Last week, André Klein — an independent author who has been self-publishing since before “indie author” was a category — hit publish on a new book. He ran it through epubcheck, the open-source validator that serves as the industry’s pedantic gold standard. It passed. Amazon Kindle accepted it. Apple Books accepted it. Thorium, the free reader, displayed it without a twitch. Then he uploaded it to Kobo.
Kobo rejected it as “corrupted.”
No explanation. No specific error. Just a red flag on a file that every other system on earth had deemed perfectly compliant. The blog post Klein wrote about the experience — titled, with exhausted precision, “Your EPUB Is Fine. Kobo Disagrees. Blame Adobe” — has been climbing Hacker News all week, and the reason isn’t hard to locate. It’s not just an author’s technical gripe. It’s a preview of what happens when a proprietary validation layer sits between you and the files you thought you owned.
The EPUB format is supposed to be the open alternative to Amazon’s walled-garden ecosystem. It’s the format you buy when you want to read on a Kobo, or a Boox, or any device that isn’t shackled to a single retailer’s cloud. It’s the format librarians recommend to patrons who ask how to avoid giving Jeff Bezos a permanent seat at their nightstand. The pitch is straightforward: buy the file, own it, move it wherever you want. EPUB is the format of digital sovereignty.
Except that sovereignty turns out to be conditional on Adobe’s approval.
The Middleware That Eats the Open Standard
The culprit, as Klein traced it, is Adobe’s Digital Rights Management layer — the DRM that Kobo uses to wrap EPUB files before they land on a customer’s device. Somewhere in that wrapping process, Adobe’s tooling flags perfectly valid EPUBs as corrupt. The validator that matters isn’t the open-source one the community maintains. It’s Adobe’s, and Adobe’s is a black box.
This is not a bug. This is what it looks like when an open standard gets colonized by middleware. The EPUB spec may be public, but the actual pipeline from author to reader runs through proprietary tooling that gets to override the spec whenever it feels like it. Your file is fine, but the gatekeeper disagrees, and the gatekeeper’s disagreement is final.
The practical consequence is that an author who wants to sell on Kobo — which, alongside Kindle, represents one of the two viable ebook storefronts — now has to debug against a validator they cannot see, for errors that do not exist, to satisfy a company that has no incentive to tell them what went wrong. Klein’s workaround, after weeks of trial and error, involved rebuilding his book file from scratch and hoping Adobe’s impenetrable validation pipeline would shrug its shoulders and let it through. It did. He still doesn’t know why.
The Customer Is Not the User
Here is the part the conventional critique of digital rights management tends to miss. The standard complaint is that DRM treats the customer like a thief — it punishes honest buyers with restrictions that pirates easily bypass. That complaint is true, but it’s also been true for twenty years and hasn’t changed anything. The more interesting problem is upstream.
Adobe’s validation layer does not exist to protect Kobo’s customers from corrupt files. It exists because the publishing supply chain has a structural problem: nobody in it trusts anybody else. Publishers want assurance that their files won’t be pirated. Retailers want assurance that the files they sell won’t generate support tickets. The platform that provides those assurances extracts rent from both sides. The fact that the rent extraction occasionally manifests as a false corruption error on a perfectly valid book is not a failure of the system. It is the system working as designed — just not for the author, and just not for the reader.
“The validator flagged something in the metadata layer that epubcheck doesn’t even look for,” a developer who maintains ebook conversion tools told me in a Slack DM on Monday. “It’s not that Kobo is incompetent. It’s that Adobe’s stack is a black box and Kobo doesn’t control it. Nobody controls it. That’s the product.”
The product is opacity. And opacity, in a market with only two storefronts that matter, is leverage.
What the Open-Source Validator Cannot Fix
There is a temptation, reading Klein’s post, to treat this as a debugging story with a happy ending. Author encounters obscure error, author reverse-engineers a workaround, author publishes, everyone moves on. The comments on Hacker News are full of suggestions: better logging, a community-maintained compatibility list, pressure on Kobo to document its validation rules.
These suggestions are reasonable and they will not work, because the problem is not technical. The problem is that Kobo’s business does not depend on being the best steward of open ebook standards. It depends on being the not-Amazon option that still manages to sell enough books to keep the lights on. If Adobe’s DRM stack occasionally eats a compliant file, that is Adobe’s problem to fix — except Adobe has no incentive to fix it, because the authors affected are not Adobe’s customers. They are Kobo’s suppliers, and suppliers in a two-player market do not have leverage.
Klein ends his post with a line that reads as resignation but lands as indictment: “And until that changes, which it won’t, if you want to make sure that your books are available on all platforms, you’ll have to keep Adobe around.”
He is right that it won’t change. The open standard lost the moment the supply chain routed through proprietary validation. What’s left is the appearance of choice — the Kobo in your hand, the EPUB file on your drive — and the quiet reality that neither one works without Adobe’s permission. Your file is fine. That is exactly the problem.