I’ve been using ISBNdb.com to provide a link from library pages for a known item book to a vendor-independent way to buy a book online. I don’t like letting Amazon or any other particular vendor get a monopoly. ISBNdb.com provides a list of a variety of vendors, with prices.
ISBNdb was the first thing I discovered that will let me do this. And ISBNdb has a really nice API, although I don’t really use it for much other than linking to the page. But I like supporting them (they get the Amazon affiliate ID etc), because the guy seems like a good guy, and the API is nice and I might want to use it some day.
The problem is that the ISBNdb.com interface is kind of messy. And it only updates prices ‘on demand’, so if you ask for prices for an item that hasn’t been looked at for a while, they might not be up to date (but will trigger ISBNdb’s crawlers to go fetch new prices, which will be there in a few minutes — which is too late for ‘just in time’ for the user!).
So then I discover BookFinder. Similar service as far as providing prices from multiple vendors. Much cleaner interface. The ads are more subtle. It does have a direct linking syntax, but doesn’t actually have an API. BookFinder actually conducts the web crawl really just-in-time, before showing you the answers. Which I think is a better interface, but leaves no good way to pre-check to make sure BookFinder had a page for the relevant ISBN. — but experience with these ISBN-based services tends to be that if if the ISBN exists, they WILL have some data.
So what do you think, should I swap ISBNdb.com for BookFinder in Umlaut?
update. Hmm, one downside of the BookFinder interface is if I just feed it an ISBN, it doesn’t actually give me feedback as to what book I’m searching for, to make sure it was the right one.
Bibliographic Wilderness 
How odd not to include the title. I like ISBNdb’s listing the basic bibliographic info once at the top but then it goes on to include the store title over and over and that’s kind of irritating. Is there that much variation in “store titles” — the term ISBNdb uses? I don’t mind waiting a few seconds for the Bookfinder to display — it has visual clues that tell me it’s working and even tells me what it’s searching for.
What would be nice is ISBNdb’s content with the crawl at the point of search feature added and a little usability/graphic design work. It has a lot of nice features for someone doing general book searching (WorldCat look out.)
Too bad you can’t take some from one and some from the other.
I like Wikipedia’s Special:ISBN page. It gives links to dozens of booksellers and libraries.
Sorry, correct name is Special:BookSources. Example:
http://en.wikipedia.org/wiki/Special:BookSources/0304522570
Scroll down that page to see all the links it generates.
Thanks for considering BookFinder.com, Jonathan. I’m the site’s founder, and I appreciate your giving us a look, and the feedback.
BookFinder.com has a library heritage (it was a spinoff from a 1996 project at UC Berkeley’s School of Information), and is built on top of somewhat FRBR-esque concepts.
There’s more we can be doing for ISBN searches, and we’ve been experimenting with ways to improve (e.g. see one of our experiments, at http://www.bookfinder.com/dir/i/Living_My_Life/0142437859/ ).
We’re strongly committed to offering live realtime results from the widest possible range of sources, including databases of specialized international and antiquarian collections. Unfortunately, this means we can’t predict whether a book will be in stock or not.
For best results, you may also want to consider linking to author/title searches on our site. Unlike most commercial book search systems, we easily handle search for the whole work, across editions, and including a wide range of non-ISBN-tagged data (and doing realtime clustering to handle resolution of ambigious cases).
Thanks Anivran. What I’d really like to do is search by ISBN (not by author-title), but then have BookFinder automatically show me all the editions, not just the ISBN I asked for. But identify the ‘work’ by the ISBN I gave you, not by author-title. Have you got the data to identify the work from the ISBN in your database already, with your ‘FRBR-esque concepts’? If so, that would be great.
In my experience and testing, I’m not happy with the poor precision on author-title searches. It’s important to me not to send my users to a ‘false negative’, many of our legacy systems perform so poorly with that kind of thing that I feel like I’m trying to regain trust by only giving ‘correct’ answers.
Likewise, while you’re doing a realtime search — I wonder if you could provide an API so my software could ask you to do a realtime search, then wait for the answers, then find out if there is availability, and only then actually present my users with the link. (I already have a number of ‘takes a while to find out for sure’ querries going on anyway, supported with an AJAXy spinner front-end). I’d use that for sure if you had it, again the idea of making sure not to send my users to a ‘false negative’ — there are a lot of links I’m presenting on the page for various services related to the book, and I try to have them all be pre-checking for the user and telling the user what they’re really going to get.
If you had those things, it would make it no choice at all to user your service.