Tuesday, August 04, 2009

The ISBN Is Dead

There are few greater supporters of the ISBN standard than I (and most of us are named "Michael" so we are easily identified); however, I am increasingly concerned about the future health of the ISBN. In its current form the ISBN is not yet dead but therein lies the problem: 'in its current form.' In order to gain entry to the supply chain, most small and medium-sized publishers will continue to buy their ISBNs from agencies around the world as they have since the 1970's. (In contrast, most large publishers have reservoirs of ISBNs sufficient to last almost forever and only occasionally buy new prefixes to establish new imprints).

Five years ago, I participated in the once-a-decade ISO ISBN revision process that resulted in the current ISBN standard. (Michael Healy ran this two year process on behalf of ISO). That revision included the expansion from 10 to 13 digits, but this was tame compared to the contentious issue of separate ISBNs for every eBook format. I support this position (although I did not have a vote in the revision) and agreed with others who viewed assigning separate ISBNs as consistent with the way ISBNs had historically been assigned to other title formats. Despite the passage of time, this issue continues to generate significant comment and has become (to me) one of several indications that the ISBN in its current form may not be sufficient to support the migration to a digital world.

A second problem the ISBN faces is driven by some down-stream suppliers who don't see the ISBN as relevant. The most prominent (egregious - pick your label) of these has been Amazon - and this is not just because no Kindle title carries an ISBN. Amazon has long been disdainful of the ISBN and, almost from the opening of the bookstore, they assigned "ASINs" to books. In his defining Web 2.0 article, Tim O'Reilly used the example of Amazon's ASIN as an indicator of Amazon's application of the principles of Web 2.0. At the time (while I was at Bowker in 2005), I took a more sanguine view in an email:
Amazon’s ASIN creation was built out of expediency. If they received a title from a publisher that (for whatever reason) had no ISBN, they assigned a number just so they could get it in their system. (Don’t laugh, we get frantic calls from publishers who are at their printer and don’t have a number). At first they were designating these as “ISBN”s which we had them change. There was never an intention to take ISBN and make something better and different. So while I would agree on your point about extending the bibliographic content, in the case of ASINs Amazon were not looking to create additional value or take the identifier to some other more valuable place: they needed 10 digits to identify a SKU. Now they have polluted the supply chain with these numbers. No other vendor has seen a requirement to create their own SKUs; there has never been a need, because the ISBN has been the most effective product identifier ever established.
Hence, at Amazon, the lack of ISBNs on Kindle titles isn't really new; although it was a fairly rare occurrence (albeit from a very large player). Others now new to the supply chain (including suppliers of print-on-demand titles) have decided not to use ISBNs. Some of these suppliers are using the Google Book settlement titles as their 'inventory' and thus, by definition, this issue becomes a significant challenge to the ubiquity of the ISBN.

A third issue concerns the rapid influx of new titles as a result of digitization programs. At this point, it's unknown whether any of these titles will be subsequently broken down into parts, (although this seems inevitable,) but that further compounds the issue of how ISBNs - or other identifiers - will identify this content.

Some may argue that, as the supply chain compacts the connection between producer and supplier becomes tighter and a specific item identifier isn't required. Maybe that's true; however, I believe it's far too early in the transition to digital content to make this judgment. Unfortunately, if we shrug our collective shoulders to these issues, this non-action will set a precedent from which we as a publishing industry will be unable to recover.

The ISBN standard united the industry from author royalty statement to store shelf and, while I emphasize the ISBN is far from dead, there are sufficient warning signs to suggest that the ISBN may be unable to thrive in the 21st century as it has over the past 40 yrs. As a community, we need to recognize that the ISBN may not be meeting its intended market need and that the future may make this deficiency even more stark. From an international perspective, ISO could help by reconvening a partial (or full) revision of the standard; it seems incompatible with the speed at which all industry changes that we can continue to live with a 10 year revision cycle. In my view, ISBN could benefit from an accelerated revision cycle while the result of non-action could be increasing irrelevance.

Into this mix I would also add that ISBN can no longer stand generally independent of other identifiers, such as a work ID or party ID. For example, while assigning ISBNs to pre-1970 titles may make an ISBN agency's revenues bulge, it may not be the most effective proposal for the supply chain. A more appropriate approach may be a combination of work ID, party ID and ISBN and, for this, we require a cohesive methodology and possibly a 'merging' of these standards in a more formal way.

