Forgetting as a Function

When the internet wants to remember, how can we learn to forget?

Sarah Welsh, The University of Texas at Austin

What is surprising is not that digital media fades, but rather that it stays at all and that we remain transfixed at our screens as its ephemerality endures. Wendy Hui Kyong Chun (2011, p. 173)

You may have forgotten the case study that guides this webtext. That is, you only would have forgotten this story I'm about to tell you again if you came across it on a particular sphere of Twitter, within a certain time frame. But maybe you didn't forget it at all. Because you can't really remember something you've never read, and it's also hard to forget something that you keep being reminded of. Tweets and images and articles are posted, circulated, forgotten, and re-posted all the time before they are forgotten again. It may even be possible that by the time you have finished reading, you will find yourself in a position to forget even this webtext itself.

There is a lot that happens on the internet, and it happens in a multitude of spaces that various audiences have engaged with, shared, or stored away in conversations, search histories, and screenshots. We might remember these events if they warrant recalling, or if we are prompted to retrieve digital ephemera that would otherwise float among various networks for as long as this internet lasts. For the most part, even if something is deleted, there are lots of different ways to retrieve at least some trace of it. "Memory" is an easy and problematic metaphor for the things that are stored online. That's because the internet does not "remember" like a human brain remembers. Liam Bannon (2006) argued that designers tend to forget about the necessity of forgetting as a human activity, which limits our options for technological design. A human mind is not an "information-processing device" (p. 5), and the design of computer networks and ubiquitous computing technologies neglects the idea that it is necessary for human minds to forget.

Though Bannon (2006) was focusing especially on ubiquitous computing, the problem still arises in numerous contexts online. Even legal interventions such as the European Union's Right to Be Forgotten have done little to help. Bannon helped lay the groundwork for scholarship that reinforces the importance of forgetting to memory, which has been reflected in design features that allow for ephemeral media—for instance, the production of digital traces on platforms like Snapchat, or encrypted messaging apps that disallow screenshots. These processes on ephemeral platforms might look more similar to slight forgetfulness than they do to something more large-scale like amnesia. That being said, one way to replicate this kind of forgetfulness online could be to resist both duplication and recall by using applications intended to destroy the data you create. This is not always an option, and it does not account for information that was distributed without your consent, for instance.

In this webtext, I will propose a social process for digital forgetting (or promoting forgetfulness of media traces that should be relatively inconsequential) using one successful example from Twitter. One example is of course not exhaustive, but it was chosen as a representative model of the ways users are learning to forget. And where Bannon (2006) argued that forgetting should be a feature of design, I propose that it also can be a function performed by users, and one that works with the constraints of technology's capacity for "memory." If our systems are not built to forget, we might consider how we can do so not (only) by combating technological functions, but by working with them. Duplicating the event in this webtext to make a point about forgetting is ironic of course, because I'm providing another avenue for it to be recalled, circulated, reproduced, re-membered. But as I will show, sometimes that's a good way to forget.

step 1: post something on the internet

On October 6, 2018, Marla Reynolds signed on to Twitter and posted an old photo of her son, Pieter Hanson, in his navy uniform. The image of her son, saccharine-sweet, All-American, and posted at just the right time in the news cycle, was primed to go viral. She captioned the image as follows: "This is MY son. He graduated #1 in boot camp. He was awarded the USO award. He was #1 in A school. He is a gentleman who respects women. He won't go on solo dates due to the current climate of false sexual accusations by radical feminists with an axe to grind. I VOTE." She added "#HimToo" at the end of the post to categorize it, adding another means for discovery. Within hours, Reynolds' post went viral, thanks to popular twitter user @KrangTNelson (2018a)—K.T. Nelson, a writer for Vice—who retweeted a screenshot with his own caption:

this is the perfect post. nobody will ever top it https://t.co/lbNxdudiGe

— KT NELSON (@KrangTNelson) October 8, 2018

The original tweet can't be embedded here because Reynolds deleted it after her post was taken up in a different way than she intended. Clicking the link above will take you to a familiar error message that fills in space for something that wants to be forgotten, leaving a trace but not the original inscription. While "Sorry, that tweet doesn't exist!" does give confirmation of a thing having existed, the ability to screenshot is one of the best archival tools we have on the internet. Nelson's retweet of this trace that once existed was retweeted hundreds of times and "liked" over 5,000 times within six months of its original posting. The copy Nelson (2018b) made through a screenshot gave the post new life:

since she deleted it here's a screenshot. one of the greatest posts of all time

pic.twitter.com/Z6TUKOjH9T

— KT NELSON (@KrangTNelson) October 8, 2018

At the time of a first draft of this webtext you're reading, the screenshot above was retweeted over 16,000 times. A simple Google search for "Pieter Hanson" will bring up countless copies of the original tweet, as well as links to articles referencing the incident, images of Pieter in his life outside Twitter, and so forth. Here's a screenshot of the original found though a quick search, as currently displayed in an article on Vox (Romano, 2018):

