Ebooks and ISBNs

[Robbeli]

Originally Posted by Jellby

For me a trillion has always been 10^18, so making it 10^12 is certainly making it a lot smaller

The actual number is even smaller. The first three digits indicate that it is a book. They are always 978 (can also be 979 in the future). The last digit is a checksum digit, so it leaves only 2*10^8 possible numbers. Considering that the ISBN includes a country code and a publisher code, so you cannot use every number, it probably leaves only 2*10^7 usable ISBN's. Twenty million books is not that much.

[HarryT]

Originally Posted by pilotbob

I never understood separate ISBN for Hardbook vs Paperback Vs Trade either. It just seems silly and a way for whom ever hands out the ISBNs to make money.

It makes like a heck of a lot easier for the customer who wants to buy, say, a hardback specifically. I often do ISBN searches on Amazon for that very reason. I'm strongly in favour of retaining separate ISBNs at least for different paper formats.

[pdurrant]

After the large upheaval of moving to 13-digit ISBNs, and given that they're now EAN-13 codes as well, the chances of the ISBN format changing in the next decade is as close to zero as makes no difference.

Originally Posted by poohbear_nc

It would seem workable to have a common 9 or 12 digit base ISBN for an ebook - with an additional last digit/letter changed to reflect the format it's in. If in some fit of logic the last digit/letter were standardized for format, it could make search filters quite effective for sorting by format.

[pdurrant]

The current rule is that each separately orderable format has to have a different ISBN. Different ebook formats only need different ISBNs if buying one doesn't buy you the others. i.e. all the current DRM formats!

It is a bit of a mess, but I haven't worked out how to answer the survey yet. Mapping my thoughts onto their questions and answers is tricky.

Originally Posted by Nate the great

An ISBN is a 10 (or 13) digit number that's basically a serial number for books. If a book came out in multiple formats (paperback, hardback, etc), each format had its own ISBN. This applies to ebooks as well; an ebook has a different ISBN for each format sold (LIT, Epub, Mobi, etc). The ISBN Agency has decided to reconsider this decision.

[JSWolf]

Originally Posted by pilotbob

Is this really what you meant to say... or did you mean to say:

"The idea of a single ISBN for each format of a given eBook is a bad idea."

???

I think 1 ISBN for a book title is plenty fine. What is the different what digital format it is in. I never understood separate ISBN for Hardbook vs Paperback Vs Trade either. It just seems silly and a way for whom ever hands out the ISBNs to make money.

Sure, each edition can be a different ISBN, but if the content is the same, what difference does it make what the media format it is delivered in?

BOb

I do like the idea of a single ISBN number for a given book with say a letter at the end designating which format it is in. That means only paying for one ISBN number and then getting the ISBN to be able to differentiate from the different versions.

[Solitaire1]

Originally Posted by Robbeli

The actual number is even smaller. The first three digits indicate that it is a book. They are always 978 (can also be 979 in the future). The last digit is a checksum digit, so it leaves only 2*10^8 possible numbers. Considering that the ISBN includes a country code and a publisher code, so you cannot use every number, it probably leaves only 2*10^7 usable ISBN's. Twenty million books is not that much.

It is actually much smaller than that. Only three digits are used to identify the book itself, so that gives the same publisher only 1,000 ISBNs for each country code (unless the publisher has more than one publisher code).

After thinking about it, a more workable solution to the problem it could be a universal book ID number (one completely separate from the ISBN) to identify a specific book. When a book is first released it gets an ID number, and all future versions of that book carry the same ID number.

To provide an illustration, the code could be as follows (each part is separated by a period):

- Language Number: A three digit code that indicates the book's language.

- ID Number: A nine digit code (allows for 1,000,000,000 titles) that identifies a specific book.

- Abridgement Number: A one digit code that indicates if the book is/is not an abridgement of the original book (not uncommon with audiobooks).

- Version Code: A three digit number to the specific format (printed hardback, printed paperback, PDF edition, and so on).

Just a thought. What do you think?

[fugazied]

I think libraries are quite capable of managing their meta-data and they are the kinds of organisations who need additional meta data defining the particular format of the book in question.

AFAIK In the library a book has the same ISBN because it is the same work, then it has additional data in their records stating if it's an ebook, book, pdf, whatever. If you get formats involved in ISBN for what is essentially the same book (in terms of words) it limits the numbers of ISBNs I guess!

[Robbeli]

Originally Posted by Solitaire1

It is actually much smaller than that. Only three digits are used to identify the book itself, so that gives the same publisher only 1,000 ISBNs for each country code (unless the publisher has more than one publisher code).

The different parts are variable. Country code is one or more digits, publisher code also. So some large publishers can have more then three digits, small publishers less.
As for your suggestion to create a totally different number, as far as I am concerned, just keep the ISBN. Publishers, booksellers, libraries, everybody is using it. It would cost way too much, to rebuild all there systems.

Originally Posted by fugazied

AFAIK In the library a book has the same ISBN because it is the same work, then it has additional data in their records stating if it's an ebook, book, pdf, whatever.

O no ! A real librarian would want a different ISBN for a pBook and for an eBook. Because the bibliographic rules say so, and because handling these books differs.

