Refreshing Our MemoryHow Algorithms Are Warping the Power of Recall

Digital drawing with a lilac background. A woman with long black hair stands on a purple road that extends into the horizon. Floating around her are snapshots of memories, including someone in a hospital, hanging out with friends, and a sunset.

Illustrations by Ariel Davis

This article was published in Touch Issue #93 | Spring 2022
Whoever is in charge of Snapchat’s yearly recap needs to be fired,” Mya Roschel, a high school student from Arizona, wrote on TikTok in late 2021. The visual that she deems, in all caps, “OUTTA POCKET” is a computer-generated graphic of a disembodied smiley face on a bright pink roller coaster, text above it reading, “You had a lot of adventures.” The picture that follows is of seven people, likely selected by Snapchat’s algorithm because of the number of faces and the unfamiliar location; what says “adventures” better than a group photo? The casket the group is adorning with white-and-blue flowers—Mya’s father’s—has presumably not registered.
 
In her 2021 book, The Disconnect: A Personal Journey Through the Internet, Roisin Kiberd considers her own fraught brushes with an algorithmically generated online past. In her case, it’s the Facebook Timeline whose “Memories” elbow their way into our feeds without warning. When the Dublin-based writer is presented with one of these “twee, occasionally horrifying” posts, she feels “confronted with loss rather than gain.” For the most part, Kiberd tells Bitch, she’s referring to lost time. Describing herself as “severely anorexic” during college, Kiberd says it was “a lost time in my life” that she could do without reminders of. “It was a period that was wasted,” she says. “So I really don’t want to see pictures of it.”
 
At their most rudimentary, “memory” features on social media resurface images posted around the same time in successive years: If it’s July 10, 2021, the algorithm will remind you of what you posted on July 10, 2017, or a photo you shared two years earlier. More advanced incarnations sort your photos into “albums” (for instance, a mix-and-match array of pictures from every July on record in your photos) or short movies—say, photos of you with a particular friend—set to generic, royalty-free music. These features don’t know how your life has evolved. They don’t know that the friend hugging you in a Facebook slideshow has blocked you on the platform. They take what you have fed them and assemble a narrative.
 
In “Manufactured Recollections: What Does a Platform Want You to Remember?,” a 2019 essay for Real Life, trend strategist Sara Reinis wrote that the memory features used on Facebook, Instagram, Snapchat, and others create “a regular cadence of induced nostalgia on a mass scale, and [normalize] a new inflection point in how and when millions of people revisit images.” Forcing us to relive our pasts as we documented them—either as social media content or in private photo collections—is just one of the ways that tech companies infiltrate our personal lives without explicit consent. Having so much visually cataloged and archived is a privilege in many ways—there is an undeniable appeal to indulging in nostalgia, and a wealth of scientific literature has recognized what The Journal of Business Research calls a “longing for a lived past” as a healthy and prosocial response. But as our personal archives grow to unmanageable sizes, should technology be the arbiter of what is worth revisiting? 
 
Ad for Hand in Hand: Domestic Employers Network
Kiberd notes that “there’s always an element of mourning, when a culture retreats into nostalgia.” The covid-19 era has been one in which a critical mass of us have lived primarily through our memories, relying on a recorded past to project a more hopeful future. On difficult days, desperate and claustrophobic, I’ve been grateful to return to posts from before, years that felt easy to romanticize in the snaps I’d taken of sweaty, happy friends in clubs, dozing on trains, and beaming in dimly lit restaurants. But that life—that version of myself—is dead. I can’t relate to the green girl in the pictures: her body is unrecognizable, hair unforgivably flat. Her ambitions are meaningless to me; too many of her friends are no longer mine. Three years after the pandemic began, we are mourning lost versions of ourselves, regardless of whether we truly want them back. 
 