This incident is not a new phenomenon, and the relentless circulation and manipulation of this image was due in part to the #himtoo hashtag that went viral during Brett Kavanaugh's widely criticized confirmation hearings for the Supreme Court in September 2018. Marla Reynolds' tweet was duplicated, disseminated, and reappropriated in thousands of new posts, many in the service of parody, where similar language was used with a different image. Collections of these parody images, along with the original tweet itself, were compiled into articles for websites like Vox (Romano, 2018), Quartz (Singh-Kurtz, 2018), Slate (Schwedel, 2018), and others.

This is all to say that Reynolds' tweet blew up, and the photo circulated intensely for a few days. Pieter Hanson was now a meme, when ironically, Pieter Hanson the person had never created a personal account on Twitter. What ended up propelling this incident to a stop was the smart response from Pieter and his brother when they found out about what their mother had posted. As it turned out, Reynolds had used his image in support of a viewpoint her son did not subscribe to. #Himtoo had gained steam as a category on Twitter for conservative-leaning accounts to find memes and voice concerns about women falsely accusing men of sexual misconduct. She posted the image at the perfect time and place for it to go viral: because of the hashtag, it was easily discoverable by liberal accounts looking to poke fun at the way #himtoo was being used. The way Pieter responded was an easy narrative for news outlets to distribute, in a time when divisive American politics had reached yet another unpleasant climax.

Pieter Hanson was sitting in class when his friends began texting him screenshots of his mom's tweet, retweeted thousands of times by thousands of Twitter accounts. His photo now existed in a space where it would continue to proliferate beyond his control. Hanson, a former Marine-turned-marketing student, responded in a manner that exquisitely reflected what the internet had become accustomed to. After he created a Twitter account so he could intervene, he acknowledged the image his mother had posted, and took a corresponding photo with a new caption that represented his own views (Hanson, 2018). It was a case study for effective digital crisis management in spaces where screenshots can be used to circulate a story in perpetuity. To save your reputation from being relentlessly decimated, sometimes you can lean into the internet's unique ability to duplicate and save, but in a way that is kind, graceful, and attention-seeking. Using the medium to create more copies of the thing you want to forget sometimes has the odd effect of the attention cycle blowing up and then dissipating. In crisis management and public relations they call this "controlling the narrative." Rather than delete, Hanson responded with another tweet:

That was my Mom. Sometimes the people we love do things that hurt us without realizing it. Let's turn this around. I respect and #BelieveWomen . I never have and never will support #HimToo. I'm a proud Navy vet, Cat Dad and Ally. Also, Twitter, your meme game is on point. pic.twitter.com/yZFkEjyB6L

— Pieter Hanson (@Thatwasmymom) October 9, 2018

Hanson's brother also responded by posting multiple images of them together in real life, and other friends also posted their own pictures of Hanson in his life outside the Navy. Together, they started a process that saturated Twitter with another image to distribute and replicate in order to replace the original that within hours had become a parody of itself. Further, Pieter Hanson's response gained steam by becoming a new symbol for left-leaning news outlets and Twitter accounts to take up as evidence of #himtoo as a sham. Pieter's response shows that sometimes a good way to forget something online is to replace what you want to disappear with a new narrative. The kind of digital crisis management that Pieter did here—where the attention he directed quickly had the effect of a digital forgetting—is a lesson for individuals and even companies facing similar cases of unexpected virality, circulation, and repetition via relentless memory. In short, after a time period of concentrated attention, the audience moved on.

