If you've logged into your Facebook or Twitter accounts in the past two weeks, you have probably seen at least one – or more likely, six or seven – posts from an app called What Would I Say?. Simply put, it's a little mechanism that, when you give it permission, processes every status, photo caption, and comment you've ever posted to your own Facebook timeline and spits out a randomly generated status that resembles something "you would say."
The app's popularity exploded within days of its creation at HackPrinceton, Princeton University's annual hackathon, during the second weekend in November. The team behind it – a cabal of nine grad students whose usual areas of research range from computational biology to computer science to statistics – tells WIRED that in the two weeks since its inception, over 2.2 million of the bot's faux-statuses have been posted on Facebook.
Of course, What Would I Say? is certainly not the first (nor is it the last, probably) of its kind: From That Can Be My Next Tweet to @Tofu_product, there are a variety of text generators ready to plumb the digital depths of our online presences and create bizarre, uncanny valley reflections. And there's a reason we enjoy these reflective tokens, even at the potential risk of our privacy; they act as tiny digital mirrors that reflect -- and confirm -- our identities and our place in the world.
"Bots like this show you that you exist," says social media theorist and sociologist Nathan Jurgenson, who studies the interactions between our digital and IRL selves. (He's also pretty well known for his job as Snapchat's in-house sociologist.) "You’ve posted all these status updates, they really did matter, they haven’t gone away, they were recorded, and they say something about you. It’s the same thing people said when Friendster came around: We want proof that we exist."
Jurgenson says that's the most basic impulse behind our desire to engage with and share these reflective tokens, but the reason they actually succeed in entertaining us – why we find these statuses hilarious enough to share with our friends – is a little more complex. According to Jurgenson and other social theorists, "apps" like What Would I Say? (which isn't really an app, but it interacts with Facebook's API as such) mimic human behavior, but not quite perfectly, which also makes it less unnerving.
"It’s an uncanny valley situation, where [the app] reflects the self, but not too well, and not too poorly," he says. "It’s enough of you that you recognize yourself, but it’s a distorted-enough reflection where it’s not creepy."
Most update generator apps like this operate in similar ways, using human data and mathematical probabilities to construct a Frankenstein's monster of language that toes the line between too real and not real enough. What Would I Say?, for instance, combines Facebook's open API with a bot algorithm called a Markov chain: an equation that uses probability to predict sentence structures. Essentially, it takes the common phrases you use and, according to the probability of you using them, and randomizes them based on the way your words typically follow each other. It's not adhering to the official rules of grammar you learned in school — instead, it's simulating grammar based on how you actually use it. (That's why it sounds funnier and more human than, say, Siri.)
Jurgenson also says that generators like What Would I Say? and That Can Be My Next Tweet are entertaining to us for another more metatextual reason: because they mimic the language patterns of bots (or supposed bots) like @horse_ebooks, a distinctive form of speech that also provides much of the humor in the subcultural comedy ring known as Weird Twitter. "It’s this new vernacular of the internet, where you have Weird Twitter and bots that tweet these non-sequiturs and you try to find meaning in these weird phrases," says Jurgenson. "It becomes a new internet linguistics, and now we get to participate in that: This bot took our own words and put them in that new voice."
Of course, these tokens, the ones that ask us to submit our information for fun, often come with a particular level of uncertainty. Though the What Would I Say? app doesn't store data – as the team writes on its about page, there is no database, and "all computations are done client side, so only your browser ever sees your post history" – some are not as explicit in how they use (or worse, keep) data.
But we log in anyway, and take it on faith that we won't be screwed in the process. (I held out for only a few days, and then curiosity got the better of me.) British sociologist Anthony Giddens called this our "bargain with modernity": the choice we make to accept a technology, be it a commercial airplane or social media platform, without knowing exactly how it works or whether it will hurt us, in exchange for the social benefits it provides.
"We do this with identity all the time," says Jurgenson. "Social media is a massive identity-based bargain with modernity, where we’re like, 'All right, bot, I’m just going to give you all my information because it’s going to be fun, and social.' You don’t want to think about it that much, because what would that mean, if you actually didn’t participate in social media fun, these social games and in-jokes, because you were always worried about privacy? You'd miss out on a lot."