This commentary naturally leads into a discussion of the construction of bibliographic databases, which I hope to present in the future.


Inkling said...

The ISBN was never intended to work with ebooks, so it's hardly surprising that it creates problems, particularly when a standards body insists on something as messy as a separate ISBN for each ebook format. Just the cost of doing that dooms the idea, but it also makes no sense. Ebooks, like serials, need their own numbering scheme, one designed with a bit of thought to what is unique about ebooks. Off the top of my head, I'd suggest these features.

A string of numbers that describe the country of origin. Copyright limitations and marketing agreements may seem out of place in the Internet age, but they still exist. There could be a number for a global edition with no copyright, sales or distribution restrictions.

A string of numbers (as with an ISBN) that defines the publisher.

Another string that defines the basic text, i.e. Mark Twain's Tom Sawyer.

*** These first three approximate what an ISBN currently does. But the ebook standard would go further.

An additional string that the publisher can use for editions, revisions, or whatever. Customers would benefit by knowing that seeing a #122 here meant it was was more recent that #109. Helpful when OCRed text goes through correction cycles.

A string that defines whether there's any DRM and, if so, what the restrictions are. This is something that matters a lot to customers. It should be clear and unambiguous.

And finally, a string that defines the ebook format.

Note that publishers would only have to apply for the first three sets of numbers: country, publisher and text as with ISBNs. They could create the last sets of numbers by simply following the standards. They would not need to go, cap in hand, to some agency to get 'permission' for an ePub edition of a book previously released in Mobibook.

At the other end of the market, for potential customers, software could be available to extract all the meaning in the numbers. Given a long string, it could tell a potential customer that this is the Mississippi River Press's 13th (2011) edition of Mark Twain's Tom Sawyer in ePub version 2.1 with no DRM.

Standards would also settle, one and for all, just what a particular DRM means. No uncertainty, as with Kindle books, as to how many devices an ebook could be moved to. If Amazon sells a book with a certain DRM, they'd have to play by its rules.

That's the rough idea of a scheme for identifying ebooks that would be a benefit rather than a burden. I'm sure others can improve upon it.

--Michael W. Perry, Inkling Books, Seattle

Nadine Laman said...

Oh don't be silly. ISBNs work just fine and they should be assigned to ebooks. It is NOT cost prohibitive to the publisher. There is no reason not to assign an ISBN to ebooks.

Use the same ISNB across the board for each format of the ebook, regardless of which ebook reader it goes on, eventually every ebook will be readable on any device once there is a standardized format. Remember Beta and VHS?

Not assigning an ISBN to an ebook is like saying it isn't a real book, it doesn't count. Just makes it less alarming that there are pirating issues, if it isn't a real book.

Nadine Laman, Glendale, AZ - USA

Glyn Pope said...

I published a book through lulu. Wasted money on an isbn. Sold more books that didn't have an isbn through another publisher turnermaxwell.
I think there is some snobbery to isbn's. Authors think that they aren't proper writers without one. When you see the amount of crap published that does have an isbn my point is proved!
see http://glynpope.blogspot.com/

Nadine Laman said...


True, an ISBN does not guarantee quality of content. It does make it easier to digitally order, distribute, and track in markets outside of Amazon. But for a self-constructed network, it isn't NEEDED. It is simply an identifier like an auto tag/licence plate.

Nadine Laman, Glendale, AZ - USA

Anonymous said...

The ISBN weakness - specially in the UK and US - is cost. You cannot both sell at commercially inflated prices and expect it to remain an industry standard.

The challenges to the existing system come from new entrants who often are publishing on a shoestring.

It is common now to publish in a dozen ebook formats. Where ISBNs are still free, as in India, resistance to their use is inevitably not strong.

How much do Nielsen and Bowker charge - and why?

Joseph Harris

Glyn Pope said...

'Where ISBNs are still free, as in India, resistance to their use is inevitably not strong.

How much do Nielsen and Bowker charge - and why?' said Nadine Laman

Yes - why aren't ISBNs free in the UK and US?

Todd said...

There are a long list of issues here that need to be discussed in detail. I'll post something on www.niso.org/blog later today.

In short:
1) It is part of the existing standard that ISBN numbers should be assigned to each different e-book version

2) There are sufficient ISBN-13 numbers to last.

