Why Is Google Spending $10 Million on Fflick? Perhaps to Predict Box Office Success.
When I first read on TechCrunch that search giant Google is in the process of acquiring the movie-tweet analysis service Fflick, it triggered a memory that prompted me to start digging through my Gmail account. Once that digging was done I had found a year-old paper produced by two researchers at Hewlett-Packard that in turn led me to an interesting theory about one reason Google may be shelling out for this service, which at first glance looks like nothing more than one of dozens of consumer recommendation engines geared toward movies.
This research paper was produced by two social-computing researchers at HP Labs: Bernardo Huberman and Sitaram Asur. It’s titled “Predicting the Future With Social Media” [PDF here], and it looks at Twitter as a means of predicting the box-office success of newly release films based on the number of people tweeting about them and the sentiments contained in those tweets.
They argued that Twitter was a far better predictor of box-office success than the motion picture industry’s “tracking” reports that studios have used for years. In fact, the two researchers said at the time that Twitter could predict with nearly 98 percent accuracy whether a movie would be a hit or a flop in its first weekend of release. For the study, they mined nearly three million tweets referring to 24 different movies over a time period of three months.
Fflick does some sentiment analysis of its own, but uses that data to help Twitter users decide whether they are going to buy a ticket to a movie based on whether their Twitter friends liked it. Could it be that Google wants to mine that same sentiment data to help movie studios predict box-office sales?
As I said, this is only a theory–one that I admit I’m stretching to the max. I can’t find any connection between the two researchers and Ffflick’s four founders, or its investors, which includes the Founders Fund, though there needn’t be one for my theory to be close to the mark. Fflick was started in August of last year, about five months after the paper was published. And the paper itself was widely covered at the time, in particular by the Los Angeles Times.
Since neither Google nor Fflick is commenting on this deal, which is supposedly still pending, I thought it was worth suggesting as a possible motivation on Google’s part.
Electronista thinks it may have something to do with forecasting popularity on Google’s forthcoming YouTube movie project and the need to predict.
I did check in with the paper’s principal author, Huberman, by email to ask what he thought. His reply: “Sentiment analysis of tweets is great for marketing studies and Google wants to go there since they have search going on with Twitter.” Time will tell if this is what Google has on its mind.