There’s no shame in squeezing a little serotonin out of your lived past to weather a long and lonely period of isolation. But every coping mechanism has limitations. In a June 2021 dispatch on Catapult, Adam Dalva characterizes an “imagistic despair” that has emerged from the digital distortions of the self. “Instagram face” and Snapchat dysmorphia are old news, he notes; Zoom dysmorphia, resulting from frequent encounters with webcams, is the freshest pathology to fear. “When we’re stressed [like, say, during an epochal pandemic], we age more quickly and change our eating habits,” Dalva writes. “Our masks can cause acne. Our bodies bear the imprint of our annus horribilis, just as society has made screens mandatory.” Factor in the impact of every confrontation with a past self—captured from a flattering angle by a designated “Instagram boyfriend”—and the dysmorphic effect is compounded.
 
In her essay, Reinis argues that the algorithms behind memory features have an undue influence on us: “They convert images into prioritized signifiers of memories and package them into an explicitly hierarchical system of ‘importance,’” she writes. “These images and the way they are algorithmically organized don’t merely remind us of the past; they help shape how we think of ourselves in the present.” For instance, the algorithm’s ability to recognize and group together images of food may well give its user pause before each meal: If this is a dish you may lose to the sands of time, perhaps an attempt to save it through documentation is warranted. 
 
Reinis also notes that, in an online world where everything is content, our choices have absorbed a “commercial logic.” Is it even worth taking a photo that social media algorithms won’t bother to boost? An image of a sunset or a forest, lacking Instagram friendly human faces, isn’t likely to do numbers. “If a thought can’t be retrieved, then it’s not a useful thought,” 26-year-old former Google employee Kevin Moody told The Cut in December 2021. In a different but related context, one might ask: What use is a photo if it cannot be adequately cataloged? 
 
The “anniversary model” established by apps has implanted in us a seed of anxiety—to record your surroundings now or risk having nothing to look back on—for which the high-stakes socializing of the ongoing pandemic era provides fertile soil. Now we look around the room and worry what might, in the future, be taken from us. Almost every plan is a special occasion. We take pictures in the way we always have, to lay claim to our existence, to prove that we were here and we were happy, but with far greater fervor. The much-maligned “photo dump” (a series of disparate images intended to represent the poster’s everyday activities, often over the course of a month) is an undeniable outgrowth of this impulse: At this strange, frequently barren point in time, almost anything is Instagram worthy. 
 
Digital drawing of a woman on a screen, with other smaller iphone screens on top of the one she is on, spiralling into smaller pieces.

Illustrations by Ariel Davis

For Reinis “it doesn’t matter what is recollected, only that recollection occurs on an ongoing basis according to a regular schedule, which corresponds well with platforms’ demands for a consistent flow of content.” The 2015 press release for On This Day, a memory feature shared by Instagram and Facebook, frames this imperative as the user’s own desire: “People often look back at old photos and other memories they’ve shared on Facebook, and many have told us that they enjoy products and features that make this easier.”
 
It’s not hard to see how a sense of obligation can be created from these services, encouraging an even greater reliance on the companies that provide them. Kiberd sees the concerted effort by tech companies to harness nostalgia as a kind of emotional manipulation, to ensure the longevity of their platforms in our lives: “It’s a really creepy and fascinating inverse to how they lay claim to our futures,” she tells Bitch. “Prediction is such a fundamental part of how [social media platforms] profit: predicting our behavior, predicting what we want to buy. So it makes sense that they draw on the past to bring out our emotions, if only to keep themselves in our futures.” 
 
This, in fact, might be Silicon Valley’s greatest success to date. In a June 2021 Atlantic piece titled “I’m Scared of the Person TikTok Thinks I Am,” Kaitlyn Tiffany documents the increasing characterization of recommendation algorithms as actual extensions of the self, noting that her own TikTok viewing habits have begun presenting her with “videos that only a horrible person would like.” The fact that TikTok serves her videos of age reveals, teeth-whitening tutorials, and White Claw drinking games is something that Tiffany takes personally because, well, it is personal. Julian McAuley, an associate computer science professor at the University of California San Diego, specializing in recommender systems, only confirms her fears: “When you say ‘my algorithm,’ it’s kind of right,” he tells Tiffany. “There’s one algorithm, but input into the algorithm is everything you’ve historically done. So everybody is getting different recommendations based on their historical actions.”
 