3) There are links between ISBN, the ISTC (Text Code), the ISPI (Party Identifier), and the DOI (Digital Object Id)

4) More work on developing and improving these links is being launched by ISO TC 46 / SC 9

Will link when the complete post is up.

MC said...

Glyn - this is less a debate about pricing however to answer your specific question, any ISBN Agency is able to charge for ISBN's on a cost recovery basis. This is a standard (and appropriate) ISO accommodation to effectively run one of these operations (particularly in larger agencies such as the UK and US). Having said that the UK, US, Germany and one or two other agencies are unusual in that they are run by commercial companies. Most of the 170 worldwide agencies are departments of the national libraries of those countries. Many are run well (some also charge) but in most of those markets the book supply chain is no where near as sophisticated as the US, UK or Germany's, and ISBN's are not as fundamental in many of those countries as they are in the US, UK, Germany (and France, Italy, Scandinavia and a few others).

MC said...

Michael W, These are great comments and should be part of the discussion however I think you are starting to merge together the ISBN with the need to change core bibliographic data that 'rides along' with an ISBN. That's a whole new post.

MC said...

Todd - thanks for the post. I think we are all aware that each eBook format gets its own ISBN but thanks for restating it clearly. And I really hope there are enough ISBN-13s! (although I know that empirically to be the case).

Alicia Wise @wisealic said...

It's high time for the industry to rally round the ISBN.

Without the ISBN, properly applied to uniquely identify different editions (whether electronic or print) publishers will have to bear the higher costs of communicating about each book in a different way with each bookseller. This could only be in the interest of the very largest booksellers, and their interests are not always ours.

It costs something to administer ISBNs, and this cost needs to be recovered in some way. Perhaps the charges made at present are too high - I do not have enough nformation about this to form a clear view, and feel this would be an interesting issue to explore further - but the costs of administering ISBNs do need to be covered in some way.

Robin Tobin said...

Yet another excellent post here. I am in midst of moving bibliographic information on behalf of about for 210 publishers or so (for a couple of distributors) and not sure what impact in terms of bibliographic databases the ISBN actually has.

For people like us who work with the title information you would never use the ISBN to identify the title in any event as it is notoriously unreliable (data entry issues at some point in the chain usually) and obviously with a database then needs to create relationships to other information in the publishing process (print estimates; rights contracts; royalties etc etc) you would have to have a unique unmodifiable identifier for the purposes of ensuring you could do so.

In Onix parlance I guess for most bibliographic systems this would be the Record Reference and the ISBN in terms of the construction and management of the system is neither here nor there in any event.

Anyway, yet again thanks for the post and realise my perspective is limited but we do build a lot of those bibliographic systems over a number of years and the mention of systems did spark the thought.

Brian Green said...

Here are just a few further points of information from the International ISBN Agency.

1) An initial Vote on whether or not to revise the ISO ISBN standard took place in 2008, 3 years after publication of the 2005 ISBN standard. There was no vote to revise it. Subsequent ballots take place every 5 years (not 10).

2) The Amazon ASIN is a good example of the sort of proprietary identifier that gave birth to the ISBN 40 years ago, when the trade realised that the supply chain simply could not cope with multiple identifiers for the same product.

3) It was primarily the libraries, wholesalers and retailers on the ISO ISBN working group who argued for separate ISBNs for each e-book format for discovery, ordering and tracking purposes. They continue to argue that need and, in most countries, the publishers seem happy to respond to it.

Susan Cross said...

As a first time author I am using a small self-publishing house who is handling the ISBN for me. Admittedly, I know little or nothing about the process but if books are starting to be purchased increasingly through websites, the question arises whether or not the ISBN is necessary? I am a baby boomer so I follow traditional rules, but will the younger generations who self-publish continue to go along? Interesting material for an uninformed person like me.

Fran Toolan said...

This is a topic I find impossible to stay away from. I agree with Michael that the ISBN as we know it will be a dead animal in the electronic world.

I think that it bears remembering that the genesis of the ISBN was to help the book publishing supply chain, but the ISBN never worked for Mass Market titles that were sold on retail outlets other than bookstores. That’s why we had to play the whole UPC game. And because bookstores wanted to sell other products than just books, the ISBN eventually had to morph into an EAN.

The ISBN is at best a tool for publishers and data aggregators. Publishers use it to report sales, and even record costs, most back office publishing systems have a strong tie to the ISBN for inventory control purposes, and virtually all royalty systems report using the ISBN as the roll up number for sales reporting. Aggregators use it for listing titles that are available.

