Manufactured Recollection

What does a platform want you to remember?

As digital images and image-sharing apps have taken us beyond the physical and technical limitations of film, we have entered an era of constant photo taking. Estimates suggest that well over a trillion photos are now being taken annually, leading to overflowing personal photo archives that are increasingly difficult to sort through manually. My own Apple Photos library extends back to 2008 and includes more than 30,000 photos — three times more photos than days I’ve been alive.

To help sort through the volume of digital photographs, social media apps (Instagram, Facebook, Snapchat) and personal photo libraries (Apple Photos, Google Photos) have deployed algorithms to serve as curatorial assistants. They have, for instance, transformed my 30,000-plus photo swamp into a somewhat manageable and navigable retrospective of all the experiences I’ve attempted to capture with digital photography. But as a result, I am remembering much of my life through the algorithmic frameworks of these third-party companies.

Sorting algorithms automate the process of looking back, giving a sense that if something is worth revisiting, they’ll let me know

These algorithmic systems have become widespread, not only sorting or surfacing images but contributing to how they become meaningful. Their features across different platforms are more or less standardized: They can automatically produce albums of vacations and other highlight reels, and quantify the exact number of selfies I have taken (a number that shall remain an intimate secret between the iCloud and me). They can draw on popularity metrics to determine what to resurface. They can create slideshows of specific friends using facial recognition technology or of specific subjects — I can type “food” to see a full catalog of all the dishes I have photographed over the years. They also try to quell too-many-photos anxiety with notifications prompting me to revisit images or albums from years past, automating the process of looking back and giving a sense that if something is worth revisiting, they’ll let me know.

Through such “Memories” features, algorithms increasingly contribute to both the content and cadence of what we remember. This infuses our self-understanding with a commercial logic, as the interests of social media platforms, advertisers, and technology manufacturers are blended with individual memory-making practices.

Our photographs exert powerful influence over our memories. Since its advent, photography has played a central role in cataloging personal experiences and family milestones, isolating particular moments as potential triggers for future remembrances. “No object is more equated with memory than the camera image, in particular the photograph,” writes Marita Sturken in her book Tangled Memories: The Vietnam War, the AIDS Epidemic, and the Politics of Remembering, “Yet memory does not reside in a photograph, or in any camera image, so much as it is produced by it.” Because of a framed picture on the mantle, we can think fondly on how beautiful our mother looked on her wedding day. Due to a particularly awkward braces-laden grimace in a yearbook photo, we experience a surge of embarrassment and a wash of uncomfortable seventh-grade feelings. At times, we even create false memories in response to a photograph.

Photographs carry with them both the promise of objectivity (“pics or it didn’t happen”) and the emotional weight of subjectivity (“this sunset is beautiful”), making them a powerful force in personal identity development, linked to both self-documentation and self-expression. As camera technology becomes more accessible, subjectivity itself is increasingly shaped by the camera’s capabilities, which molds how we see and ultimately how we remember. Phones and image-sharing sites have made cameras even more central to our everyday practices — making it easier to take and share photos continually as a routine mode of communication. Consequently, we view photographs not merely as relatively rare artifacts capturing particularly significant moments but as prosthetic extensions of ourselves and our interior lives.

When algorithms intervene in how and when we interact with our photographs, they secure a deeply emotional inroad to our identity-forming practices. They convert images into prioritized signifiers of memories and package them into a explicitly hierarchical system of “importance.”  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 and how we might think to document our lives and articulate ourselves in the future.

“Memories” features across photo platforms function together to form a fairly uniform landscape of manufactured recollection that underpins the lives of nearly everyone with a phone and a social media profile. For users who are presumed to be overwhelmed by the amount of photos they’ve taken over years, these features are meant to be reassuring, while further embedding our personal histories within particular platforms. Nothing will be lost to the sands of time, the features promise, and no special effort will be required to sift through swaths of content to experience the emotional spike of nostalgia.

One of the earliest memory features is based on a simple algorithmic function: showing images you posted or captured on the same day in years past. Originally popularized by the standalone app Timehop, a version of it is now implemented on Facebook, Instagram, Google Photos, and Apple Photos, creating a regular cadence of induced nostalgia on a mass scale and normalizing a new inflection point in how and when millions of people revisit images.

