April 20th, 2012

ISBNs for ebooks, what?

This is a second attempt at this post; it got way out of hand.

Do ebooks have ISBNs?

Sometimes. On Amazon, ebooks with page numbers may have a Page Number ISBN associated with them which indicates the paper edition of the book that the page numbers match. That's kind of cool -- nice work! In general, however, Amazon appears to use ASINs (awww, my baby has really grown up to be something so cool! Yet I don't feel bad for giving my baby such a funny name. It's _still_ A SIN that so much of Amazon's internal software relied upon 10-letter/number identifiers and any replacement for the original ISBN had to fit into that space. Thank Goddess for base 36, amirite?) for ebooks. It's less obvious what BN is using; perhaps someone else can figure it out.

It's worth pointing out that when the rest of the world went to ISBN-13, Amazon's detail page shows the ISBN-13 and the ISBN-10 associated with a book, but if you look at the ASIN (in the URL), you'll notice it's _always_ 10 letter/numbers. Not 13.

BISG is a sort of standards group for some fraction of the book industry. What do they think about this?

Well, here's a quote from BISG_Policy_1101.pdf:

"An ISBN identifies the registrant as well as the specific title, edition and format of the Book. It is mainly used within the supply chain for ordering, listing, sales tracking and stock control purposes."

Yes, Virginia, ISBN is some bizarre predecessor of useful things like PLUs or SKUs and _there's a standards group trying to make book retailers everywhere use it_. Even for ebooks. This _might_ be reasonable in, say, Canada, where you can get ISBNs for free, but in the United States, you have to buy them. From Bowker.

In any event, I think BISG is Not Happy with Amazon, and neither is this blogger:


who Nate over at The Digital Reader pointed to in The Morning Coffee for the day, and which distracted me into digging out where the assumption that BISG had anything to do with ISBNs came from, much less ISBNs for ebooks.

I would also like to point to this from the policy statement, example 1.1:

"Example 1.1
The file format in which a particular Digital Book is available is changed.
In practice: A publisher assigns a unique ISBN to a non-device specific Digital Book in EPUB format. In addition to being made available to Consumers without further modification, the EPUB is provided to, and altered significantly by, a third-party in order to create a new file format that renders on a specific device or software program.
The third-party has created a separate, device-specific Digital Book different from the publisher’s Digital Book (the EPUB); this newly created Digital Book should be assigned a unique ISBN (or proprietary identifier)."

That looks suspiciously to me like a tunnel for Amazon to drive ASINs through. "It's just a proprietary identifier."

I feel like BISG screwed up by embedding usage rights in the ISBN, but I wasn't privy to the debate so I don't know. By embedding usage rights but assuming DRM is separate, it would seem to preclude a digital receipt solution being implemented across DRM systems (but Pottermore may be proof to the contrary).

This seems obviously wrong:

"The experience and expectations of a Consumer patronizing an online retailer to buy a Digital Book or visiting an online library to borrow a Digital Book are no different than a Consumer patronizing a physical bookstore or library."

Standards documents often suck, but this seems unusually poorly thought out. Want to read it in all its dozen-ish page glory?


There was bloggy commentary when it was released:


Among others. I'm still reading the January 2011 Summary Findings document. It's really a laugh a minute.

Standards! We Must Have Them!

That was sarcasm.

Once upon a time, I wrote test code for a C/C++ compiler. This was before they let me work on the compiler itself, partially out of desperation. In any event, I'll never forget hearing my coworker repeatedly propose taking the ARM out to a gun range and using it for target practice. The coworker in question was not a gun sort of guy, either (I think he was vegan, actually, not that those are necessarily incompatible).

A little later, I worked for a company that was attempting to develop a new web browser (this was just before browsers suddenly became Free) and had the Brilliant Idea to use an SGML rendering engine and actually implement HTML the way its propaganda said it had been devised: as an instance of SGML. Ha! A quick perusal of the way HTML of the time was actually written suggested an SGML rendering engine would not actually be able to render any of it, it was so badly formed. I wound up using Lex (but not YACC, altho I thought about it) to create a tool that took Typical HTML and produced well-formed HTML that was probably pretty close to what the person had intended, and that the SGML engine could cope with.

Then the project was canceled and I went to work elsewhere, at an internet bookseller, might have heard of them. Starts with an A. There, I worked on combining information from a variety of sources to create a catalog that was the basis for detail pages and used at many points in the supply chain. Before I got there, ISBNs were being used more or less as SKUs, but there were a lot of problems with this. A bunch of small publishers abused their ISBNs (treated them as ranges and incremented the check digit rather than using it as a check digit, reused ISBNs, etc.). Also, we had items listed in a source with "fake" ISBNs that we would have liked to sell also. When Amazon wanted to branch out into additional categories of items, we needed a definitive solution BUT it had to fit within the 10 letter/character field because all of our internal software was based on that. I thought that was A SIN, but I proposed a base 36 solution to our problems, and we collectively pretended that stood for Amazon Standard Identification Number, except S.K., who figured it was arc sin.

I think it's safe to say I have a small amount of experience with standards and what's wrong with them.


The ISBN was created by a retail chain in the 1960s to help it keep track of books. It then went on to do what lots of standards do (create jobs for persnickety people whose ability to make reasonable cost/benefit judgments is congenitally impaired). When article numbers and product codes got invented (a little hazy on how that got started -- I suspect a chain in Japan somewhere, then jumped to Europe and then became international), space was carved out within one/some of them for ISBNs to live in, leading to our 13 digit ISBN world that plays better with scannable product codes than it did in its 10 digit incarnation.

With this history, consider these statements, appearing on the same page of ISBN International's 2010 study summary:

"Retailer needs and requirements are an important driver of product identification behaviour."


"US e-book stakeholders appear to be unconvinced that there is any business case for assigning ISBNs to separate e-book versions (where there is lack of consensus on whether these versions represent separate products). This is particularly true of the publishers interviewed, most of whom produce a single master file and use intermediaries to convert this to different formats. (The International ISBN Agency also notes that the vast majority of e-book business in the trade sector is through a limited number of dominant channels, each of whom has published a different set of requirements for ISBN assignment by publishers, as well as assigning their own proprietary identifiers to sell direct to consumers.)"

I'm going to supply a translation for that: the publishers ISBN International talked to didn't want to put any money into separate e-book versions since what they were doing was supplying an electronic master file (or, at times, a paper book) to Amazon and Amazon was doing the conversion. Possibly other people were also. And Amazon couldn't imagine any reason to buy an ISBN for all these ebooks when they could create their own, more compatible with their own systems identifier for free.

I'm serious when I say standards become job-creators after a certain point.