Goodreads Librarians Group discussion

If you look on the search page you'll notice a new data source is now available: Barnes & Noble. We'll need your help to figure out any issues with the new data, so please report it here!

Some notes on how it's currently working:
- Books that are already sourced from Amazon won't be updated. This is probably most books, so the change isn't huge.
- For books that B&N has that Amazon & Goodreads do not have, we will import them. It will be interesting to see what kind of stuff appears. I know for instance that many of the B&N ebooks will start to appear.
- Imported B&N books will have a 5-60 second delay before their cover images appears. The reason is that B&N has nowhere near as cool image resizing abilities as Amazon does.

Ooh, this is exciting!

Indeed!

Otis, imports from B&N are showing up with a blank source field. Example.

Also, it's been several minutes, and still no image. But B&N does have one.

Edit: Image showed up sometime in the last 30 minutes. Guess I just wasn't patient enough. Still no source listed though.

I agree that this is exciting. I'm glad Goodreads has an additional data source.

Thanks- this is awesome! GR is getting better and better all the time!

Rivka: you're right the source wasn't showing up correctly - will be fixed shortly!

Very cool!

Otis wrote: "Rivka: you're right the source wasn't showing up correctly - will be fixed shortly!"

Oh, that is cool. It's a link!

I've noticed that (I think) B&N data is overriding the bookcovers from Amazon. Two examples:

http://www.goodreads.com/book/show/97...
and
http://www.goodreads.com/book/show/47...

When I noticed the covers weren't the right ones on my shelves, I deleted the ones that were there and it reverted back to the correct cover from Amazon. The next day, the strange covers were back. Checking the log, it only has my changes there, so my guess is it's the B&N data? Any way to keep it from overriding the existing covers?

I'm having the same problem with this book

Oops - it shouldn't have been doing that - we'll fix it.

Although we are evaluating if we can replace all the Amazon data (mostly the images & descriptions) with B&N data. This will allow us to not have Amazon as the exclusive bookseller on our book pages anymore, among other things.

We are building scripts to look at the differences in the data for the most popular books - but it's interesting that Amazon and B&N had different covers for those titles. Do you think that is common?

I'm not sure how common it is - with "Unbearable Lightness", I think it's a case of the same ISBN and a different cover that we see all the time. "Amulet" is really the same cover, but a crappy quality version with a weird yellow tint. Those are the only ones I've noticed on my shelves though, so if it's overriding for all the covers, then not very common I guess?

I saw it too for this book. I had just read it and checked/completed the data and the book cover was correct. Then I noticed a week or so later that the cover was off...the color of the author's name and the title are weird. The log showed no one changing the cover but when I deleted it, the right one showed back up.

Now I just checked it again, and the weird one is back. So it looks like it's related to this B&N issue.

Plus, B&N apparently doesn't have a full size cover. The large thumbnail is much smaller than the Amazon large thumbnail.

This is another book that looks to have gotten an overriding B&N image that wasn't as nice as the Amazon one.

(I deleted the B&N one)

Found yet another books where it looks like a less nice image from B&N has overridden a better Amazon cover. The large thumbnail of the B&N version is so much smaller than the Amazon one.

(haven't yet deleted this one...I'll leave it for now so Otis can see)

...so far, I'm not impressed with the B&N cover images

Sorry Isis, but I'm getting a lot of covers for books from B&N that aren't available through amazon.

I'd prefer having covers to no covers at all.

I don't mind that we are getting covers from B&N that we otherwise wouldn't have, but I don't like the fact that poorer quality images are replacing nicer ones.

It seems like it would be more useful if a B&N image was only added if there was no available one from Amazon.

It seems like it would be more useful if a B&N image was only added if there was no available one from Amazon.

I agree, Isis. Allowing it to add, but not override images would be a nice compromise. And maybe the option for librarians to manually override Amazon images with B&N ones? Sort of like those "Refresh info from Amazon" links we have on the edit pages?

A couple more that I noticed browsing through my shelves: here and here.