[charleski]

Originally Posted by Falbe Publishing

I suppose the case could be argued that every format needs a unique identifier

If different formats are sold separately, then they need different ISBNs, otherwise you break the entire rationale of the system. Distributors and retailers need to be able to refer to a specific product solely through the ISBN. If not, then they'll just end up developing another parallel system and we'll be in more of a mess.

but there's NO WAY I'm going to pay for an identifier for every format of every ebook I produce

Yeah, that's certainly an important concern. Another is the inefficient nature of the current system, whose segmentation must mean there are large amounts of allocated-but-unused ISBNs.

If the almost-trivial change of moving to 13-digit EAN codes was a 'large upheaval' then I shudder to think what will happen when real change is needed .

[pilotbob]

Originally Posted by Solitaire1

ID Number: A nine digit code (allows for 1,000,000,000 titles) that identifies a specific book.

Actually, if you allow for 9 characters not just limiting it to 0-9 you have 0-9 and A-Z maybe skipping O and I to not confused them with 0 (Zero) and 1 (One). This would give you a heck of alot more than 1billion codes. So, you could even make it a bit shorter and still allow for alot of books. (Perhaps someone that is better at math than I am can figure how many that would allow. I always forget how to figure this out.)

But, I like your idea... the root of the ISBN would be for the content then a few extra digits for language, format, publisher, whatever.

Then search engines could easily adjust to show you not just the format you searched for if you look for the full ISBN but for all the formats of the same book.

BOb

[kovidgoyal]

Why bother with trying to encode all this information in a number? Modern e-book formats (e-pub) have fairly good support for metadata in a simple to access form. Just use it for search engines.

The fundamental purpose of an id number is to unambiguously identify something. In the digital age, when books are so easily modifiable by users (they can convert, edit metadata, even insert content), it's pointless trying to uniquely identify anything. Instead ISBN should be used to point to the "canonical" digital edition of a particular logical book. That is the digital edition registered as authoritative by the author/publisher with the organization that administers ISBN.

[DawnFalcon]

Originally Posted by Robbeli

O no ! A real librarian would want a different ISBN for a pBook and for an eBook. Because the bibliographic rules say so, and because handling these books differs.

Well, Baen went from ISBN to DOI and then to their own system because of the costs of dealing with a large number of "editions" of their books, and I can't blame them.

[kennyc]

Originally Posted by JSWolf

The idea of a single ISBN for every format of a given eBook is a bad idea.

Oh, wait, I misread this. I think there should be a "base" isbn and then a "format/edition" designator.

[Solitaire1]

Originally Posted by pilotbob

Actually, if you allow for 9 characters not just limiting it to 0-9 you have 0-9 and A-Z maybe skipping O and I to not confused them with 0 (Zero) and 1 (One). This would give you a heck of alot more than 1billion codes. So, you could even make it a bit shorter and still allow for alot of books. (Perhaps someone that is better at math than I am can figure how many that would allow. I always forget how to figure this out.)

But, I like your idea... the root of the ISBN would be for the content then a few extra digits for language, format, publisher, whatever.

Then search engines could easily adjust to show you not just the format you searched for if you look for the full ISBN but for all the formats of the same book.

BOb

I thinking of strictly numbers when I mentioned nine digits. A problem I've run into when mixing letters and numbers is that you'd have to eliminate certain letters (in addition to O and I, you'd also have to eliminate C, D, L (if lowercase is included), and Q) to avoid potential visual confusion. A numbers-only code also eliminates potential language difference confusion (to use a chess example, in algebraic chess notation the bishop can be written as "B" or "L" depending on the language).

As far as including publisher information in the code, I think this code should be based strictly on the book itself. Anything independent of the text of the book itself (like the publisher) should not be included in the code.

As an example, with a book in the public domain many publishers from around the world can publish the same book. Each published book would have a different ISBN, but the ID Number for each released version of that book would be the same. With a computer search, it would be easy to find all versions of the same book. Additional parts of the code would allow you to pinpoint the specific version of the book you want (such as "I want an unabridged paperback version of The Wizard Of Oz in French." or "I want an abridged AportisDoc version of Silas Marner in American English.").

[radius]

Originally Posted by pilotbob

Actually, if you allow for 9 characters not just limiting it to 0-9 you have 0-9 and A-Z maybe skipping O and I to not confused them with 0 (Zero) and 1 (One). This would give you a heck of alot more than 1billion codes. So, you could even make it a bit shorter and still allow for alot of books. (Perhaps someone that is better at math than I am can figure how many that would allow. I always forget how to figure this out.)

Hi pilotbob,

Think of it this way...

When you have a single digit number which can be any of 0..9 there are ten possibilities right?

If you have two places for digits, and the rightmost one can be 0..9, there are ten possibilities for that digit. If the leftmost place can also be 0..9 then there are there are ten possible first digits, and for each of those there are ten possible second digits, or ten times ten (ie: one hundred) possible combinations.

ie: For the first digit 0, there are ten possibilities for the second digit (0..9).
For the first digit 1, there are ten possibilities for the second digit
....

Makes sense that from 00 to 99 there are one hundred combinations right?

The same thing applies if you allow possible "digits" other than 0..9

So if you add 24 letters in addition to the 10 existing digits, then you can have (24+10) * (24+10), or 1156 possible combinations with a two place number.

With 9 places, you can have 34^9 or just short of 61 * 10^12