Walmart has been losing market share to Amazon for years now. But the tables have turned with Walmart's Vudu, the brick-and-mortar retailer's surprisingly successful entrée into the movie download business. Since acquired by Walmart last year, Vudu's share of the online movie rental market has exploded from an unmentionable 1 percent to become the third most popular such service with 5.3 percent market share. (Bear in mind that Apple's iTunes owns 65 percent of the rental market leaving little pie for the competitors to split.) Amazon hardly grew at all.

The leapfrog represents the first big success of Walmart's new Silicon Valley-centric approach to growing their online business. Part of that strategy is the @WalmartLabs group, described by some as the "world's largest startup," and as the name implies, experimenting in the social sphere is the top priority. @WalmartLabs is more or less a new name for an old social media startup called Kosmix, which Walmart acquired this spring for $300 million. As Mike Cassidy of Silicon Valley's Mercury News explains, Kosmix mines social networks like Facebook and Twitter for information not only on what people are buying--which is the core of Amazon's recommendation engine--but also on what they're talking about:

That information could be used to steer shoppers to products based on their hobbies, for instance. Or the Kosmix technology could be used to analyze Twitter tweets in neighborhoods surrounding specific Walmart stores. That intelligence could help store managers decide on inventory. Should they expand their sporting goods department or maybe their video gaming offerings?

Oddly enough, Kosmix founders Venky Harinarayan and Anand Rajaraman worked at Amazon after the company acquired their e-commerce startup Junglee in 1998. Harinarayan and Rajaraman understand the strengths and weaknesses of Amazon's recommendation engine because they helped build it, and according to TechCrunch, the new engine they're building for Walmart will be based on their competitor's weaknesses:

At Amazon, explains Rajaraman, there's too much emphasis placed on your purchases for recommendations. "Purchases are a window into your interests," he says "but they're a small window." Anyone who has ever purchased a gift at Amazon only to be suggested similar items upon every subsequent login knows the problem. Instead of recommending more travel guides for Paris, long after your trip was complete, or more power tools, long after Dad's birthday has passed, Walmart plans to recommend things it actually knows you like.

Anticipating privacy concerns, @WalmartLabs will allow shoppers to opt into its socially powered products, but they'll also receive rewards in the form of discounts for their participation.

The first @WalmartLabs products are due out this season, and it will be interesting to see how they'll affect Vudu's success. It's not difficult to imagine how Vudu would use the @WalmartLabs technology. Like Amazon, Netflix recommends movies based on what you're watching, but it seems like a social movie recommendation engine could build upon that model nicely.