Algorithms seem like perfectly pleasant things that are eager to help you find what you're looking for. But they lack that human touch. And the way Scribd sees it, a little dash of humanity is just the sort of thing you need when you're trying to figure out what book to read.

That's why the subscription ebook service is revamping its browsing and book discovery experience. The new feature, which debuts on Scribd's website Thursday and rolls out to the company's mobile offerings soon after, looks to emulate the experience of walking into a book store and getting recommendations on what to read from the frustrated liberal arts major working behind the counter.

"We felt that book discovery could be a lot better than what it is today," Jared Friedman, Scribd's co-founder and chief technology officer, told me as he showcased the site's new features. "The rise of subscription services gives us the opportunity to raise the state of the art for book discovery today."

How Scribd helps you find books

Instead of using Book Industry Standards and Communications codes to catalog its available books, Scribd has an in-house editorial team adding tags of its own that cover everything from the book's subject and style to where it's set and when it was written. The idea, Friedman says, is combine these editorial insights with algorithms to make recommendations that take a subscriber's reading history into account.

"We have 500,000 books in the Scribd library, so it's really important we organize that library well to connect the right reader with the right books," he added.

Among the things you can expect to see at Scribd are topical collections based on the tags assigned by Scribd's editorial team, top books and authors, and handpicked favorites by Scribd staff. That latter feature looks to mimic the process of shopping at a book store, right down to the notes from Scribd's editors as to why they recommend a particular book.

In my walkthrough of Scribd's new recommendation service, we took my love of biographies in general and my interest in Stephen Ambrose's biography of Dwight Eisenhower in particular along with the fact that I've just read a book about unusual maps to produce book recommendations. Scribd suggested I read The Journals of Lewis & Clark or How the States Got Their Shape to scratch my history itch. The Eisenhower connection led me to a staff-curated collection featuring 26 hand-picked books about assorted presidents.

Existing Scribd users who have a history of using the subscription reading service will obviously get more tailored recommendations right now than newbies to the service. Still, Friedman says, new Scribd users will be able to benefit from editorial recommendations as well as popular reads on the site. "The system definitely gets better the more you use it," Friedman said.

Increased competition

Scribd rolls out its souped-up discovery features at an interesting time, with Amazon--the 800-pound gorilla of the book retail world--muscling in on Scribd's market with its own $10-a-month subscription ebook service. But don't assume Thursday's announcement is some sort of hastily-assembled response to Amazon's newfound interest in ebook subscriptions. Friedman says Scribd has been working on this revamp for the past year--and plans to continue perfecting its approach to book discovery.

After all, Scribd has more practical motivations beyond staying one step ahead of Amazon for making it easier for its subscribers to find something worth reading. "Everyone who sells books wants to help you find new books," Friedman said.

Personalizing recommendations has been a tough nut to crack for the companies that sell digital goods. After all, few companies have had enjoyed more success at selling music, movies, and apps than Apple, and yet, that company's recommendation tools haven't really evolved beyond suggesting similar offerings that other customers have bought or pointing users toward lists of top purchases and downloads (And the less said about efforts like Ping, the better.) Other online retailers of various shapes and sizes can't really boast of handling discovery any better.

So has Scribd stumbled upon the magic formula for helping you find something new to read? It's hard to say from seeing one demo, though the initial debut looks promising. The proof will come as users get a chance to put Scribd's book discovery efforts to the test.