She’d been abandoned repeatedly this year. Her fiancé. Her parents. Her best friend. Even her landlord had effectively pushed her out. And now, the cracks were starting to show, visible to anyone who cared enough to look.
I typed:Seeks partner who is reliable, emotionally present, and committed for the long term.
My finger hovered over the “enter” key.
This was a massive violation of privacy. I knew this beyond a shadow of a doubt.
And despite that—despite knowing I’d regret this in the morning, despite every ethical principle I claimed to hold—I pressed enter anyway.
The progress bar appeared, creeping forward slowly as it analyzed Holly’s data against my own profile.
“This is stupid,” I muttered. “This is?—”
The results appeared on my screen, an astounding 98 percent compatibility.
I blinked and leaned forward—as if those six inches would change what I was seeing.
It didn’t.
Ninety-eight fucking percent.
That wasn’t a number the algorithm regularly spit out. No. It was one that was rarely—if ever—seen.
It must be broken. Or I’d entered the information incorrectly.
I checked the data inputs and then re-checked them before running the program a second time.
The number changed.
Holy shit.
Holly and I weren’t 98 percent compatible. We were a99 percent match!
A slow, dizzy feeling rolled through me, a mix of awe and nausea.
The highest match the app had ever produced—ever—was 97 percent, and those two were married with twins and a third kid on the way.
Ninety-nine wasn’t just rare. It was literally unheard of.
My hands were shaking, and my heart was hammering so hard I could feel it in my throat, behind my eyes, in the tips of my fingers.
Ninety-nine percent.
I’d built this algorithm. I knew exactly what that number meant. It meant compatible life goals. Similar communication styles. Complementary personality traits. It meant that statistically, if we got together, we were endgame.
Beyond that, it meant that I wasn’t crazy for feeling this way about her so quickly. It justified the way I’d taken one look at her and thought, “mine.”
But what if I’d made an error? What if I’d unconsciously skewed the inputs because I wanted a specific result? Confirmation bias was real, even among people who built systems designed to eliminate it.
I ran the calculation a third time with slightly different parameters—more conservative estimates of her values, less optimistic interpretations of her social media posts.
Our score dropped back down to 98 percent.
The algorithm didn’t lie. It couldn’t.
My pulse kicked hard, too loud in my ears. I pushed away from the desk and stood, crossing the room. My office suddenly felt too small, the air thick and suffocating.
I scrubbed both hands through my hair, grabbed my glass, and paced to the window. Snow had started to fall again—fat flakes illuminated by the lamp on the corner of my property that drifted lazily to the ground in soft, slow spirals.