We fixed it so as of 5min ago only BNN sourced books are getting new images. More updates shortly :)

Thanks, Otis!

We all appreciate how hard you work on this site and how you listen to all our opinions :)

Ditto what Isis said! Thank you! :)

Ok so in order to better compare the images from B&N and Amazon to see the differences, we made a page that shows the covers from all books with over 1,000 ratings. We need your help to scan it and let us know how many look different. Please disregard the size difference. Also please note the page may take several minutes to load!

http://www.goodreads.com/system/bn_bo...

We also made files for other fields. These will load faster as there are no external images to load:

http://www.goodreads.com/system/bn_bo...
http://www.goodreads.com/system/bn_bo...
http://www.goodreads.com/system/bn_bo...
http://www.goodreads.com/system/bn_bo...
http://www.goodreads.com/system/bn_bo...

Otis wrote: "Ok so in order to better compare the images from B&N and Amazon to see the differences, we made a page that shows the covers from all books with over 1,000 ratings. We need your help to scan it and..."

*whistle* That's a lot of book covers!

Working from the bottom, in the last 10% of the page I see 20 different covers and 2 where the B&N cover is (size aside) of a dramatically lower image quality. How many books in total are there?

I wasn't counting the number where there's no Amazon or no B&N cover -- I assuming you can count those programmatically...?

Scrolling down from the top, perhaps 5-10% of the way through the list, I found 10 cases where the covers were different, two of which were covers for completely different books.

There are 5947 books listed in the images file. If Michael and Cait both were right in looking at 10% of the file, that means 10-20 images out of every 600 are off (or 1-3%). A few more estimates would be nice still.

Based on this, 1-3% is not huge, but not an insignificant number. So the question becomes if the difference is acceptable. Are Amazon's images correct or are B&N's? Or are they both right and thus it doesn't matter? Maybe someone can look at 4-5 books where they are different and do case studies?

I'm inclined to think that the B&N images are the better ones, but I'm not always the best at tracking cover art for editions -- does someone else want to look at these? It's the next 10% up from the bottom (I didn't want to load the images in this thread, but you can cut and paste the quoted text into a local file for yourself), with the Amazon image first and the B&N second:

<html><body>
henry huggins - beverly cleary
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
lord of the rings and the hobbit - jrr tolkien
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
ark angel - anthony horowitz
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
castle - julie garwood
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
the enchanted forest chronicles - patricia c wrede
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
the castle in the attic
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
the skin i'm in
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
tracks
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
childhood's end
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
outer banks
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
the little engine that could
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
cereal murders
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
seedfolks
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
body of lies
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
ex machina the first hundred days
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
things not seen
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
young goodman brown
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
daughter of the blood
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
the poetry of robert frost
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
v for vendetta
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
charly
<img src="http://ecx.images-amazon.com/images/I...
<img src="http://images.barnesandnoble.com/imag...
</body></html>

(By the way, in the bottom 20% there are 121 covers that Amazon has and B&N does not and 9 that B&N has but Amazon does not, which is a little painful!)

Hey guys,

Just another update on Barnes & Noble. I've included a link to descriptions below so you can see how Goodreads, which may have been edited by librarians, compare to Amazon and Barnes & Noble for some of the more popular books on the site. This should give you an idea for how the descriptions might differ. We're in the process of switching data from Amazon to Barnes & Noble, so we'd appreciate any recommendations on what to do based on this data.

Here's the link to descriptions. It's a big file and may take awhile to download:
http://www.goodreads.com/system/bn_bo...

Here are the options we're considering:
- leaving Amazon descriptions and just hiding them places we can't use them
- switching entirely to Barnes & Noble
- letting librarians decide on a case-by-case basis

Please let us know if you have any feedback!

Ben

Why are you switching at all? I thought you were supplementing Amazon data with B&N data, not replacing it.

I'd be inclined to leave librarian edited descriptions alone. They have likely been customized several times over. It would be a shame to see that data overwritten with whatever happens to be on B&N.

