Page 26 of Swiped
The man stopped in front of him. His eyes were glassy and nestled in a thick grove of wrinkles. “Skin lice!” he cried, and fell onto Rami’s lap. He writhed for a second against his thighs as Rami’s mouth fell open in shock. Then the man grabbed the lollipop from Rami’s mouth, popped it into his own, and shoved off.
The bus jerked to an emergency stop as people cried out in the pain from smashed toes and elbowed ribs. The doors screeched open, and the man and his new lollipop jumped off. Rami’s mouth was still open as he watched him shuffle off intothe dark night, looking until it was just his own dumb reflection staring back at him. It was a familiar image. Because no matter what happened, it seemed like his nights always ended just like this. And maybe now he had skin lice? Rami pulled out his phone to google the possibility as the bus rolled on.
* * *
Nat sat alone in the flickering fluorescent light of the BART train. Her head throbbed with the sparkling wine and recent crying, and Jo and Eric and of it all. Thankfully, the car was mostly empty, except for a couple of teenagers making out in the accessible seats. The hum of the train through the tunnel was almost enough to cover the sounds of their smacking. Her headphones were in her other, larger, non-date-appropriate purse, of course.
Her phone buzzed. It was a text from Jo in their BeTwo group thread.
Jo:How’d it go, boss? We win this thing?
Nat scoffed as Jo sent a GIF of a woman anxiously chomping gum, as if Jo were actually worried about her. Justin “loved” the message, and Nat closed the thread. She was definitely not responding to that.
Instead, she opened a new message to Sara.
Nat:Whatcha doing?
Sara:Working on my résumé.
So fun!!!
How was the date?
Nat:EPIC FAIL!!!
Meet at the spot?
Sara:I was gonna try and finish this tonight . . .
I suck at proofreading even without a martini in me
Nat:Send it to me! I’ll polish it up for you
The train eased to a stop with a shrill squeak, and the doors opened with their pneumatic hiss. A loud group of yet more teens entered the car. Fantastic. One of them produced a portable speaker and began blaring an upbeat disco-inspired jam that made frequent use of an air horn sound. Even better.
Sara’s text buzzed back.
Sara:You’re the best!!!
OMW
See you there
Nat sighed and regarded the glass and metal rectangle glowing in her hand.Et tu?she wanted to ask. She opened her app. There was a crop of new messages and a bouncing heart in the menu bar, which meant that she had secret admirers to unlock, a feature that was available instantly behind a paywall or for free every twenty-four hours. But Nat knew her own algorithm, and so she knew that she had to swipe on as many profiles as possible in that crucial first forty-eight hours of creating an account in order to keep from getting buried in thedaily crop of new users, and also feed the algorithm enough data to get the most matches. She started swiping right on profiles, barely even taking in the pictures, let alone their names and pithy twenty-word bios, before the screen filled with a greenYESheart and flipped to the next man in the queue. Why not? The more the merrier.
She’d initially based BeTwo on a system called collaborative filtering, which was basically the same system that streaming services used to suggest what movies or music someone might like based on what they’ve already watched or listened to. If Bob liked smooth jazz, it showed Bob songs that other people who also like smooth jazz had listened to, thus making Bob feel satisfied and deeply understood by the magic of lifeless code.
Most dating apps lifted that model and stopped there. They sorted users based on broad categories like race, gender, job/income, educational level, age, and once someone matched with enough people of a certain type, it showed them that subset of profiles pretty much exclusively.
Of course, Nat had decided to honor the revolutionary insight that people usually liked more than one type of music, and so her version of collaborative filtering ended up being a lot more nuanced. Years-of-her-life-level nuanced, but it meant that BeTwo’s match suggestions were based on more factors than any other app that she knew of.
The teens kept blasting their jams. Nat kept blindly swiping right.
Something that had always bothered Nat was the fact that most dating apps’ collaborative filtering didn’t just group users based on the demographic categories, it also ranked them based on the majority preferences for each category — or in other words, it ranked users by popularity. Each and every user was given a score based on how likely they were to get right swipes compared to every other user on the app. So, when a userinputted their age, it was compared to data on how many “yes” swipes a user of a similar age had received from their potential matches. That meant that if the user pool skewed young, which they naturally tended to do, users over thirty-five were automatically ranked lower in the dating pool — even before taking other factors like ageism into account. Nat had felt that an algorithm reflecting that Bob likes smooth jazz, and so probably isn’t into heavy metal, was one thing. An algorithm reflecting that Bob likes twenty-five-year-olds and not thirty-five-year-olds, never mind the fact that Bob himself is forty-three years old, and then punishing those thirty-five-year-olds because of Bob’s preferences, was entirely different.
She’d labored to create a novel algorithmic approach because, as she’d learned through her many betas, most algorithms didn’t surface their low-ranking users nearly as often for matches, because most dating apps had a vested interest in making users feel that their match pool was brimming with the hottest singles around. In fact, many apps didn’t show low-rank users to higher-rank users at all. They just disappeared into the dregs of the user pool, and matched only with users that the system had deemed on-or-below their level of swipe-ability. In other words, it was a likeability contest where not coming in first place would essentially banish you to Siberia.
Nat had thought that approach was not only a moral outrage, but it was also lazy. It didn’t take a genius to see how this could do more than just perpetuate the worst biases of society — it would literally codify those biases with every single swipe. It was part of what she’d been trying to say when Tracy had brought up breaking toxic social tendencies in the panel discussion. That was exactly why her algorithm was so special. It was why it was worth protecting. It was why she had to beat Rami.
