The Social Graph

A few weeks back there was a lot chatter from the usual talking heads about Facebook's new graph technology-related announcement. As usual, I have not heard anything about the immense potential of graph, which is more or less the core of facebook's immense technology potential. My sense is that if I was Oracle, I would immensely concerned about the innovations in data technology of Google and Facebook. For those who know how to write code, I would suggest you play a little with the graph APIs, most of which have been available for many years. It is also a technical read, but a really fascinating in-depth dive into Facebook technology: the social graph.

I remember talking about the value of Graph in June 2011 to a banker analyst friend who had a sell rating on the company's stock. Obviously I was right ... Here is also an interesting point of view that ties up Graph with big data ... Let's just say that some are slower than others ;-)

Why Facebook's Graph Search Really Does Matter: Big Data + NLP
[Reproduced from Forbes]

By Marrying big data with natural language, Mark Zuckerberg will know more about you than your closest friends.  Ever since Facebook unveiled its Graph Search last month, pundits have opined that it’s everything from a “killer app” that will crush companies ranging from Google to Yelp to a powerful new ad targeting technology to nothing more than a glorified extension of the “like.”  In reality, it is none of these things.

Facebook, after all, is a generalized social network and can’t hope to compete with Google’s long commitment to, and expertise in, search, nor can it hope to unseat specialized services like Yelp.  I very much doubt it will bring measurable lift to ad targeting efficiency (social activity is just one of a plethora of behaviors tracked) and it certainly is much more than an extension of the “like.”

What Facebook Graph search really signifies is that it is entering the race to marry natural language processing with big data.  The list is getting to be a long one and now includes:

Apple Siri:  Probably the simplest (but most popular) version is Apple’s Siri which is available on iPhones and iPads.  It’s still a bit buggy, but works reasonably well if you speak slowly and clearly.

Microsoft Kinect:  While best known as a gesture interface for the Xbox, Kinect also takes voice commands in natural language.  Microsoft has now  integrated Kinect with Windows 8 and, with a research budget of nearly $10 billion – most of it dedicated to cloud services – we can expect Kinect to become an impressive marriage of voice, gesture and data.

Google Now: Without much fanfare, Google has integrated its natural language processing platform directly into its searchbox.  Its Google Now service aims to not only search, but actually predict what you might want to know.

IBM Watson:  While it became famous for beating humans in the game show Jeopardy!, IBM’s Watson is being geared up to tackle industries ranging from medicine to finance.
So what’s really notable about Facebook Graph Search isn’t that it’s so
new and different, but that Facebook is willing to throw their hat in the ring against well entrenched rivals that are far better capitalized (Google spent a $1 billion on infrastructure just last quarter!).  This truly is the future of computing.

---

Who Is Searching Whom?

While the marriage of big data and natural language processing is exciting, it’s also frightening, because Facebook Graph Search and the other platforms aren’t just search algorithms, they are learning algorithms.  As technology historian George Dyson puts it, it’s not really clear how much we’re searching the search engines and how much they’re searching us.

When we contact a call center, we’ve become used to hearing a recording that says, ““This call may be recorded or monitored for quality and training purposes” and just assume that a supervisor might be listening in. However, it is increasingly machines that are analyzing our voice patterns and personal history.

That’s what the company Mattersight calls predictive behavioral routing and it can separate our personality into one of six categories that can determine how we will react to different approaches.  Similar methods can be used to monitor our corporate e-mail and phone traffic.

If that seems far fetched, Forbes earlier reported that the technology has gained traction among major corporations and has already been deployed to over 30,000 call center seats (and that was two years ago, the number is most likely far higher now).

Chances are, you’ve already been analyzed by Mattersight’s algorithms, probably several times.  Somewhere out there, a computer is getting to know you better than even those closest to you.

So, no, Facebook Graph search isn’t notable just because it can help you find a nice restaurant or a pleasant place to get a drink.  It marks another step into a future where corporations are able to peer inside our lives to a far greater extent than we would ever allow a government to do.

To be honest, I’m not sure how I feel about it.