LibraryThing now allows you to keep track of the languages in your collection. If you don’t want to do that, you don’t have to. If you do, the changes are far-reaching.

• Every book has three fields: primary language, secondary language and original language.
• Languages are drawn from Amazon, your library record or the whole LibraryThing collection (see below).
• The catalog shows “language” and “original language” fields. Go to “change fields” to see them.
• Language can be edited within your catalog, much as tags are.
• Power edit has a versatile “set language” feature.
• Each language has its own dedicated page (eg., French). At present, these only show the most popular works originally in that language.
• Your “Fun statistics” page crunches the numbers on the languages in your collection.
• I’ve adopted the full MARC specification for languages, so you can catalog your Arawak and Elamite holdings. In most circumstances, however, you’re given a shorter list, with the option to see the full one

Not right? Don’t blame LibraryThing!

LibraryThing does its best, but it won’t always get the language right without some help. The reason has to do with the source of the data:

If you find your books through libraries, the languages are picked up from their catalog’s MARC record. That’s the theory. In fact, as we’ve discussed on the Google Group, library records are surprisingly sloppy with languages. (If you doubt that, click the “card” icon and look at the MARC 008 and 041 fields.) Polyglot libraries will cleanup. Of course, if you don’t care about the language field, you don’t need to look at it.

If you find a book on Amazon, LibraryThing guesses based upon which Amazon you used. That’s the best I can do, unfortunately. Amazon doesn’t tell me the language.

Because of the way “works” operate, if you leave the “original language” blank, LibraryThing will make a guess based upon the other copies of the work in the system. As elsewhere, these guesses appear in green. Green guesses are updated daily.

Let’s talk!

This is one of the more extensive changes I’ve made, with tentacles all over the functionality and code. Sometimes the “why” of a feature is complex, but I had to do it that way. Other times, I may have taken the wrong route. I’m guessing people come up with some great suggestions for changes or new, derived features. (And, as someone will surely point out, the system still has problems searching and sorting diacriticals. I’m working on it.)

I’ve set up two discussion threads in the Google Group, one for philosophy/functionality discussion, and one for bugs. I’m looking forward to what people have to say. Gratias tibi ago, Thingamabrarii.

I’ve added “why” links to the new “Pssst!” book recommendation feature. It’ll give you an idea where your recommendations are coming from.

I’m nearly finished with yet another recommendation algorithm. Then I’ll let them duke it out and users can decide which are the best. Or maybe I should have an algorithm to recommend an algorithm for you.

I just launched the new “Pssst!” feature, which examines your catalog against patterns in the catalogs of LibraryThing’s other 30,000 users to produce book recommendations just for you. LibraryThing already suggested books on a book-by-book basis, and matched you up with users who had similar libraries. This completes the circle.

To use the feature you need to have cataloged ten books. If you have, check our your recommendations. If you haven’t, check out my recommendations—ancient civilization and web design, it got me. (Incidentally, yes I have read Thucydides. I just have another 2,000 books to catalog!)

Recommendations are available in three types:

• People with your books also have…
• Similarly-tagged books
• Most popular books you don’t have

In each case, I’ve divided the recommendations into “fiction” and “non-fiction” (a rough, algorithmically-determined label). You can click a link to exclude books by authors already in your collection, so a mystery buff doesn’t get showered with minor Agatha Christie novels.

Needless to say, the point is not commercial. Although you can find commercial links, I didn’t add any to the recommendations page. I hope you’ll agree that LibraryThing—with no funding and building on a real community—produces recommendations as good or better than the ones you’re getting from Amazon, and ones not tilted toward current offerings or what you bought last week.* Community beats commerce, even in commerce.

Let me know what you think here, or discuss it at length on the Google Group.

