A few weeks ago my work team had a question we weren't sure how to answer. We publish open textbooks as part of the library's mission to provide access to high quality learning resources, and they are always assigned both an International Standard Book Number (ISBN) and a Digital Object Identifier (DOI). Every book we publish is different and our next book — to be released early next year — threw up a new challenge. We will simultaneously be publishing two different versions: one in English, and the other in a different language. Having never done this before, the question was: should each book be assigned its own DOI, or since they are the same work, should they share a DOI?
When I looked for guidance on this, I found the answer fairly quickly, but it led to something a little more interesting. So here we go:
The ISBN rules are very clear that all variations must be assigned unique identifiers:
You will need one ISBN for each edition and format of your book. The ISBN identifies the publisher as well as the specific book title, edition, and format (hardback, paperback, ePub, PDF, Mobi, Audio…). Any variation of a book would require the use of a unique ISBN to identify it properly.
So each of our versions definitely needs to have a individual ISBN. But what about DOIs? Well, the DOI system gives us flexibility to assign at any level:
- A DOI name can be used to identify physical, digital and abstract entities.
- An entity can be identified at any arbitrary level of granularity.
It turns out the rules - such as they are - for DOIs are a fair bit more flexible. You can assign a DOI to any "object". It could be a work, a particular edition of a work, or even a single chapter. So the answer to our original question is "you can choose either". Given the two edititions will have different ISBNs and are not identical, we decided to assign a separate DOI to each as well.
While I was investigating this, however, I found something else. It turns out that it's possible to combine an ISBN and a DOI, into something called ISBN-A! At least, it's possible in theory. The ISBN-A is essentially an ISBN formatted into a DOI URL structure, registered both as an ISBN and a DOI. It's a cool concept, particularly useful for ebooks. The kicker, however, is that in order to mint an ISBN-A, you have to be a recognised ISBN registry. This kind of makes sense, because otherwise anyone and everyone could possibly create competing ISBN-As.
However, ISBN registries do not operate in a competitive market environment. In Australia Thorpe-Bowker hold the monopoly on ISBNs. If you want one, you have to get it from Thorpe-Bowker, and nobody else is allowed to sell you an Australian ISBN. Pretty good business if you can get it — simply assigning algorithmically-derived numbers to people for a hefty fee. Why bother innovating if you can just print money uh, I mean barcodes? Anyway, Thorpe-Bowker don't assign ISBN-As, and don't appera to have a heap of incentives to do so any time soon. This means that if you operate in Australia and want to use an ISBN-A, you simply ...can't.
So an obviously good tool — that has been agreed to and designed by the International DOI Foundation and ISBN International — can't actually be implemented due to yet another lazy monopoly. This is why we can't have nice things.