But within the first 24 hours, Marla Reynolds' original tweet—along with and because of Pieter's response to it—spawned articles from the Guardian (O'Neil, 2018), the New York Times (Garcia, 2018), Today (Holohan, 2018), the Washington Post (Flynn, 2018), NBC News (Rosenblatt, 2018), the Verge (Maloney, 2018), Time (Hoffman, 2018), and the BBC (Gerken, 2018), among others. The original tweet—along with Marla Reynolds' account—has since been deleted, but it has by no means been forgotten. It now exists in countless forms, in countless parodies, and within days, it had turned into a symbol for a very familiar narrative: the complicated relationship between conservative parents and their liberal-minded children. While deletion seems like it could be an easy way to forget, it's never that simple when "memory" online is public and collective. In this situation anyway, Hanson had an astute understanding of the way the internet works: in order to forget information online, you have to replace it with more information and direct attention from it in a different way. Hanson's mother deleted her Twitter account, but this did not stop the post from proliferating and circulating. The life that your posts are given beyond the moment you hit "share" provides records for later accountability if your post is reprehensible, but at the same time our nearly infinite ability to recall leaves little room to change your own narrative.

step 2: redirect attention

As this example shows, the repetition that is possible in digital spaces is relentless and unforgiving. Duplication and circulation make it extremely difficult for the collective consciousness of the internet to "forget" something that existed on social media. That is, unless you are able to successfully redirect the attention (of the people who noticed). Forgetting online functions not by directing attention away from "X" by trying to delete it (which is basically impossible and can exacerbate the problem), but by redirecting attention from "X" to forget it by contextualizing it in a different way, similar to how Pieter Hanson responded. This is simultaneously a problem with and a function of a digital economy of attention within systems of distributed intelligence as exemplified by the way social media operates. Forgetting has never been an essential part of a capitalist system that seeks to direct attention to specific commodified information in a vast sea of it, but in some cases it might be something that we (as humans, societies, governments, corporations, users) need. If we think about our social media practices as existing within an ecology of attention, or, the formulation of a digital existence that is both "collectively sustainable and individually desirable" (Citton, 2017, p. 203), forgetting becomes more important because it is also essential to life itself.

Friedrich Nietzsche (1980) argued that in order to survive—to truly live—humans need both the "historical and the unhistorical," or memory and forgetting (p. 10) The ability to remember, but also to forget, means that we are not bogged down by the past and thus have the ability to use historical knowledge for growth. Bradford Vivian (2010), who has argued for the importance of forgetting by thoroughly interrogating memoria, detailed the impulse towards remembering in our present day and age where, "archival documentation, revivals of communal heritage, and commitments to preserve memory at all costs consequently hold widespread cultural priority" (p. 4). In a 2009 review essay, Vivian considered how Avishai Margalit's (2002) Ethics of Memory, Paul Ricoeur's (2004) Memory, History, Forgetting, Marc Augé's (2004) Oblivion, and Miroslav Volf's (2006) End of Memory together illustrate how forgetting might be a "productive and even desirable aspect of collective life, ethics, and decision making" (Vivian, 2009, p. 90). This review, along with Vivian's other work, helps establish a theoretical grounding for the merits of forgetting when traditional lines of rhetorical inquiry see it generally as memory's foil. This work also serves as a productive starting point for considering memory and forgetting in the age of the internet.

Beyond curated digital archives which have been theorized at length, boundless stores of data that the internet allows for signal what Vivian (2010) called "archival obsessions," where memory is "lost or lifeless" (p. 33). Here, there is an overflowing of objects and traces that are left to be forgotten (and then recalled). Though Vivian's idea is about public memory in the context of written histories, the same might be applied to our digital repositories, both institutional and ad hoc, where deleting is not forgetting, and screenshots abound. The focus on forgetting signals a shift to a digital age where, as internet scholar Viktor Mayer-Schönberger (2010) argued, we are facing the possibility of knowing what it means to lose forgetting to memory by losing our ability to truly delete. For Jeff Pruchnic and Kim Lacey (2011), the future of memory "as a rhetorical force," will likely be tied to the future of forgetting (p. 491), and for Collin Brooke (2000), we are only beginning to understand the dark side of our "distributed memory" often via information overload, and through technology, we are learning "what it means to remember too much" (p. 792). However, as some of the examples I have used in this webtext illustrate, we may not lose forgetting so much as we will have to invent new ethical orientations toward forgetting. That is, there are ways to work around what current and long standing algorithms and mechanisms allow, redirecting attention from the internet's relentless memory.

On the internet, on an average day, humans generate about 2.5 quintillion bytes of data. This often-cited number likely originiates from an IBM market research group, and is probably higher in 2019 than when one article citing this number was written in 2013 (Jacobson). Nevertheless, that number looks like this:

1,000,000,000,000,000,000,000,000,000,000.

