January 27, 2013 at 09:00 AM EST
Facebook’s Categorial Imperative
Facebook's newest feature adds some much-needed relevance to the huge proportion of its data hoard that no user has seen or, if we're honest, thought about, in days, weeks, or years. But Graph Search is ultimately nothing more than a handy sorting algorithm, and it's indicative of the fact that really, Facebook doesn't understand the first thing about us.
Phi mask in FB blue

Facebook’s newest feature adds some much-needed relevance to the huge proportion of its data hoard that no user has seen or, if we’re honest, thought about, in days, weeks, or years. But Graph Search is ultimately nothing more than a handy sorting algorithm, and it’s indicative of the fact that really, Facebook doesn’t understand the first thing about us.

(As a quick excusatory aside, this doesn’t mean I think it’s useless, of course: The tool they’ve created will make it a snap to find that one picture of your friend on that camping trip in July of 2009 (or was it 2010)? And “Friends in Seattle who like Poker” is a great way to set up a card game. These problems, trifling though they may be, are solved. Also, a “sorting algorithm” is not in any way inherently bad, and many useful and powerful things can be described as such, so no slight intended there either.)

The idea struck me when they were demonstrating how to drill down in search: “Friends of friends in California who read Harry Potter, like mountain biking, and speak Spanish and English.” Leaving aside the questionable utility of such a finely-pointed query, it drove home the fact that Facebook’s conception of each of its users is an endless series of nested categories. Zuckerberg’s joke slide showing a galaxy of pull-down filter boxes was more revealing than they let on.

An individual, to Facebook, is the sum of their interactions with the site — can never be more. You are a collection of data, beginning as an empty vessel when you sign up, and gradually growing in complexity and depth. This much is self-evident.

Facebook has a categorial imperative: Its reason for being may be to provide a service, but its means for being is to systematize individuality.

—Let me unpack that.

The measure of a man

Everything you do on the social network contributes, in one way or the other, to your being categorized ever more specifically. The categories to which you belong may be obvious (male/female) or subtle (interpretation of “like”), and there are more or less depending on how much you interact with Facebook.

If I decline to explicitly state my political preferences (or it simply never occurs to me), I’ll be excluded from searches like “Friends who are Democrats.” Yet, surely it’s obvious from circumstantial evidence — my liking a story about Obama, my linking to a left-leaning charity — that I am one. Facebook may be able to calculate the likelihood of my voting on the left side of the spectrum, but its model is crude, correlative. And of course it doesn’t dare to suppose, for a number of reasons.

Certainly such a conclusion is obvious to you and me. We humans (let it be said without boasting) are the most finely-tuned detectors of social tendencies, and the most sophisticated engines for predicting behavior, that have ever developed or been developed.

And is it not the goal of a system like Facebook, which probably observes ten lifetimes worth of human behavior every day, to ape that behavior, so to speak? And may it not do so successfully?

I have my doubts. When you ask an engine of social knowledge for friends who like Indian food, and it returns all your friends who have liked the “Indian food” page or concept or whatever it is, that’s telling. Expecting it to do more would be a little crazy, since as I said, it’s just a sorting tool. What’s telling is not that there are limitations, but that you bump up against them instantly.

Knowing a person, and the pieces that make them, isn’t a hard thing to do, exactly. You know hundreds, probably thousands of people. There are extensive records in your head, going back years: pictures, words, smells, ideas, and all. But you — you incredible machine, you — have interpolated all of that information into a single idea, a wavefront informed by every word and look you’ve ever exchanged with someone. It’s one of the most sophisticated things humans do, one of the most effortless, and also one of the most inimitable.

I don’t mean to place the burden of totally simulating human social consciousness on Facebook. What I’m saying is that Facebook’s self-imposed limitations, as well as the limitations of a system to which all information is submitted voluntarily (as opposed to unconsciously), mean that all it can ever do is resurface information that you yourself decided not to internalize — probably for good reason.

Graph Search lets you find things you already had and, apparently, didn’t care enough about to record them yourself. You can’t ask it for friends who are free tonight, or friends who have suffered from depression, or friends who have good taste in books. You could make a better guess on these things — these searches that matter outside of Facebook, which is to say searches that matter.

Recall that the value we get out of Facebook is in the new things, not all the rest of the stuff. New things like friends’ photos and status updates, or news from the sites and brands we like. We dip our fingers into the river and the next day, when we return to the bank, where is the river in which we dipped our fingers? Facebook knows, and Facebook can tell you. Why? What use is a map of yesterday’s river?

Set theory

Certainly we asked for a way to track down that errant photo or acquaintance in the city you were visiting. And while what Facebook gave us addresses these needs, it irreversibly reveals how superficial is its understanding of people. People as categories. Nested categories. Demographics.

As others have no doubt already pointed out, Graph Search resembles the drill-down demographic targeting of modern advertising. Facebook already had a nice big bucket labeled “males age 18-35 who like Baseball and Apple.” Put “Friends who are” in front of it doesn’t automatically make it a social tool for ordinary people (though it can be put to use). Facebook’s fundamental system of defining people is not the way people define each other. It’s the way people are reduced to sets of high-priority characteristics — it’s profiling, for tracking, targeting, and marketing. Facebook thinks this way because this was the way it was built to think — this is its categorial nature and its categorial imperative.

A social network that operates on non-human principles can’t ever be anything but a ledger, crammed with cold observations. And while some may think that this sterile data can be tracked, shuffled, recombined, and interpreted, alchemically, turning the silt of the river into gold, I disagree.

What emerges from these kabbalistic manipulations is useful to advertisers, of course, and also (once they are allowed to get their hands on it properly) academics like sociologists (and they are both welcome to it, I say), but for individuals it will rarely amount to more than a momentary diversion, or a modicum of convenience — not that such things aren’t at a premium these days as well.

In other words, any value provided by Facebook over and above the service itself will be rooted in the mechanical reduction of human data, and useful primarily to those who think in categories rather than individuals. You know the type.

Facebook is an amazing service and tool, and deserves respect both as a trailblazer and as a bricklayer, putting down first the notion and then the fact of a world-spanning social network. But it’s also ready to be replaced with something just as ambitious but far more personal.

A social application that knows a person the way that person knows another is not a trivial advance. It does not exist now and likely will not for quite some time. Facebook is not it, nor is Google’s shadowy profiling of every passerby, nor any other community or group on the web or emerging today.

Yet if people are going to use computers and the Internet (by any other name) to connect with one another, then this is the destination. Perhaps it is the boundary, towards which progress bends asymptotically, but I think not. Computers don’t think the way we think, they don’t see how we see or remember how we remember. But these are largely technical hurdles, tall enough that they look like insurmountable barriers — so we spread out along them until some long-legged visionary shows us the way over. It’s happened a hundred times before, and it’s happening in another hundred ways right now. Wait for it — ask for it — or build it.


Stock Market XML and JSON Data API provided by FinancialContent Services, Inc.
Nasdaq quotes delayed at least 15 minutes, all others at least 20 minutes.
Markets are closed on certain holidays. Stock Market Holiday List
By accessing this page, you agree to the following
Privacy Policy and Terms and Conditions.
Press Release Service provided by PRConnect.
Stock quotes supplied by Six Financial
Postage Rates Bots go here