It happened on a Friday night back in February, and I should have seen it coming. Linda was at her book club, I was too tired to do anything productive, and scrolling through Netflix felt easier than finding something else to occupy my brain. You know how it is when you’re just looking for background noise while you zone out on the couch.
I picked some Danish show because the picture looked moody – detective by the water, very atmospheric. Seemed harmless enough. Two episodes in, I’m snoring with the remote still in my hand, drooling on the throw pillow Linda bought last Christmas. When I woke up Saturday morning, my Netflix homepage looked like it belonged to a completely different person.
Apparently while I was unconscious, the algorithm decided I was obsessed with Scandinavian murder mysteries. “Because you watched The Killing,” it announced helpfully, then proceeded to show me about twenty different versions of detectives in turtlenecks standing over dead bodies. I mean, I work with databases and spreadsheets all day – I should understand how these recommendation systems work, but seeing it happen to me was something else entirely.
The logical part of my brain knew exactly what was happening. The system found one data point and ran with it, creating what the tech people call a “filter bubble.” Instead of showing me variety, it was doubling down on this assumption about my viewing preferences. Classic algorithmic overreach. But here’s the weird part – instead of being annoyed, I found myself thinking maybe the algorithm was onto something.
“Maybe I am the type of person who enjoys slow-burn crime dramas with terrible weather,” I thought to myself. It’s ridiculous when you think about it, but there was something almost flattering about the algorithm’s confidence in knowing what I wanted. Like it understood me better than I understood myself. So instead of clicking “not interested” on all those Nordic noir suggestions, I figured I’d give one more a try.
Big mistake. Huge.
Within three weeks, I’d watched four complete series from different Scandinavian countries. All grim, all involving complicated family secrets, all featuring detectives with serious personal problems. I started developing opinions about Swedish versus Danish subtitling styles. My coffee consumption went up. I found myself wearing more dark colors to the office, which my colleague Janet actually complimented me on.
When people asked what I was watching lately, I’d launch into explanations about the moral complexity of Nordic crime television. Me! The guy who spent the last decade watching nothing but cooking shows and the occasional action movie. I was suddenly acting like some kind of expert on international television, all because an algorithm made an assumption based on me falling asleep during one show.
That’s when it hit me how these recommendation engines actually change who we are. They don’t force us to watch anything, but they make certain choices so much easier than others that we end up following the path they’ve laid out. It was simpler to embrace being a Nordic noir fan than to fight against every suggestion Netflix threw at me.
The really embarrassing part? When I ran out of shows on Netflix, I started hunting down Finnish crime dramas on other streaming platforms. Platforms I’d never heard of before, with names I couldn’t pronounce. I joined online forums where people discussed lighting techniques in the second season of shows with unpronounceable titles. I started recognizing character actors across different series and feeling weirdly proud when I spotted someone from a previous show.
Even Amazon got in on the act. “People who watched Bordertown also bought these wool sweaters,” my shopping recommendations informed me, displaying an array of chunky knits perfect for brooding near fjords. I actually considered buying one until I remembered I live in Chicago, where I’d wear something like that maybe five days a year. The absurdity of it all was starting to sink in.
The strangest part was how quickly this algorithmic identity became part of my actual identity. When friends asked for TV recommendations, I’d confidently suggest the latest Scandinavian crime series, positioning myself as someone with sophisticated international viewing tastes. “I love the pacing,” I’d say with a straight face, when honestly, six months earlier my idea of good television was whatever cooking competition was trending.
This isn’t just a Netflix problem, obviously. Every platform we use has these recommendation systems now, and they’re getting scary good at predicting what we’ll engage with. Sometimes they nail it perfectly, other times they latch onto weak signals and amplify them into caricatures of our actual interests. My mother googled “garden hose” one time three years ago and still gets ads for increasingly elaborate irrigation systems. She lives in a condo with no yard.
What makes these systems so effective is how they gradually change the menu of options available to us. Once Netflix decides you’re into romantic comedies, it shows you more romantic comedies, which increases the likelihood you’ll watch one, which reinforces its theory about your preferences. Before you know it, you’re trapped in a feedback loop of the algorithm’s own making.
I understand the technical side of this stuff better than most people my age. In my younger days working with different software systems, I sat through plenty of meetings about recommendation algorithms and user behavior prediction. We’d discuss collaborative filtering and content-based recommendations like we were solving world hunger. The goal was always engagement – keeping users on the platform longer by serving them increasingly targeted content.
What we never talked about in those meetings was the psychological impact. How a temporary interest could get algorithmically amplified until it became a defining characteristic. Nobody speculated about how these engines might transform a Chicago accountant into someone who could hold a twenty-minute conversation about the visual style of Norwegian police procedurals.
The effect gets even weirder when the algorithm is partially right but in a funhouse mirror way. A couple years back, YouTube decided that because I watched two home repair videos when my kitchen faucet was leaking, I must be obsessed with extreme DIY projects. My homepage became filled with videos of people building underground houses with primitive tools and converting school buses into luxury apartments.
I have zero interest in living in a converted school bus. I’ve never wanted to build anything underground. But at two in the morning when I couldn’t sleep, I found myself weirdly captivated by a forty-minute video of some guy constructing an elaborate mud and bamboo swimming pool complete with filtration system. “That’s actually pretty clever,” I caught myself thinking, despite living in a rental apartment where I’d get evicted for digging up the parking lot.
These recommendation engines are insidious because they don’t feel like manipulation – they feel helpful. “Top Picks for Paul,” Netflix tells me, as if it’s doing me a favor by curating options I’ll supposedly enjoy. This creates the illusion that the filtering is neutral and beneficial, when actually it’s anything but.
What we’re dealing with is a form of soft coercion. The algorithm doesn’t force us to watch anything, but it shapes the environment in which we make our choices. It’s like having someone rearrange all the books in a library so that certain sections are easy to find while others are hidden in the basement. Technically you can still read whatever you want, but the deck is stacked toward specific outcomes.
This creates a weird new dimension to identity formation – one where we develop tastes and preferences in partnership with machine learning systems. I never chose to become an expert on Danish crime television, but when the algorithm assigned me that identity, it was easier to lean into it than resist. The path of least resistance led straight to the foggy streets of fictional Copenhagen.
It raises some pretty fundamental questions about autonomy and choice. If your streaming service influences what you watch, which influences your conversations and cultural references, which influences how you see yourself – who’s really curating your identity? When your preferences are shaped by algorithmic suggestions, are they still authentically yours? At what point does personalization become manipulation?
These days I try to deliberately confuse the algorithm as a small act of rebellion. I’ll watch a cooking show, then switch to a horror movie, then jump to a documentary about insects. “Try to categorize me now,” I think smugly, though I know these systems are sophisticated enough to handle contradictory data points.
Some platforms have started including “surprise me” features that inject randomness into recommendations, which I appreciate. It’s like having a circuit breaker for algorithmic assumptions. Sometimes these random suggestions introduce me to great content I never would have found otherwise. Other times they remind me why filtering exists in the first place.
Maybe there’s a sweet spot between algorithmic assistance and personal autonomy. A way to get helpful suggestions without being locked into increasingly narrow categories. Until someone figures that out, I’ll keep letting Netflix make wild guesses about my identity while I watch fictional Scandinavian detectives solve particularly grim murders in beautiful but hostile locations.
And honestly? If you need recommendations for Finnish crime thrillers featuring morally complex investigators with troubled personal lives set against gorgeous but unforgiving landscapes, I’m your guy now. The algorithm decided it, and who am I to argue with the machine?
Paul’s a Chicago accountant learning to survive in the cloud-software era. He writes candidly (and funnily) about being tech-competent but perpetually one update behind. His motto: technology is great—once someone explains where they’ve hidden the settings.
0 Comments