The back story:

Long-time users may remember a previous feature that also produced book recommendations. I took it down because the math was so complex that it was slowing the site down, even when I required people to wait 20 minutes for the results. This new algorithm is much faster, relying on the existing book-by-book recommendations. (Book recommendations are regenerated on a sliding schedule, at night.) The old feature was also made before the “works system,” so it threw up a lot of books already in your catalog. Even so, I want to bring back suggestions focused on most-similar users, if I can do it in a way that doesn’t kill the server.

I agonized over the name—Recommendations? Suggestions? I even contemplated “Pimp a Book” (I own the domain). “Pssst!” sounds a little gimmicky, a little “marketing-ish,” too “Pssst!®.” The deciding factor was length. I wanted to give it its own tab—it’s a major feature already, and I have plans to expand it. The other options were just too long.

Lastly, I want or plan to extend the feature in the following ways:

• A way to mark what you think of the recommendation—good, bad, totally off-the-wall.
• A way to add a book to your wish list.
• Tracking the list over time, so you could find out the “new” recommendations.
• A “why” button. (For example: “Suggested because you own The Hobbit and Sexing the Cherry.”) That’s very expensive to do for all books at a time. One-by-one, I could “Ajax” the answer in, but so much Ajaxing—people will click “why,” “why,” “why” like mad—makes me nervous.

Oh, I renamed “extras” to “Tools and Toys.” I’ll be adding to that tab quite a bit in the coming weeks. Have the cool toys—chicklets, Thingamabrarians, that nifty LibraryThing-Outlook-PDA conversions, etc.—never made it off the blog or Google Group.

*I mourn the fact that my Amazon recommendations have flipped, from obscure Throwing Muses albums (of course, I have all of them, but Amazon don’t know that), to baby-care books. O for lost youth!

I think I’m going to tear my hair out with jealousy. Today’s New York Times has a gushy article about the media trading site Zunafish.com. Kindly look at the chart below, LibraryThing vs. Zunafish according to Alexa.

Update: WOW. That’s a big bounce! Let’s see if they hold on to much of it. Nobody ever holds the first-day bounce.

Zunafish is the long red line at the bottom. There are at least a dozen trading sites doing better. Heck, my ancient history hobby site is crushing them! Or check the blogs. Google Blog Search lists 2,235 blog posts about LibraryThing. Zunafish? Seven, five posted today! (They’ll no doubt be more soon.)

Nor is Zunafish a totally new deal. They opened in January. As the NYT writes, “Mr. Bloom and Mr. Elias said that the circle of traders had been limited so far — they did not disclose figures.” You’re not kidding. According to Alexa, 4,570,852 web sites have more traffic. Four-million.

But, as the article states, they’ve raised $485,000. I guess that buys them a PR campaign. No doubt their numbers will spike now that they’ve landed the NYT article.

It seems so terribly unfair. Press should follow success, not create it. LibraryThing’s traffic currently outranks booksellers Biblio and Booksense, all trading sites except Peerflix (eg., PaperbackSwap, Lendmonkey, FrugalReader, Bookins, SwapandSave, etc.), Amazon’s AllConsuming, the much-heralded Basecamp.com, and on and on. And yet LibraryThing’s press coverage has been largely restricted to The Christian Science Monitor‘s electronic edition and a piece in my home-town paper. Instead, LibraryThing’s grown on word of mouth.

I know. The answer is to get funding and to hire a PR firm. Forgive me for being idealistic, but it shouldn’t have to be that way.

I’ve added a tab-delimited export, for use in Microsoft Excel and elsewhere. Go to the Extras tab to download the file to your desktop.

Tab-delimited supplants the former CSV export, which had some drawbacks. (I’ll be working to get it up to snuff in the near future.) Among other things, it handles special characters better, and includes all your fields. Problems remain with non-Latin characters, such as Chinese and Georgian.

So far, testers on the Google Group have reported no problems. If you find one, go ahead and post it there, or here.

In other news, also check out the Google Group for a cool new beta feature I’m developing, and asking for comments on. It’s not ready for blogging yet.