All over the rest of the retail world, manufacturers assign their own identifiers, and retailers assign their own. Reporting systems have evolved to the point where most system can map multiple numbers together so that sales can be reported using the identifier needed by the recipient.

Publishers should assign ISBNs based on the way they want to measure what’s happening with their product. So, if a publisher wants to track their business at the high level of “Print” and “Electronic”, then they should only assign 2 ISBNs. If they want to track/report at the level of HC, PB, TP, PDF, HTML, whatever, they should assign ISBNs to each. The wholesalers, distributors and retailers can assign any number they want as long as they can map it back to the ISBN assigned by the publisher.

I think the whole notion of a single identifier for the entirety of the chain is what is really dead here. Publishers don’t have a linear chain anymore, and if they try to control all the branches in the electronic delivery world, they will simply fail.

Suzanne Allain said...

When I started my new publishing company, LeMoyne House, earlier this year, I purchased 10 ISBN numbers for $275.00. That seemed like a lot of money to me, a small, start-up publisher. But I figured it would at least cover the first 5 books I released, and $55/title was not an unreasonable amount, in my opinion.

However, I quickly discovered that those 10 ISBN's I purchased barely covered two books, once I realized that I was technically supposed to use a different ISBN for each digital format. Do the powers that be realize how many different digital formats there are? Even if you just release a book in Microsoft Reader, eReader, Adobe Reader, and Kindle reader formats, you've used four of your ISBN's at $27.50 a pop, plus one for the print version. That seems a little excessive. You're saying to publishers they need to spend nearly $150 on each title just to register ISBN's, and that doesn't even cover all the digital formats.

I don't think it's unreasonable to require one ISBN to be used for the digital edition of a book, but it does seem ridiculous to me to require a different ISBN for every digital format.

Suzanne Allain, LeMoyne House

Glyn Pope said...

If a publishing house says, 'yes we will publish your work but you pay for the ISBN' are they really publishing it for you?
There are examples of this happening.

Anonymous said...

Here at Antellus we don't use ISBNs anymore. The issue for us was cost, and since Bowker jacked up its prices for number blocks we did not see any alternative but to abandon the ISBNs in favor of our own catalog numbers. See, back in the day, "Books In Print" used to use the individual publishers' numbers to list the books. Now, Bowker is using the perceived need for the number by booksellers (promulgated by Bowker)to make money. In order to stay in business and continue to publish good quality books we saw that the only barrier to this was the cost of the numbers. In some countries, the numbers are free, so we are placed at a disctinct disadvantage. When "Books In Print" is opened up again to all numbers, the true picture of the publishing industry will be revealed. There are a great many more books in the bibliosphere than are listed.

Marty B said...

The ISBN is dead. Long live the ISBN...

In my opinion, the ISBN was dead when it became an EAN. After all, an ISBN now follows the EAN standard. Having said that, I still believe the ISBN/EAN has and will continue to serve a valuable purpose. Only chaos will result from not having a single numbering scheme. As just one small example, how do you think book jackets, trade reviews, 1st chapters, etc., get linked to the book? All via the ISBN. Without that single linkage, costs go up, not down. (More on this below - see the *asterisk).

Publishers do not pay for individual ISBN's/EAN's - they pay for a prefix, the size of which is based upon the number of books expected to be published. In fact, Bowker does not assign the ISBN # (aside from the prefix), the publisher does, within the rules of the standard. Aside from having to generate a bar code, generating an ISBN once the prefix has been obtained does not cost publishers money.

I agree with Brian Greene's comments: the emergence of multiple identifiers does nothing to help the publishing community. I believe that Amazon's insistence on assigning their own ASIN does cost them money because somewhere in their system, they have to maintain a relationship between the ASIN and the ISBN/EAN to facilitate efficient ordering between themselves and publishers. IMO, they should have only used an ASIN where the ISBN did not exist, which I suspect is less than 3% of time. I sometimes can't order a book on Amazon because I'm not sure if the edition they display is the same as one I already own or a newer one. Why is this so? Because they don't display the ISBN, so I can't compare.

