Tuesday, January 22, 2013

Facebook’s Graph Search

Have you heard of Facebook’s Graph Search? You should. The buzz has been everywhere these few days. Wired Magazine has an article on it, but it’s a little too much of a cheerleader for Facebook for me. Still it gives a good overview of Graph Search.

Forbes had a more nuanced article, found here. The Forbes article references a short Gizmodo article, also worth reading, linked here. Techcrunch has a decidedly negative take on Graph Search. Their article can be read here.

I am not a heavy user of Facebook, despite being an early adopter (I joined in 2005, when I was a senior at a university on the East Coast.)  I am what you might call an “unenthused” poster of links, videos, photos and (god forbid) status updates. No Timeline for me, thanks.

[As an aside, I am also a smartphone holdout, despite being a bit of a geek. I carry a super basic Nokia phone without even a camera. I've long maintained that the deficit today is not of technology or connectivity, but rather of attention. Everyone's attention is spread way too thin these days.] 

Now, with the unprecedented “discoverability” of personal data enabled by Graph Search, I think I will be even less inclined to post stuff online. This is George Costanza’s fear of worlds colliding, writ even larger than what people have already experienced. Any user of Facebook will be able to query the database of user-generated data stored on Facebook and interpret that data in ways that you probably didn’t anticipate when you ‘liked’ something, posted a comment, or were tagged at an event.

I’m not even sure if Graph Search will be a useful feature for most users. I certainly don’t think I would use it, except perhaps for cyberstalking (there, I’ve said it. Not that that’s considered deviant behavior these days; all of us online have PhDs in cyberstalking).

The reason why I don’t think Graph Search will be useful to me is because I rarely have the habit of asking my friends’ opinions when I need to buy something, hire a service, watch a movie, read a book, play a computer game etc, all hooks that Facebook can use to monetize its database of personal information. This is because I search for information independently, and I have a very good idea of what I like and what to look for.

Even for people who value their friends’ opinions and would want Facebook to give them the results of an instant poll of their friends (based on friends’ past ‘likes’ and behavior), most people wouldn’t base their decision making on the opinions of ALL of their friends. I certainly wouldn’t trust the opinion of my friends on certain matters.

You wouldn’t ask a non-foodie about her favorite makan places, would you? And you would probably have zero use for another friend’s tastes in music and books, if for example, you privately think that he’s a Neanderthal (but you love him anyway).

And this is assuming Facebook users religiously post all of their opinions and ‘likes’ which are immutable and unchanging and completely context-independent. We are not even going to go into the thorny thicket of problems that have to do with users being facetious, or ironic or just plain obfuscating.

Or the related problem that even for someone who assiduously prunes his Friends List, and who resists adding new Friends unless he has known them a while, one third of the people on my list are people I don’t actually know…and another one third are people I have not spoken to in years.

To put it simply, Facebook data on is not clean. The signal to noise ratio is abominable, and careful interpretation is needed to make sense of the data to generate actionable insights. That sounds like an awful lot of work to get a restaurant recommendation near Ion Orchard on a Friday night, one that you agree with at the end of your dinner and that you could not have just as easily pulled from hungrygowhere.

More likely than not, the results of Graph Search are going to be dominated by the people on your Friends’ List that are perpetually online, post status updates incessantly, and have the attention span of hamsters. On steroids…and coffee.

No comments: