A technical document few readers will ever see is generating an unusual volume of argument across the book trade. The draft specification for EPUB4 — the first major revision of the publishing industry’s standard ebook format in over a decade — has circulated in working-group form for months, and with each revision the debate sharpens. At peace on its format questions for years, the trade has largely forgotten how contentious its own plumbing can be. EPUB4 is a reminder.
The headline changes are real. EPUB4 proposes to replace the format’s venerable XML backbone with a profile of HTML, aligning ebooks with the web platform that has meanwhile swallowed the rest of digital publishing. It strengthens accessibility affordances, formalising structures that screen readers and reading systems have had to infer. It modernises multimedia handling, acknowledging that ebooks increasingly contain audio and interactivity whether the traditionalists approve or not. And it tightens the relationship between the format and the distribution metadata that surrounds it, recognising that a file that cannot describe itself correctly to a storefront is a file that will not sell.
Each of these ambitions has a constituency and a counter-constituency. Large trade publishers, managing backlists in the hundreds of thousands, see the HTML alignment as a path out of the conversion treadmill: build once for web and book, extract ebook as a view rather than a translation. Developers who have spent a decade working around EPUB3’s eccentricities greet the cleanup with open arms. But the very thing that charms technologists alarms the accessibility community, which notes — with the authority of having dragged the industry toward WCAG compliance largely against its will — that a specification moving fast toward a moving target (the living web platform) risks embedding fragility where stability matters most. An ebook, they point out, is one of the few digital artefacts a reader can reasonably expect to open, legibly, in thirty years. Web pages make no such promise.
Reading-system vendors, whose apps and devices must implement whatever emerges, oscillate between welcoming the simplification and dreading the migration cost. A format change at scale is not a software update; it is an act of industrial archaeology across backlists nobody has budgeted to retouch. Retailers have been characteristically quiet, which the veterans of previous format transitions recognise as the calm of parties deciding whose problem this will be.
Beneath the technicalities runs a commercial question the working group cannot settle by engineering. EPUB3’s failure to fully displace EPUB2, more than a decade after its own triumphant draft, demonstrated what the trade prefers to forget: formats in this industry do not succeed by superiority but by adoption, and adoption follows the path of least friction across the widest installed base. The publishers pushing hardest for EPUB4’s modern features are those whose lists are digital-forward — interactive children’s books, enhanced audio-text hybrids, educational content. The publishers with the deepest backlists, and thus the most conversion exposure, have been measured in their enthusiasm, which in standards politics reads as a veto threat.
The realistic prognosis, say those inside the process, is a profile-based outcome: a core specification strict enough to guarantee longevity, with optional modules for the enhanced features, and — the crucial concession — a commitment to backward-compatible rendering that lets the existing corpus age in place rather than demanding resurrection. Even that compromise carries a cost: complexity deferred is complexity compounded, and a format with optional modules behaves, in practice, like several formats wearing one name.
Why does any of this matter to the reader, the bookseller, the working novelist? Because the invisible decisions are the load-bearing ones. The last format war gave the trade a decade of distraction and a generation of orphaned files. EPUB4’s draft is the industry’s chance to demonstrate that it has learned the difference between a standard and a preference — to build plumbing that outlasts the companies that argue about it. The debate will run through next year and beyond. The deadline that matters is not the specification’s publication date. It is the day the first major backlist silently fails to open somewhere it used to.