book supply chain, are dependent on EPUBCheck, the software that automatically checks an EPUB for conformance to the spec. Most retailers will simply reject any EPUB that doesn’t pass EPUBCheck.
Thanks to the test suite created for EPUB 3.3, the current version of EPUBCheck already conforms to it, and most of the supply chain has already implemented it. The same is true of Ace by DAISY, the accessibility checker that most producers of EPUBs use.
We’re good to go, folks! EPUB 3.3 should be a firm, stable, and enormously useful standard for a long time to come.
Bill Kasdorf is principal at Kasdorf & Associates LLC, a publishing consultancy. He is a founding partner of Publishing Technology Partners.