No special effort will be required to sift through swaths of content to experience the emotional spike of nostalgia

At its core, this feature constantly affirms that posts ought to be revisited and reshared with periodic regularity, with virtually any previous image possibly warranting an “anniversary” and a celebration. “On This Day” tells a story of our own significance back to us that is strangely indifferent to the particular “content” of our past. 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.

A more sophisticated algorithmic process is involved with AI-generated “Highlights,” another now standard feature that tries to assesses the likely importance of images and assembles them together, layering on themed music, illustrations, and titles. Facebook develops seasonal highlight albums, using cheery cartoons to direct you to view, for example, your “Winter Memories.” Instagram gathers “Suggested Highlights” from your Stories under titles like “Mexico” when you have posted with a different geolocation than usual, or “Together” when you have tagged another user’s account in multiple posts. On its app store description, Google Photos claims its automatic albums let users “tell better stories, without the work. Automatically get a new album with just your best shots after an event or trip.” Hence you can trigger an AI-generated “Love Story” of yourself and a partner complete with acoustic guitar swelling in the background, or notify the app that someone in your camera roll has passed away so it can piece together an “In Memoriam” slideshow.

The idea is, as Google Photos product manager Tim Novikoff told the Verge, to have AI “make movies that are emotionally powerful — that make you really smile, or even make you cry and reminisce and show your family.” In other words, they purport to do the work of investing your images with emotional significance without your participation, so that the viewer can consume rather than produce their recollections. In the same vein, Apple Photos automatically creates a highlight reel when it detects a large sum of photos taken in a short period of time, even if it is unsure of the exact occasion — in those cases it simply labels the album “A Big Day.” These auto-populating album tools promise to sort through our thousands of photos to find identifiable themes and extract important moments — making us audiences of ourselves as the algorithms piece together our “best” stories for us.

These personal story-creating mechanisms are superimposed on all platform participants — no one can opt out of being part of someone else’s highlight reels. And the same ranking procedures and pre-assembled categories hold firm for everyone, creating a standardized script for memory and identity formation across the lives of millions. “Highlights” encourage us to rely on externally imposed criteria of relevance, freeing us from the responsibility of intentionality: Rather than considering the potential significance of a moment while capturing an image, we can take an endless amount of snapshots with the assurance that algorithms will eventually produce relevance for us after the fact. The sheer number of photos then becomes a key indicator of significance for the algorithms. Users may thereby learn to associate a quantity of photos taken (and data produced for platforms) as the surest way to mark a qualitative experience.

To help sustain the creation of the photos and data they depend on, nearly all “Memories” features are accompanied by suggestions to share. For example, when a user receives an “On This Day” notification from Instagram, clicking on it will lead to a pre-assembled Story post with the date it was uploaded and the word “Memories” superimposed. Every time we are greeted with an “On This Day” post, we are urged to reshare once more so as to gain a second chance at being relevant on others’ timelines. While remembrance and photography have always been oriented toward social sharing — think of signed yearbooks, family photo albums, Christmas cards — automated suggestions that push for constant image recirculation in a monetized environment add new commercially dictated incentives.

Under the influence of this kind of manufactured recollection, memory is firmly tied to the act of creating and circulating more content (#tbt posts, etc.). As Facebook product manager Jonathan Gheller told TechCrunch, “Consuming memories from the past and resharing them rewires our relationships.” But “Memories” features rewire relationships in such a way that makes commercial platforms indispensable mediators. In this environment, even before we have revisited something, the memory is already packaged as content. It is already prepared to have advertising sold against it and to be leveraged to fine-tune further targeting. As a result, personal and commercial spheres continue to collapse into one another, with tech companies forging the pathways they want us to take.

In this context, memories become susceptible to being evaluated according to performance metrics. The feedback loop makes us keenly aware of the impression and engagement we get on our images when they are construed and publicized as memories. Facebook regularly provides automated reports on engagement, prompting us to “remember” which photos received the most likes and comments in previous months or years. On my Facebook homepage, for instance, the company will resurface a post such as my “Most Commented on Photo from 2015,” with animations and colorful text bubbles dancing over a picture of my mom, my sister, and me. This constant reminder of how memories perform in the public eye links what we remember — and what we might come to think is worth remembering — with what drives engagement for platforms.

