How Instagram's Algorithm Killed Visual Weirdness

In June 2016, Instagram replaced its chronological feed with an algorithmic one. The change was announced as a way to show users “the moments we believe you will care about the most,” prioritizing posts based on engagement signals rather than recency. Almost overnight, the visual texture of the platform began to shift. The glitchy, diaristic, and deliberately off-kilter aesthetics that had flourished in the chronological era—vaporwave collages, lo-fi self-portraits, surreal photomontages—started to recede, replaced by a more uniform visual language: high-contrast, aspirational lifestyle shots, cohesive color palettes, and the polished consistency now synonymous with influencer culture. The common defense of algorithmic curation is that it merely reflects what users want. If feeds are smoothing out, the argument goes, it’s because people prefer the familiar. This is a tidy story, but it mistakes a feedback loop for a mirror. Algorithms don’t just surface existing preferences; they reshape the incentive structure that determines what gets created in the first place. When a platform’s ranking system rewards content that reliably generates likes, shares, and time-on-screen, it creates a predictable gravity: the safest, most easily processed material rises to the top, while the strange, the ambiguous, and the experimental get buried. Creators, especially those who depend on the platform for visibility, adapt accordingly. What makes the June 2016 switch so instructive is that it was a single product decision, not a gradual cultural drift. Before that date, Instagram’s feed was a raw, time-ordered stream. You saw whatever the people you followed posted, in the order they posted it. That environment tolerated—and sometimes rewarded—visual risk-taking because there was no gatekeeper deciding what deserved attention. A photograph of a half-eaten meal, a deliberately overexposed selfie, or an abstract digital composition could sit next to a professionally lit portrait, and both had equal claim on your scroll. After the change, the algorithm began calculating which posts were most likely to hold your gaze, and the metrics it used—dwell time, tap-through rates, comment sentiment—tended to favor content that was immediately legible and emotionally easy to process. The weird stuff didn’t disappear, but it stopped circulating. This is not a nostalgic lament for a lost golden age. Algorithmic curation brought genuine benefits: it reduced information overload, helped users find relevant content amid the noise, and allowed the platform to scale without drowning everyone in a firehose. Acknowledging those gains is important, because the problem isn’t curation itself. It’s the specific optimization criteria that platforms choose—and the way those criteria, left unexamined, become a conformity engine. Research on algorithmic awareness supports this view. A study published in Frontiers in Psychology found that users who are conscious of how algorithms operate can experience “compliance and compulsive use,” particularly when they identify with the algorithmic information they receive. In other words, knowing that the system is watching and ranking your output can lead you to shape that output to please the system, often without realizing you’re doing it. The British Psychological Society has similarly documented how social media algorithms reinforce narrow aesthetic ideals—not by dictating them from above, but by creating an environment where deviation feels costly. For independent creators, artists, and writers, this dynamic is especially corrosive. The very idiosyncrasies that make work distinctive—an odd sense of humor, a messy compositional style, a refusal to stick to one palette or subject—are exactly the things the algorithm is least likely to amplify. Over time, the constant feedback of likes and reach numbers trains you to sand off those edges. You start posting what performs, not what surprises. And because everyone else is adapting to the same signals, the result is a monoculture of interchangeable posts that feel competent but rarely memorable. Reclaiming an edge, then, requires a two-part move: technical and psychological. On the technical side, creators can deliberately break the pattern that the algorithm expects. Instagram itself now offers a chronological “Following” feed, though it’s tucked away and not the default; using it can help you see—and be seen by—your actual network rather than the engagement-optimized version of it. More broadly, diversifying beyond a single platform reduces dependency on any one ranking system. Newsletters, RSS readers, and platforms like Tumblr or Are.na, which still operate on chronological or curator-driven logics, allow work to circulate without passing through an engagement filter. But the deeper shift is psychological. It involves revaluing the very weirdness that the algorithm trains you to suppress. This isn’t about ignoring metrics entirely—that’s rarely sustainable for someone whose livelihood depends on visibility. It’s about recognizing that the metric is a measure of one thing (short-term engagement) and not a verdict on the work’s worth. The historical counter-examples are instructive: many of the visual subcultures that defined early Instagram—think of the surreal, color-drenched worlds of artists like Arvida Byström or the deadpan, anti-glamour photography of Petra Collins—gained traction precisely because they didn’t look like everything else. They weren’t optimized for the algorithm; they were optimized for a specific sensibility, and that sensibility found an audience because the platform’s structure at the time allowed it to surface. The homogenization of visual culture isn’t an inevitable consequence of technology. It’s the result of a series of design choices, and the June 2016 feed change was one of the most consequential. If a single product decision could accelerate the smoothing, then a different set of decisions—by platforms that prioritize diversity of expression, and by creators who refuse to sand off their edges—can push back against it. The algorithm may reward the average, but the work that lasts rarely comes from the middle of the bell curve.