Hoping to make its recommendation feature better, the DVD mail-renter Netflix, Inc. (NASDAQ: NFLX) is offering $1 million to the first person or group able to "reach a certain level of accuracy in recommending movie picks" based on customer personal preferences. Something I would absolutely not call a foolish waste.The company has released 100 million anonymous movie ratings to use as data points. The prize will go to whoever manages to develop an algorithm that can outperform currently-used Cinematch with at least 10% improved accuracy. Algorithms, also used by online retailers such as Amazon.com, Inc. (NASDAQ: AMZN) advanced rapidly for a time but have more or less plateaued.
Perhaps because of recent (or should I say, perennial?) data-release and net privacy dust-ups, Netflix made of point of stating that the 100 million movie ratings made available for prize-seekers have been stripped of all personal and account data. Hm, I wonder if that's the last we will hear about that?
However, I'm less interested in that than curious and anxious to see what kind of results they will get with this "wisdom of crowds" approach to R&D. Forming a "wise crowd", as defined by author James Surowieki requires four elements: diversity of opinion, independence, decentralization, and aggregation. This contest, assuming $1 million is a price sufficient to attract the interest of enough mathematicians, engineers, and programmers to generate a vast pool of ideas, has all four ingredients.
(via BetaNews)
Entrepreneur's Journal: Putting together a solid sales contract
The week in preview: Earnings from Walmart, Macy's, and other retailers
Six top trades for the rest of 2009
Reader Comments (Page 1 of 1)
10-02-2006 @ 8:39PM
Randy Smythe said...
$1 Million dollars and plenty of attention!
The winning $1 million dollar amount may actually be small in comparison to the publicity this exercise will generate both for Netflix and those inventive "geeks". Kudos to NetFlix for their marketing genius and they get a better mousetrap out of it also.
10-03-2006 @ 10:40PM
Cinematch Guru said...
The link from Cinematch to Cinetech is incorrect -- they are in no way related.
10-03-2006 @ 10:51PM
Michael Canfield said...
Thanks Guru.Changed the cinematch link.