It's not all trash, of course. But there's enough trash in The Atlantic that makes me not want to bother sifting through it on a monthly basis. But this month's issue had a kinda thought-provoking piece on online dating data, the kind being collected by OkCupid as we speak. More and more people are pouring out things into an online dating website than they'd ever be comfortable sharing in public (or private). OkCupid, for example, has a sign-up quiz which is used to match you with other users on the site on the basis of your dating profile. The quiz asks you pretty intimate stuff that other dating websites sort of ignore, such as how many partners have you had relations with, do you have a STI, are you into kinky shit, etc. In addition to volunteered information collected by quizzes, the site is constantly collecting more "passive" usage statistics such as search queries, how long you visit profiles, what pictures you're looking at, who you're messaging (and who you're not messaging), and how you respond to instant messages.
As a social scientist, I can tell you that people are freely giving up what we researchers pay literally millions of dollars a year to collect. As someone who is primed as an academic to think constantly about protecting the rights and privacy of the people I study and who is in many ways constrained by the weight of just how far we must go to protect confidentiality, it's always amazing to me that sites like Google, Facebook and OkCupid just have this incredibly intimate data basically just compiling itself. People sign away a lot for the privilege of using these sites and it's actually at the benevolence of the site founders that their privacy isn't even more encroached, honestly, since we know most people have no idea how their data is used or what, exactly, they're signing away when they sign up for a site or click through a waiver. But still, from where I sit professionally, this is amazing stuff that we social scientists are only just starting to wrap our heads around. Most of us aren't trained to understand it as data yet and right now that's basically the only thing holding us back from moving further into telling you, for example, how your search queries predict on average how many more years you'd have of dating before you marry your first spouse, when compared to other users who match your dating profile. How terrifying is that? Marketers and academics do this kind of shit with the expensive and limited data we have at our disposal now. Imagine if we had more of it.
Personally, I'm fascinated with OkTrends which is at least some attempt to make use of the ridiculous amount of data the site is collecting and convey it back to users in a way that is both approachable and informative. In fact, quite recently, OkCupid's team of datacrunchers came out with a pretty neat finding showing that gay men, for example, virtually never try to search for or contact straight users and are no more sexually promiscuous than are straight users. These kind of findings are not only interesting, but they are changing how we understand social behavior (while also debunking stupid myths that have never had data behind them one way or another).
The piece in the Atlantic brought up one way for understanding how this exponentially increasing store of data can have negative consequences. Algorithms are used by OkCupid to better set you up with a potential match. Most dating sites do this on a pretty superficial level. If you're devoutly Christian, most are going to veer you away from singles who list Atheism as their religious view. OkCupid is a little more advanced in that it's using far, far more data to construct its algorithms than Match.com. It's not perfect, not by a long shot, but it is moving in a direction where it knows better than you do what you want to find in a date. It knows how others whose site activity looks like yours are doing in their dating trajectory and it's going to start using algorithms to shape your site usage to ensure you a slightly better batting average.
What does that really mean, exactly? OkCupid site founder, Sam Yagan, notes that the sites' main goal is largely just to get people get a conversation going with other users. It wants you to message and connect with another user. Whether or not you have a successful date out in the real world is not really something it considers because the site itself can't control real world interactions (yet). But it can try to tailor the site to ensure that more of its users are connecting with other users who will be receptive. One of the problems OkCupid's data team has found is that black women, for instance, have a really hard time getting men of all races to reply to their messages.
Now comes the interesting stuff: the moral questioning. Does Yagan start using algorithms to limit the users black women can see to mostly only those men--based on statistical evidence--who are most likely to respond to messages from black women to better ensure a successful match? If he did, to what extent would users even be aware that it was even going on?
It gets more complex when you start thinking about other variables of interest and what responsibilities a dating site has to its user base. Online dating is getting more and more sophisticated and the questions sites ask are getting more probing. You are torn, as a user, from revealing the data because the reason you're on this site in the first place is to find a match; if you withhold data, are you hurting your chances of being matched with someone perfect for you?
Even though most of the site's quizzes are useless distractions telling you which Harry Potter character you're most like, some of them are actually not too dissimilar from instruments we use as social scientists to screen people for things as important as depression, suicide, and domestic abuse. From the site founder's perspective, if you have questions being answered on your site that can serve as predictors of mental instability and social dysfunction, do you have a moral obligation as a dating site to weed those users out of the dating pool? What about infidelity? Do you steer cheaters towards other cheaters? Or what about health predictors such as eating and exercise habits? Do you factor in predictors for impotence, obesity, diabetes, cancer or heart disease into your matching calculus? What about social class? Do you match the rich with the rich? Poor with the poor? Again, what are your obligations once you have that data? Do you do anything? What can you do? What's to stop you from doing anything?
I think these questions are important to ask because we know more and more people in America are turning to the internet for dating. Because online dating presents far more data than you could ever process on your own and because more and more people are signing on, algorithms are increasingly necessary to help you find a compatible partner. It's difficult to think websites such as Facebook and OkCupid might have this much influence in shaping social networks, but the reality is they already do. So long as we enjoy this data for our personal consumptive needs, it's still important for us to start thinking critically about what implications this could have on the shape of contemporary society.
0 comments:
Post a Comment