McAuley was speaking specifically of TikTok’s eerily precise “For You” page, but memory features are powered by similarly uncanny forces. Now that smartphones and image-sharing sites have made cameras a central part of our everyday interactions, our photographs are, again in Reinis’s words, “prosthetic extensions of ourselves and our interior lives.” And so their recollection, curated in ai-generated highlights—whether in the form of albums on Facebook or cheesy “concept movies” on Google Photos—perhaps makes us more cyborgian than we’d like to think. 
 
Annalee Newitz, a journalist and science-fiction author, says that constant confrontations with past photos can mean “that we become really alienated from who we are. We start to think of our experiences as things that are transactional and almost separate from us.” The difference between “the official social media memories” created for us and “what we actually remember,” Newitz believes, will only widen. Reinis argues that commercial platforms make themselves “indispensable mediators” by taking charge of our reminiscences. While there is truth in both viewpoints, it’s worth considering whether they perhaps overstate the influence of what is a clunky and imperfect technology. When the content that the algorithm chooses to regurgitate is so clumsily selected and assembled, it’s difficult to accord them much omniscient weight. The “concept movies” fed to users of Apple and Google Photos, for instance, are widely regarded as cringeworthy, possessing a fraction of the emotional influence they are credited with. The aim of this feature, according to a 2016 interview with Google Photos product manager Tim Novikoff in The Verge, was to create films “that are emotionally powerful—that make you really smile, or even make you cry.” 
 

The metric for what’s worth remembering is fundamentally unknowable.

In reality, the evidence to the contrary is staggering, and often—as in the case of Mya Roschel’s macabre Snapchat recap—darkly comic. On Twitter, actor and comedian Eva Victor laments her iPhone’s tactless creation of a “concept movie” dedicated solely to a burst of photos she took while having a panic attack in a shoe store before a job interview  (“because someone told me I should wear a heeled bootie and not my sneakers”). The slideshow flips between images of the comedian’s feet—one inside a hot-pink ankle sock, the other in, indeed, a heeled bootie—before panning to a capture of a Google search for the term “funky shoes big footed women,” and finally concluding with a pair of exhausted-looking selfies. At least for now, the robot-human literacy gap is wide enough for us to be able to laugh at these efforts.
 
But as we slip deeper into the reality of being able to catalog and retrieve—theoretically—all our life’s experiences, we lose a degree of autonomy over what French philosopher Jacques Derrida (I’m so sorry) called “archive fever”—a drive to document that becomes a compulsion to collect everything, leaving the archive overflowing and unreadable. This is the very problem that Big Tech purports to solve via memory features, promising that its algorithms will remind us of everything worth remembering. But the metric for what’s worth remembering is fundamentally unknowable. We change, we move, we make new friends, we outgrow old pastimes. More importantly, when Apple, Google, and Facebook continually demonstrate that their users are simply data to mine, why trust them to begin with? 
 
I often find myself reaching for my iPhone like a phantom limb. Too large for the majority of my pockets, it is a far less consistent presence in the moments that I feel I most want to preserve in digital amber. More often than not, I realize that retrieving it would only break the spell of what I’m trying so hard to hold on to. So I trust my mind instead. It’s a memory just for me, and the risk of it flitting away is intrinsic to an immeasurable, uncapturable value.  

 

Jenna Mahale, an Indian woman with long brown hair, poses against a wood fence with a lavender shirt
by Jenna Mahale
View profile »

Jenna Mahale is a freelance journalist and editor living in the United Kingdom who is extremely, extremely online. She writes and edits primarily for i-D, covering film, art, music, books, beauty, politics, and digital culture, especially frog memes. Find her on Twitter @jennamahale.