Screen Shot 2011 12 09 at 9 37 23 AM

This is an important matter that has created quite a bit of controversy in the industry.  From the press release:

The Book Industry Study Group (BISG) announced today the publication of a new Policy Statement detailing best practices for assigning ISBNs to digital products. Developed over the past 18 months within BISG’s Identification Committee, BISG Policy Statement POL-1101 addresses the critical need to reduce product identification confusion in the market place in order to provide the best possible consumer-level purchasing experience. 

BISG encourages all member companies and other industry stakeholders to download the Policy Statement online at and work toward adopting the suggested guidelines as soon as practical, with a target for new product introductions of no later than March 2012. The best practices are applicable to content intended for distribution to the general public in North America, but could be applied elsewhere as well.

The Policy Statement has been endorsed by BookNet Canada, a not-for-profit agency dedicated to innovation in the Canadian book supply chain, the National Information Standards Organization (NISO), where content publishers, libraries, and software developers turn for information industry standards that allow them to work together, and IBPA, the Independent Book Publishers Association.

The following excerpt starts on page 6 of the 12-page Policy Statement: 

“Separate ISBNs should be assigned to all unique Digital Books for ordering, listing, delivery and sales tracking purposes. In general, there are three major factors that determine the need to assign unique ISBNs to Digital Books. 

— Content
If two digital books are created, one an exact textual reproduction of a Physical Book and the other an enhanced version that includes video, audio, etc., then the two Digital Books are unique and different products, and each requires a unique ISBN.

— Format
If an EPUB format, a PDF format and a Mobi format (among others) are created, each format should be assigned a unique ISBN. This is similar to creating a hardcover and paperback edition of a Physical Book and should follow the same rules regarding ISBN assignment.

(Note: When the application of DRM software is part of the transaction with the Consumer (as frequently happens in the US) it does not constitute the creation of a new format as the term is being used in this Policy Statement. In this case, DRM is not a format: it is a wrapper around a product. An EPUB file with DRM software applied is still an EPUB file, a PDF file with DRM applied is still a PDF file. In this case, DRM is not part of the product, it is part of the transaction. An ISBN is a product identifier, not a transaction identifier.) 

— Usage Rights
If a Digital Book is made available with different usage rights in different markets (e.g. adjusting the usage settings so that printing is allowed in the version going to the education market, but not in the version going to the retail market), each version should be assigned a unique ISBN.

(Note: As described in the note under ‘Format’ above, usage rights specifically applied to a publisher’s Digital Book using DRM software, such as Adobe Digital Editions or Apple FairPlay, as part of the transaction between the vendor and the Consumer does not fall into the above category and does not require the assignment of a unique ISBN.)”


  1. A few years back, when this issue was first being debated, I had discussions with those involved in this dispute, including representatives at Bowker, which sells ISBNs in the U.S. That proved quite informative.

    The problem with BISG’s policy statement, and the reason it stirs up controversy, is that it’s imposing a centralized, 1970s era view of book databases on a publishing environment that’s grown much more complex, rapidly changing and fluid. We no longer live in a world where books are categorized as hardback/paperback and into editions issued a year or more apart. At the very time when we need a better system of tracking books and particularly digital books, we’re being asked to continue to use the a system developed some 40 years ago when IBM mainframes reigned supreme. That makes no sense.

    PROBLEM 1. Someone at Bowker explained this to me, although here I provide my interpretation. With 70s-era ISBNs, to discover the ‘meaning’ behind a book’s identification number, booksellers and book purchasers must go to a central data source such as Bowker. There’s nothing about an ISBN that defines a book’s content, its digital format, or its DRM. For information about that, you must consult Bowker or sources that get their data from it. And because of that, as noted above, when almost any of those features change, an entirely new and unrelated ISBN and is needed. Oddly enough, the one situation where a new ISBN isn’t needed is with DRM, perhaps the one factor about an ebook that matters most to a consumer and many retailers.

    PROBLEM 2. The data we get from Bowker is doubtful at best. Bowker isn’t the publisher, so It knows nothing about an ebook. The quality of that data it displays depends on the willingness of publishers to adopt all these complex rules, to buy numerous ISBNs at rapidly rising prices, to assign new ISBNs to editions that differ only in the suffix to the file that exits from automated processors like that at Smashwords, and finally to notify Bowker’s clumsy data entry system, in detail, again and again, precisely what what each of those particular ISBNs means. Any failure to do that renders these ISBNs worthless for anything but pricing data. Large publishers can perhaps assign staff to do nothing but manage that data. Smaller publishers will simply muddle through, perhaps learning to care little about a scheme they never liked anyway. ISBNs will become so useless, they’ll die the death of a thousand cuts.

    It’s not hard to envision a 21st century system, distributed rather than top-down, that’d be far better. Like ISBNs, there’d be a publisher code followed by what you might call a basic text identifier that’d be the same for basic and enhanced versions and all editions. That’s a digital equivalent of a book’s title. That’d be followed by additional codes that identify editions out to at least three digits to take into account how easy it now is to create a new edition. Other descriptor fields could identify the sort of enhancements (i.e. video and audio), the digital format, the version of that format, and the DRM used.

    The advantage of this system is obvious. Given the digital equivalent of the title, the only thing a company such as Bowker would issue, virtually everything that publishers, retailers, and buyers need to say or know is contained in those well-defined descriptive digits. Software that creates these books, retail systems that process them, and reader apps for finding books, can parse those meanings automatically. No tedious klutzing with otherwise meaningless ISBNs required.

    It’d make life far easier for publishers. Bringing out a new edition, would only mean incrementing the edition field by one. Adding enhancements would change another field. Adding an EPUB 4 version, when such exists, would only require incrementing that field. Easy, easy, easy. And cheap too.

    Sellers would also love the new scheme since it’d make product tracking easy as pie. Displaying all the various versions of a title would be simple, since they all have the same text identifier. Parsing out what each version means wouldn’t require any hand labor or consulting of messy and perhaps inaccurate Bowker databases. It’d only require parsing certain, unambiguous fields.

    Savvy book buyers and readers would also love it because purchasing mistakes would be rare. The numbers would tell them precisely what they’re buying, i.e. this is the latest edition with audio enhancements but no video, in ePub 3 with no DRM.

    Personally, the advantages of the latter system are so obvious, the only reason I can see that such a system hasn’t been developed and adopted is that Bowker, a company that seems mired in the 1970s, refuses to develop them and groups like BISG simply lack the vision or the gumption to precede on their own. That’s sad.

    And in the end, I think something much like what I have described above will be adopted. Its advantages are simply too great. Publishers and consumers will end up paying twice: once for implementing a system that’s doomed to fail and once for converting to something much better.

The TeleRead community values your civil and thoughtful comments. We use a cache, so expect a delay. Problems? E-mail