These kinds of loops are most salient on Facebook, but “success” metrics like these cast a shadow across most platforms. For instance, Instagram allows anyone with a personal profile to switch their account to a “business” account, which unlocks a new vault of metrics including what time of day posts perform the best, how each post performs in comparison to others, and suggestions to put paid promotion behind top-performing posts. Third-party apps and sites are also increasingly springing up. Each December for the past couple of years, feeds are inevitably taken over with posts created through apps like “Top Nine for Instagram,” which displays a grid of people’s most-liked photos of the year in one succinct celebration of audience favor.

This kind of prodding teaches us to put the idea of performance at the forefront of how we form memories. It suggests that one should remember what attracts external attention — that the point of memory is approval, and that memory follows from potential monetization. Constant reminders of performance and prompts to post ask us to adapt and optimize our behaviors in accordance to the “market research” that algorithms are constantly doing on our behalf. Algorithm-supported “Memories” features that ostensibly help us manage our excessive photos in fact pull us deeper into a cycle of constant production. They gamify our memory-making practices to capture more attention.

The memories placed at the forefront of our minds inevitably look ever more similar to the advertising campaigns and the influencer posts that surround us

It is crucial to remember that most digital photo libraries are commercial enterprises. They have a vested interest in increasing our dependency on them and engagement with them. Memories features are partly an attempt to use the affective power of nostalgia to intimately connect users with platforms and reshape subjectivity with ultimately commercial prerogatives. As these tools supplement and shift our memory-making practices, personal narratives are increasingly co-authored by algorithms. With this increased intimacy between us and photo-sharing platforms, how can we withdraw? Even as evidence mounts of how these platforms abuse our trust and exploit our data, how do we divest from them when they are becoming intertwined with our understanding of who we are?

The question of what these platforms call us to remember ought to be accompanied with a question about what these platforms are making us forget. Studies have demonstrated that repeatedly revisiting certain things causes the forgetting of non-target information, a phenomenon known as “retrieval-induced forgetting.” Commanding subjects to recall a particular set of information can cause information on the sidelines to fade.

As algorithmically manufactured recollection on platforms continues to take hold, we must ask what memories are left on the outskirts. What experiences are illegible to or unvalued by a commercial system? What does it mean for our subjectivities at large that we are all building our memories around same scaffolding? Over four decades ago, Susan Sontag posited that photography enables “an aesthetic consumerism to which everyone is now addicted.” As a multiplier of photography’s influence, algorithmically fueled “Memories” features bring us deeper into a supercharged aesthetic consumerism that shapes our personal narratives along the lines of influencer culture.

It is important to note that I am not calling for a retreat from the practice of photography nor am I appealing to the idea that there is some pure, untouched space of “true” memory that is being compromised. Outside forces, social influences, and emerging communication tools have always shaped how and what we remember. Socrates famously bemoaned the spread of the written word, seeing it as an incomplete representation of knowledge that enables forgetfulness. Memory-making practices have never existed in a vacuum. However, there are crucial differences about this new era of algorithmically controlled manufactured recollection — most notably, scale and uniformity.

In a world where the same kinds of algorithms determine both the personal memories we should revisit and the ads that perform the best, the memories placed at the forefront of our minds inevitably look ever more similar to the advertising campaigns and the influencer posts that surround us. The social consensus around what is “worth” remembering is becoming more and more tethered to the same ideal, in which all memories are visual, polished, and commercially viable.

As anyone moving through this life knows, formative experiences are much more textured and complex than a well-performing, aesthetically pleasing snapshot. In the face of shifting notions of memories (and who or what should dictate them), we must work more actively to stay connected to experiences that are not so easily captured and packaged within this system. This doesn’t necessarily mean avoiding digital photo libraries. Rather, we can diversify our approach to reminiscing through maintaining off-platform records, introducing other mediums, and otherwise investing more energy to that which escapes the purview of the camera.

To mitigate superimposed commercial narratives on our self-understanding, we must pursue individually or communally determined structures of importance rather than offloading that responsibility to a commercial algorithm. We can also explore how to preserve connections to experiences and thoughts beyond the cues of images. A superfluity of images makes it easy to conflate them with memories, but remembering is truly boundless.

Sara Reinis is a strategist by trade who thinks obsessively about the internet. Her research and writing interests surround social media, visual culture, and new technologies.