Are book cover images switching completely over to B&N? From experience, I've found B&N covers to be of lesser quality than Amazon, especially with regards to the image size.

We are switching because we are very limited by Amazon. Primarily, if we completely switch, then we gain two things:

1. The find at links on book pages won't have to have Amazon as the exclusive bookseller anymore.
2. We can build an iPhone app and a android app.

In fact you'll notice #1 already applies to B&N sourced books.

So the plan is:

1. For images, replace with B&N, and let librarians fix any images that aren't perfect. From visual scans we've only found 1-2% of the images on B&N to differ from Amazons. We will have a source on images now so you'll know if it came from B&N or a librarian or Amazon.

2. For descriptions, we're not sure yet. But we can't use Amazons or derivative's of Amazons. We will also be adding a source on descriptions so we know where they came from. We'd love feedback on what to do here. Checking out Ben's file to see the differences would be a great help.

3. For other book meta-data which is factual in nature, we will use B&N to supplement and confirm what we have from Amazon. We will be very careful here not to undo any work Librarians have done. Because the information is factual and fairly consistent across various sources, I'm not too worried about this one. The source of the book will represent where the meta-data excluding the image and description came from, so where B&N has a record we can confirm we will update the source.

Once steps 1-3 are done then we are free of Amazon's constraints!

Thanks Otis for the information, it now makes a little more sense.
As long as Librarians have priority ie their work stands over Amazon and B&N descriptions/images I'll be happy.

Otis, This all sounds terrific.

Re: For descriptions, we're not sure yet. But we can't use Amazons or derivative's of Amazons.

What happens with the huge number of books already in the database?

It sounds fine with me to go with B&N and have libriarians edit images/descriptions as needed.

What happens with the huge number of books already in the database?

That's what we need to figure out, as we need to update them! The cases we need to decide are:

1. Books where our description matches exactly Amazons and is different from B&N's. Do we update to B&N's?
2. Books where B&N has no description but Amazon does. Leave it as from Amazon and keep it as an Amazon sourced book?
3. Books where our description is X% or more similar to Amazons (so it's a librarian-edited version of Amazons original description). Do we move to B&N's or leave it as an Amazon sourced book?

For books with no description we use B&N. For books where our description is significantly different from Amazons we assume it was added by a librarian and leave it alone. For books where B&N has the same or similar description as Amazon and us anyways, we leave it alone.

The problem with #1 is that while it may mean the description came from Amazon, it may also mean that both GR and Amazon have a mutual source (via different channels), like the back-of-the-book/bookflap-description.

Otis wrote: "1. For images, replace with B&N, and let librarians fix any images that aren't perfect. From visual scans we've only found 1-2% of the images on B&N to differ from Amazons. We will have a source on images now so you'll know if it came from B&N or a librarian or Amazon."

I completely understand the desire to separate from Amazon because of the limits it places on the site.

I guess, just from experience using this site, Amazon, and B&N over the past 2 years, Amazon seems to handle data from the genre I read much better. From B&N, I've seen quite a few covers that are much poorer quality, much smaller in size, fewer available descriptions, less series info, etc.

So for me, the switch is a catch-22. But like I said, I can see the point. I just wish it didn't mean more librarian work needed to be done on books I read.

Otis wrote: "We are switching because we are very limited by Amazon. Primarily, if we completely switch, then we gain two things:

1. The find at links on book pages won't have to have Amazon as the exclusive bookseller anymore.
2. We can build an iPhone app and a android app."

Ah. I'll be honest. Except for eBooks, I primarily buy books through Amazon, so I don't care about seeing links to other booksellers. And I don't have a web enabled phone, so I don't care about iPhone apps and the like. So, from my perspective, there is no immediate gain and lots of potential for loss. But I seem to be in the minority on both fronts.

I glanced at Ben's file, but without titles or authors it looked like a game of "guess the book".

rivka wrote: "The problem with #1 is that while it may mean the description came from Amazon, it may also mean that both GR and Amazon have a mutual source (via different channels), like the back-of-the-book/bookflap-description."

