Stories in Stories, Part II

Now that the Netflix Prize story has subsided from the mainstream news media, I think it warrants revisiting.  I wrote on this story about a month ago in this post.  At that time, I suggested that the mainstream media had buried two leads of the story.

To summarize, the goal of the contest is to improve the Netflix collaborative filtering engine by 10%.  For this, Netflix will pay $1M.  Collaborative filtering technology enables Netflix to predict whether a customer will enjoy a movie based on how much they liked or disliked other movies and also their rental histories.  Netflix uses these predictions to drive consumers down the long tail of available titles to both increase profitability and customer satisfaction.  Cake and consumption, all in one.  This is worth a lot to Netflix, but the problem is tough to solve.  Hence the contest with a sizeable bounty.

Story #1:  A valuable and captive pool of talent

As of this post, there are now 14,280 contestants on 12,000 teams from 115 different countries competing for the $1M prize.  This is over 50% more participants then when I reported on this topic 30 days ago.  From a recruiting perspective, I remain thoroughly impressed with Netflix's capabilities in 'name generation.'  I continue to wonder how Netflix is choosing to engage with this captive talent pool of software developers.  I suspect that this asset is being under-utilized and under-valued.  Clearly, any future hairy technical challenges that the company might face have become decidedly less hairy, knowing they have this resource pool to draw upon.

Story #2:  The Solution > $1M

The discrete goal that the company wants solved is a 10% improvement in recommendations, but they have paid out "progress prizes" of 50k apiece to 6 teams in the last month.  You can check out their leaderboard for more details. 

What's most remarkable is that the leading teams have nearly reached the halfway mark in less than 1 month.   Clearly the 2nd 5% improvement will be far tougher to solve than the first 5%, but a materially better solution in 30 days is remarkable. 

The top 6 teams have delivered a 4% or greater improvement in the recommendation results.  The economic value of these solutions to the broader B to C marketplace is worth far greater than the $1M that Netflix is offering to the winner.  Consider that the silver and bronze medalists (and others with viable but less optimal solutions) will certainly offer these solutions to other B to C online retailers.  Net result:  the worldwide overall ecommerce engine that relies on collaborative filtering becomes more efficient.  The value creations for consumers and retailers alike is huge.

I continue to believe that we HR people will look back on this as an inflection point in the paradigm for how talent is deployed to solve problems.  This is a story worth following.

Finally, I couldn't help notice that Netflix has a Sr. Staffing Manager job open today, but that the focus seems to be on hourly staffing (not technical jobs). 

Perhaps they need to run another contest...

Comments

[DR]

You say: "they have paid out "progress prizes" of 50k apiece to 6 teams in the last month".

Where did you get this information from?

You are totally wrong.

A SINGLE progress prize of $50,000 will be given annually on October 2nd each year until 2011 for the best improvement of the last year, which is at least 1% better than the previous year, unless there is a grand prize winner (total of 10% improvement), in which case the contest ends.

Read the Rules on that site.