There is a big debate within the "numbering" community of whether numbering systems should build in any meaning or not. I'm in favor of including meaning. When the ISBN-13 standard was debated, there was some thought given to eliminating the publisher identifier. I personally thought that was a bad idea. Therefore, I'm in favor of Michael W's idea to incorporate an extension of the ISBN to indicate the attributes of ebook editions. However, the media industry tried this many years ago by manipulating the last digit before the check digit of a UPC to denote format (VHS, Beta, CD, Vinyl, etc.), but because all of the movies studios/record labels didn't adopt that standard, it never really worked. (But check out most CDs and you'll see that the number before the check digit is almost always a "2"). I really liked the idea because it provided a generally foolproof way to link together different media versions of the same intellectual work. (Admittedly, the book publishing industry is a little different because the paperback is frequently published by a different publisher than the hardcover, whereas the DVD and the VHS (back then) or the DVD and the Blu-Ray (today) are published by the same company.

[*Years ago I had to build a way to link movie reviews from Variety to metadata from Bowker's video directory. There was no number associated with the review. So we built a matching algorithm based upon words in the title plus the release year of the movie within one year. The algorithm took into account numbers ("1" and "one"), symbols ("&" and "and"), etc. We expected an 80% match rate. We got 30%. The rest had to be matched by hand.

More recently I had to match a database of music tracks used in television programs (a "cue sheet") with the program itself in two different databases. I thought we'd get a 60% match rate. After all, how many different ways can one input "Seinfeld: The Soup Nazi"? We got a 5% match rate.

In both cases, an industry standard number would have solved the problem and substantially reduced costs. ]

MC said...

5/9/2009 From Andy Weissberg at the ISBN agency via email.

I have been reviewing your original blog posting, “The ISBN is Dead,” and the follow up comments that have been generated on the blog page. While the commentary and structure thereof does present some well-constructed points of view about all of the relevant issues and challenges, there are several components of your post that are somewhat or definitively inaccurate, semantically-challenged and in some cases, inflammatory in nature. I have not responded or made comments on the blog yet, hoping that we could have a discussion before doing so.

Had you known or investigated some of the facts, and based on your own past, current experience and position as a though-leader and consultant to publishers at large, I would have assumed that before publishing such a blog, you would have perhaps connected with me or others copied on this email in advance of doing so…nevertheless, what has been published has been published and is now countering all of the critical milestones we’ve been achieving as a supply chain to directly attack and address these issues, as well as some of the highly informative educational efforts that are being coordinated on International levels, with various trading partners, competitors and others in the space.

In advance of a phone call or meeting, I would like to share the following information to clarify some key issues for you, and perhaps others, should you wish to do so:

1. Although not “required” within their trading workflows, “Downstream suppliers” as you have described them still very much do regard the ISBN as an essential mechanism for cataloging electronic products and their distinct manifestations for discoverability and trading via their systems. In the interest of not violating confidentiality, I can tell you that two of the largest wholesalers in the world, as well as one of the largest publishers in our industry have recently purchased a large volume of ISBN prefixes to cope with the challenges as described in your blog. Irrespective of the discounted business models associated with such transactions and efforts, it is the downstream suppliers who ultimately agreed that the ISBN was still a qualified, “relevant” identifier for discovery and trading purposes. Even Amazon and its subsidiaries are now working very closely with Bowker in these areas.

2. On the topic of ISBN assignment to out of copyright works – your statement, “For example, while assigning ISBNs to pre-1970 titles may make an ISBN agency's revenues bulge, it may not be the most effective proposal for the supply chain” – is inaccurate and inflammatory in nature. In addition to working collaboratively with many libraries as part of a FREE ISBN PROVISIONING GRANT PROGRAM to assign ISBNs to out of copyright works, for some time now, Bowker is also working with a host of digitizers, consortiums, associations and POD providers to freely or on a highly discounted basis assign ISBNs to these works, orphan works and otherwise. Relating to Google Book Search, without ISBNs, how else would you propose that these works are identified within a standardized bibliographic database leveled playing field?

3. We realize and appreciate the pricing-related concerns expressed by publishers about the costs associated with assigning ISBNs to multiple e-book formats, as well as workflow-related concerns. Although not public, The US ISBN Agency will be introducing a completely new and discounted pricing schedule for ISBNs in 2010. We also continue to invest in, develop and make available a number of value-added services that publishers and authors can leverage in conjunction with their ISBN purchases to enhance and improve the discoverability of their products and content in digital environments, and drive sales. You can see an example at http://seo.bowker.com/product/book/Virtualpolitik/9780262123044/html.