Which Algorithm Are You?
Photo Credit: Shutterstock.com/Gertan
Stay up to date with the latest headlines via email.
I love online quizzes.
The one I currently love most is “ How Millennial Are You?,” which the Pew Research Center put out on Friday. You answer 14 questions, ranging from whether living a very religious life is very important to you to whether you have a piercing in a place other than your earlobe, and your score tells you what generation you most resemble (Silent, born 1928-1945; Boomer, born 1946-1964; Gen Xer, born 1965-1980; Millennial, born 1981+). I turn out to be a typical Gen Xer, which lops a generation off my chronological age, and thanks to a “Modify Your Response” feature, I see that if I’d just cancel my landline and stop reading a daily newspaper, I could totally be a 20-something, except for the looking-in-the-mirror part.
The Pew quiz is based on data. It correlates your answers with a national sample polled by Princeton Survey Research. Also data-driven is “ How Y’all, Youse and You Guys Talk,” the New York Times’ most viewed and most emailed article in 2013, a quiz based on 350,000 responses to the Harvard Dialect Survey. Time.com had its biggest day ever when it posted “ How Much Time Have You Wasted on Facebook?,” an app that ingests the timestamps on every post in your Facebook feed.
On the other hand, Buzzfeed’s quizzes — its most popular, “ What State Do You Actually Belong In?,” has racked up more than 40 million views — are unmoored from data, unless you count as evidence the stereotypes from which its writers reverse-engineer the questions. Sites like Zimbio, which specializes in quizzes like “Which Character from [Modern Family, Vampire Diaries, South Park] Are You?” don’t even pretend to have an info component to their infotainment. They’re pure attention bait, designed to hook you, hook the friends you share them with, drive up web traffic and sell your eyeballs to advertisers.
But it doesn’t matter whether these quizzes are based on data or not. When you tell a site which of a dozen brands is your favorite fast-food chain and which name you’d choose for your baby, you’re adding new data, making big data bigger and enabling number crunchers to discover clusters and patterns no one had seen before. Technically, it’s child’s play to match up what you disclose on a quiz with whatever else you’ve disclosed to other data bases, from Twitter to car loan applications to retailer loyalty cards. Much of this information is commercially available: the terms of service you agree to without reading almost always permit selling your data to data brokers. Not only do you not get paid for this; you also make it possible for companies, government agencies and hackers to figure out who you are, often down to your name and address.
“Behavioral advertising” is the term for targeting consumers based on data they’ve provided, and I’ve surprised myself by kind of loving it. When Amazon’s algorithm tells me what books I might like, based on what other people who bought the books I bought also liked, I almost always discover something new that, in fact, I like. The same is true for video on Netflix, music on Slacker and (though less frequently) the products Google’s algorithm thinks I’ll buy by reading my Gmail, and the people Facebook’s algorithm thinks I’ll want to friend from network analysis of my peeps and posts.
I’m also glad that issue campaigns and political candidates can target ads and canvassers based on entertainment preferences, voter rolls and (conceivably, anyway) which puppy picture I think is cutest. If the best way to get the Senate to ratify a climate change treaty is to mobilize the voters most likely to punish their senators for siding with carbon polluters, I’m glad the data to do that exists.