The mathematics of identifying niche behaviour and interests promises to bring like-minded people together. But it may be tearing society apart
Toronto Star | August 27, 2010 | Ryan Bigge
Last month in Slate.com, Mark Oppenheimer wrote about the latest trauma inflicted by the e-book revolution. It wasn’t lower prices turning novelists into ever more wretched paupers or copyright quandaries that make it impossible to share your favourite e-book with friends. No, the problem for Oppenheimer was that e-book readers make it impossible for randy bibliophiles to judge prospective lovers by their collection of book covers.
“As the Kindle and Nook march on, people’s reading choices will increasingly be hidden from view,” writes Oppenheimer. “We’ll go into people’s houses or squeeze next to them on the subway, and we’ll no longer be able to know them, or judge them, or love them, or reject them, based on the books they carry.”
His panic turned out to be premature, however. Two days before Oppenheimer published his lament, alikewise.com launched “a dating site that allows you to find people based on their book tastes.” Sadly, alikewise.com is not an anomaly, as specialty dating services are becoming increasingly common — Apple cult members have Cupidtino.com, indie rock fans visit Tastebuds.fm and followers of Ayn Rand’s Objectivism have theatlasphere.com. And while such sites should eventually serve to cull their respective populations through vicious inbreeding, they reflect a larger problem — the triumph of algorithmic niche culture.
As Devin Leonard explains in the August issue of Wired, Hunch.com is trying to personalize the Internet by soliciting people’s opinions, beliefs and tastes and then mining the data “for correlations that provide precisely tailored recommendations for each user.” Hunch.com is not alone, with similar services provided by GetGlue.com. The interest in these sites are obvious — there’s big money in artificial serendipity. In September of 2009, Netflix awarded a $1 million prize to a team of statisticians and computer engineers who created movie rental recommendation software that was 10 per cent more accurate.
The problem isn’t that the educated guesses of Netflix or Hunch.com are inaccurate — quite the opposite. But the mathematics behind the niche-ification of everything threatens to destroy the very fabric of democratic society. Or, at the very least, create some very nasty blog postings.
In a May article in The New York Review of Books about the Tea Party movement and the “politics of the libertarian mob,” Mark Lilla refers to Bill Bishop’s 2009 book The Big Sort: Why the Clustering of Like-Minded America is Tearing Us Apart. As Lilla notes, “People with higher degrees who care about food and wine, support gay rights, and want few children but good Internet connections have been gravitating to urban centres on the two coasts, while churchgoing families that drive everywhere, socialize with relatives, and send their kids to state universities have been heading to the growing exurbs of the southern and mountain states.” This, as you can imagine, causes problems for politicians trying to find consensus among an increasingly polarized electorate.
But according to Adam Sternbergh, things are not entirely bleak. In the January 2010 issue of New York magazine, he explained that we are still united by events like Avatar and the launch of the iPad, but the binding mechanism is now pre-announcement buzz and speculation. “Once we experience something en masse — or even as we experience it — we splinter off to our myriad forums to broadcast our personal takes.”
Another reason to avoid complete despair is that even a NASA supercomputer is unable to persuade us to enjoy certain algorithmically generated suggestions. Writing in the New York Times Magazine last October, Rob Walker explained how the Internet radio service Pandora was slicing songs into their atomic parts of enjoyability to better determine listener matches. The problems Pandora has encountered (music fans made irate by suggestions such as Celine Dion or Journey) will be familiar to anyone ever set up on a blind date by friends.
Which highlights the biggest blind spot of the algorithmic niche — its target audience is irrational, unpredictable, contradictory human beings. And not taking this into account appears to be the largest predictive failure of them all. If we can’t trust Pandora to pick a great song, it’s unlikely that alikewise.com can help us locate a soulmate.
As is so often the case, a popular sitcom provides necessary wisdom and perspective. During the first season of Modern Family, a recently remarried Jay explains to his 30-something son Mitchell that opposites not only do, but should, attract. “We’re both with people different from us and that’s gonna create stuff,” notes Jay. “But you want different. Your mom and I were perfect on paper and you know how that ended.”
It’s good advice, although undercut slightly by the episode’s conclusion, wherein Jay’s wife Gloria mistakenly believes her husband is about to marry a life-sized statue of a dog butler named Barkley. Then again, love is also blind. Which means it doesn’t matter if the boy or girl sitting across from you in the subway is reading Stieg Larsson or Gary Shteyngart. As long as they’re cute (and hate Journey) the complex and irrational numbers that comprise the algebra of love will take care of the rest.