At launch, we were careful to error on the side of caution and use a version of the BookLamp Engine that was optimized for a single purpose: provide book results for a preset number of titles, with no flexibility. Thankfully our servers handled the launch traffic without crashing, and we rolled out our more robust “live” engine over the weekend – this is the real engine behind the site. From the outside perspective, you won’t notice any changes. In fact, if we did our job correctly, you’ll generally notice no differences at all. The book suggestions will be exactly the same in this round.
The differences, though, are profound behind the curtain. Where before the engine was serving up results that – fundamentally – had been calculated beforehand and rolled out when needed, the new engine (our “real” engine) performs all the calculations on the fly. This means that the comparisons we use for determining which books are similar is now happening very quickly at the point you perform a search, not cached in advance.
Why is this important?
This is important because it will enable features such as “Play with DNA”, which is where you can use a base book to pre-populate the thematic make-up of a book, but then adjust the exact mix of those elements on the fly. In other words, you can say, “I’d like to create a bookshelf with books similar to The Lost World, but without the Monsters / Predators / Beasts, and WITH Vampires / Supernatural.” We won’t be able to promise that there is a book like that, but we’ll do our best to find the nearest one we have in our database.
We’ll be monitoring the performance of the system over the next week or so as we prepare for our next sets of features, which are going to focus on the ability to browse and discover without relying on a base book to begin with. We think this will help a great deal with the issue of not having as many books as we’d like out the gate.