Enter the 'Storyverse': Q & A with Small Demons

By Alex Dalenberg

March 29, 2013

In my other life as a business and tech writer, I spend a lot of time talking to, networking with and thinking about startups. Lately I’ve been pondering how the humanities and social sciences can harness the entrepreneurial energy coming out of the startup scene.

It’s going to be a heavy lift. A lot of ink-and-paper creatives and old school academy types are code-phobic. And then there is the not so small matter of getting paid.

But I was excited to see this Los Angeles-based literature startup: Small Demons, a literary search engine that allows users to discover new books through common themes, characters, locations and other elements. Small Demons calls these interconnecting links the "Storyverse". 

Think of it as a kind of Pandora for readers in that its algorithms attempt to help you discover new reads to fit your tastes.

Small Demons is starting out small. Founder Valla Vakili told The New York Times in February that the site’s revenue is negligible. But the company has inked content deals with five of the big six publishers, including Penguin.

What you’re seeing here is a fairly classic startup strategy: build up as big a user base as possible and hope it scales. In this case, my guess is that would be through a mix of ads and commissions on referrals from the site. 

At any rate, we caught up with Small Demons founder Valla Vakili via email. Here’s what he had to say.

How does Small Demons differ from book recommendations generated by Amazon and other ecommerce platforms?

Amazon and other ecommerce platforms typically draw recommendations from behavior— what you and other users like you have bought, what you've rated, your past viewing history, things like that. On Small Demons, context drives discovery. Whether a user is a fan of a book or author, or comes to the site via another interest point, we use that starting point—your favorite book, author, city, band, or more — to show you all the books connected to it. The data we draw on at Small Demons comes from the book itself — mining the text for references to other cultural products, and making those easily accessible to users.

Where does all the data come from? Is It just a matter of searching the text?

We index the full text of books to search for references for interesting people, places and things (ex., music, movies, fashion, food). We use the context that the reference appears in to surface the person, place or thing on our site — you'll see for every reference mentioned in a book, we have a snippet of text associated with it to provide its context. We've done a lot of fine tuning to improve the accuracy of the indexing based on the categories that matter to us (for ex., media topics in things).

How much data about books actually is digital?

Tons of data about books is digital and tons more are becoming so. There are many efforts here — publishers converting their print books to ebooks; libraries digitizing their collections (worldwide), Google's scanning projects. So in a lot of ways the texts of books are becoming more and more digital. What lags is the metadata — good, consistent metadata that describes a book and its context. There are a lot of companies that focus on the first part of that—basic metadata that describes a book (author, title, year of publication, publisher). There are fewer who've made serious progress on the context side, and that's really our focus at Small Demons.