Add to Your Library: suggested improvement

The Add to Your Library button available when browsing another user's library often doesn't give very good results, especially when dealing with traditional tales that are available from many publishers.

I'd like to suggest that you introduce one of two improvements:

1. Make it act as one would expect from the name, i.e. add a copy of the exact same book (same ISBN, same publisher, same title, etc.) to my library;

2. Give options for what info to copy into the search box in Add Books. The ISBN would be my first choice in most cases.

It used to act as you describe (in point 1) and would make a direct copy of the record, but this was changed a couple of years ago to redirect you to the "add books" search. There have been many, many threads on this topic.

And the reason it was changed to the current function was that copying existing records let to propagation of inaccurate data to a degree that was unacceptable to Tim. I can see both sides of this one: corrupt data (such as ISBNs associated wrongly) causes all sorts of problems as it moves further from its originator but the need to find the most appropriate source is appropriate when the data is sitting on screen from a reliable user whom I trust.

As for the ISBN, that is not appropriate for books without ISBNs (anything before the 1960s, for example) whereas title applies to every book.

> There have been many, many threads on this topic...

... and the one with the most definitive information to which we refer all users with this type of question is...

Doesn't there have to be a point where one trusts that turning off the plus sign has propogated enough good data that it can automatically add a book?

If it hasn't propogated enough good data then one has to concede that external data sources are deeply flawed and aren't improving the data pool. If that's the case, then there's really no harm in allowing the plus sign to automatically add the book.

I'm just saying...

5> Turning off the green plus sign didn't "propagate good data" so much as it stopped the propagation of all data within LT, forcing all book data to come from an external source.

And the threat of bad data doesn't come from the external data sources so much as it comes from users who indiscriminately edit bad data into their books.

forcing all book data to come from an external source.

Or you can add your books manually. But I agree that probably most of the really bad data is from bad user editing.

Manually adding a book means the data still comes from a source external to LT. :)

>6 saltmanz: I've been here a while - and that's always been the reason given for turning off the complete functionality. To give people the opportunity to have it go to the search books page in order to stop the potentially bad news of putting someone else's junk data into your catalog.

Okay - so we've been doing it for two years. Which means we've been forcing people to either manually enter books or look up things through legit sources for that long. According to the Zeitgeist, the manual adds aren't exactly the #1 choice for books. (Only Tim would really know the click-through rate for + sign to manual add vs. actual data source.)

But my question remains the same - if the object was "better data," when are we there? Like anything one does, there comes a point of diminishing returns. When are we saturated with "mostly good" data? As in - have the 20,000,000 books that have been added since the + sign was largely disabled added enough good data to make folks feel better about the data quality involved in using the + sign to add books?

If the answer is, "the data quality isn't that good," then I question whether it will ever really be "good." Because that means the green plus sign wasn't the key to preventing bad data - and maybe it's time to turn it back on again and stop the constant stream of requests for the feature to work as expected.

ETA - fixed numbers.

We will never be there - because people still do the process of downloading one record repeatedly and editing sloppily to make all the other records they want without searching each one. The change in the green plus behavior does nothing to improve data already in LT, or prevent downloading bad data; it just prevents replication of the bad data already in LT.

>10 staffordcastle: And there's a diminishing cost to that behavior. I'm simply asking where the line actually is.

This feature is common and expected - it doesn't exist here for the "sake of data integrity" - something some, but not all, users care about. At some point, there will be enough data integrity that you'll maintain a solid mix. You can't simply assume that the plus will pull any greater percentage of bad data than any other official source that LT currently pulls.

At some point the perceived protection of data will be outweighed by the perceived value of the plus sign working as expected - adding the book and not just going to a search screen.

I'm not making this up - it's an actual business principle.

I guess I don't understand: what is wrong with going to the "Add Books" page? It isn't that hard. (I've done it nearly 3,000 times)

12>

I don't understand it either. I mean, I understand the convenience for wishlists, or for the occasional "oh yeah I had that", but really, how hard is it to click "Add Books" rather than clicking the green plus and then type a word or two? I get the impression that most of the people grousing about the green plus never add books any other way, which must take much longer than just going directly to Add Books!

12, 13: I also use the Add Books page, and find that intuitive, but there is a plus-sign feature, and it would be nice if it worked well.

I actually think that they now have a good model for making it work well and intuitively - and without a major degree of bad data problems - on the model of the new interface for "quick links" and all the "get this book" stuff. I even started a thread about it: http://www.librarything.com/talktopic.php?topic=78354
I'm not a programmer, but it seems like it might be be pretty simple to modify the code for the lightboxes, that allows you to select particular editions to "get", in order to run an "add books" command for a particular edition, on a particular source. It wouldn't work for manual entries, but then nothing's perfect, and it would be a lot better than the current plus-sign feature.

12, 13> I think many of the people who want the green plus to work have a lot of books that can't be found in the usual sources, like comics, or railroad schedules.

My vague impression is that the plan for the green plus is connected with the idea that Tim has mentioned a few times about creating abstract MARC-related work records, and apparently Casey is working on this. So I think there is something in progress on this.

15> Indeed. Earlier this year, I cataloged my entire comic book collection on LT. Despite the fact that I have a few hundred issues of overlap with member readafew, I still had to manually add all of them.

(Granted, comics are perhaps slightly outside the intended scope of LT, but it still illustrates the point made above.)

15>

Sure, and that's fine, but unfortunately the green plus in practice wasn't restricted to sensible uses like that. Now, I'd be all in favor of a green plus on the book page, rather than the work page, that automatically used the same source as the other user, or that cloned a manually entered record, but I don't think it makes a lot of sense from the work page, other than for wishlist books where you don't care about the edition.

11: But the way that LT currently works, no matter how much good data there is, bad data of this particular type can cause nasty problems. Not because the author or title will show up wrong on the work page, but because if even a single entry has some other book's ISBN attached to it that other record will be a suggested combination. That doesn't get drowned out; there is not a diminishing cost but a fixed one.

18: How about rarer sci-fi novels where you have to compare LT libraries and Worldcat in an attempt to maybe find a holding?

if even a single entry has some other book's ISBN attached to it that other record will be a suggested combination. That doesn't get drowned out; there is not a diminishing cost but a fixed one.

So this is the real problem that needs to be addressed, not the green plus sign per se. Maybe combinations shouldn't be suggested based on ISBN alone, if the titles are vastly different. Maybe there's enough data by now that one single book shouldn't be taken into consideration at all. An ISBN should only be considered a part of the work if there are at least, say, 10 instances of it connected with the work (with possible exceptions made for uncommon works that have very low numbers to begin with). If one anomalous book is combined into a work with 1000 or 10000 copies, it should be isolated somehow and its data not allowed to contaminate the whole thing.

20>

Sure, I was referring to the general case of "hard to find items"; having a shortcut that means "use the same source that this user did to add this book" would be fine.

(I find UC Riverside has good holdings in SF, BTW).