Maybe add a field that allows librarians to specify the source of the description - bookflap, publisher site, author site, librarian generated, or Amazon/B&N. I guess the only disadvantage is that you'd have to wait for the database to be manually updated. But I guess the default would be which bookseller created the original record.

*shakes head* Don't mind me. I'm particularly unhelpful tonight/this morning.

Otis wrote: "We are switching because we are very limited by Amazon. Primarily, if we completely switch, then we gain two things..."

I have put my own images of covers from older OOP books both here AND on Amazon. I hope these will not be replaced by BN's missing cover image.

I am wondering if it is the Amazon vs BN API that is troublesome or what you have to do or pay for permission to use their data.

Being on slow dialup, like 30+% of US users, I hope that grabbing live data from off site sources does not slow your site down. One of the best things about GR has been its fast speed.

Bill

I have put my own images of covers from older OOP books both here AND on Amazon. I hope these will not be replaced by BN's missing cover image.

Nope - we aren't going to undo any librarian work - images that have been directly uploaded will remain untouched.

Otis wrote: "3. Books where our description is X% or more similar to Amazons (so it's a librarian-edited version of Amazons original description). Do we move to B&N's or leave it as an Amazon sourced book?"

For what value of X can we start calling it a GR description? I know that when I go in to clean up a generic book description that I often end up editing it somewhat for content.

I will agree to anything that gets me an iPhone app and GoodReads our of Amazon control (I love options, I wish Amazon didn't limit them), but alas, I am one voice out of many :) I do know that I've been trying to go through the B&N ebook entries and adding data to them.

I do know that I've been trying to go through the B&N ebook entries and adding data to them.

Yeah, a lot of the ebooks seem to import without images?

I do think it would be awesome if GR could move away from Amazon's data and build an iPhone app, etc....I like the idea of having different "sources" for images, descriptions, etc., too!

Vickie wrote: "Could you please, for the information of us relatively newcomers to this group, briefly describe the problem with Amazon? This is probably mostly curiosity, but I don't think I'm the only one who ..."

Someone official will probably explain it better, but this is it in a nutshell. Amazon offers it's database for free, but it puts restrictions how you present the data. They want to be the only bookseller "advertised" and they don't want an iPhone app being created, because they see that as competition to some app they're creating.

Before we started using B&N as a source of data, we were told that other data sources, besides Amazon, were just too expensive. I'm not sure if B&N is now cheaper, or if there has been an investment to move away from Amazon.

Besides the business issues, some users have a bias against Amazon for various reasons. So they personally feel better moving away from being so tightly associated with Amazon.

On the other hand, I love Amazon. I have trouble walking and getting a book delivered to my door is the difference between getting a book and not getting a book.

With respect to B&N, is there any danger of them turning around and doing the exact same thing Amazon did, demanding priority links and/or restricting applications? I'd hate to get the whole system regeared and then immediately have the "same old" new problems.

Not sure if this is directly an issue related to the B&N imports, but I was playing around on GR today and found that an author with a 3-part name (Linda Lael Miller) had new imports from B&N that had the author listed at Linda Miller. I checked the B&N pages and they list the author's full name but for whatever reason, the GR book pages had the shortened one.

Not sure if maybe the data got updated on B&N post-GR importing, or if there's some sort of bug that imported the names wrong.

Not-the-Michael, while that possibility is non-zero, it is currently unlikely for a variety of reasons.

Abigail wrote: "Some users definitely have objections to some amazon practices, and to the review culture that amazon promotes. I'm not sure I'd agree that this is (necessarily) bias, although the two are certainly not mutually exclusive."

I was thinking about the "support independent booksellers" contingent when I wrote that, which I do believe is considered a bias.

Personally, I don't see anything wrong with Amazon's business practices. I think trouble makers who want to write misleading reviews will manage to do so no matter how you set up your review system.

I believe it has more to do with specific actions Amazon has taken in the past year or two. Regardless, I doubt a discussion here is likely to be productive.