This increasingly limitless capacity for recall—facilitated by the ways we collect, store, and share—built into the algorithms and mechanisms themselves, further reinforced by our own digital practices. When it comes to the circulation of digital ephemera, digital recall is essential to discussions of memory, because memory online is just storage, and a digital textual object's existence depends on a user recalling a particular piece of information to help it gain prominence or recognition, or to be used as leverage or evidence. In Lingua Fracta, Brooke (2009) reworked the canon of memory as persistence, where digital memory aggregates, and memory is not just simply data storage.

Physical storage and collective memory together have the potential to be monumental and all-encompassing, surpassing anything conceivable by the limits of our own humanity. Forgetting as a rhetorical force, is not built into computational memory systems by design, but happens in other ways. We might forget by default through digital obsolescence, banning accounts, or in more ad hoc ways like saturating the internet with more information or playing with distraction and attention. Alongside a history of modern technology is also a history of the impulse to save and store and replicate and duplicate, in a world where even everyday minutiae has the ability to be recalled.

The processes, functions, and products of memory and forgetting are recursive (Stormer, 2013): forgetting is intrinsic to remembering, has a rhetorical capacity alongside memory, and similar to memory, is essential to elements of discourse and is crucial to history. If we are to consider memory, remembering, or recalling in a digital sphere, we must also consider the possibility, impossibility, and even the meaning of "forgetting." In a digital ecosystem where we are able to delete, we are not always able to forget because we are provided an almost infinite capacity for recall via circulation and physical storage. Our systems and the virtually effortless means to circulation don't allow for forgetting in ways that might be productive or even necessary. This infinite capacity to recall of course has benefits, and keeping a backlog of easily recallable information has the capacity to hold people accountable for what they say—however, this doesn't always work in practice, especially when the volume of information produced overwhelms the tension between recall, attention, and persuasion. People like Donald Trump, who have figured out how to manipulate this tension, serve as evidence of the capacity for recall without accountability, but even those who are out of the "public eye" could be better equipped if they knew how to help each other forget.

Colloquially, and even in scholarship that deals with the persistent theoretical demands about the digital, we often talk about "memory" online and Google's ability to remember everything (thus the persistent demand for a Right to be Forgotten). However, reframing and expanding our thinking to consider recall rather than memory specifically, may get to the heart of some of the most critical questions within and about digital memory. The kind of memory that percolates on the internet does not exist in a vacuum, but rather depends on the circulation of digital materials. Scholarship about memory in the realm of the digital often declines to distinguish "recall" alone, which is at the heart of the user's ability to "remember." We can store something online, but the memory of that thing must first be provoked in order to recall it from the vast stores that the internet provides. All of these factors must also be present for something to warrant forgetting. As long as we are given a reason to recall someone online, it exists for us, and this near-infinite ability to recall also assumes a near-inability to forget. This is made possible by both physical storage capacity (which for the sake of brevity I'll define as both servers and the cloud), and social duplication and circulation. When we want someone to "forget" online, working with the social construction of digital "memory" via circulation and repetition may help make a move towards forgetting, barring any legal action or structural changes within the systems themselves.

step 3: delete and repeat

One early example of a digital object that included an expiration date of sorts as part of its design was William Gibson's (1992) digital poem "Agrippa," which can be seen in the YouTube clip above. This poem, both a piece of literature and a work of digital art, playfully touched on the idea of the ephemeral nature of human memories. Once played, the poem was meant to encrypt after a single use, and the book it was encased in (made with photosensitive chemicals), was also meant to gradually decay. The object was meant to replicate a real-life memory that erodes or disappears or decays within our own mind. Matthew Kirschenbaum (2008) wrote about "Agrippa" extensively in his book Mechanisms but also investigated the persistence of the poem and in a blog series (2005). Also of note is that at the moment I am writing this, the blog is unavailable. But according to Kirschenbaum (in my retrieval of the blog months before), the poem was released on an early internet messaging board called MindVox the morning after its debut, at an art space in New York City on December 9, 1992. "Agrippa" was meant to disappear, and yet, because of the ways digital objects are replicated and disseminated, what Gibson meant to be ephemeral was not, and it still exists online today. You can also read the text of the poem, which exists on William Gibson's website.

Along with the persistence of digital memory, there are of course many ways the internet already forgets. In researching the lifespan of the early-1990s "Agrippa" for this webtext, there were endless amounts of dead links. And yet, the text itself also persists on YouTube through a proto-screenshotting wherein a "reader" videotaped the poem in between presentations of it. This is yet another version of the screenshot memory that exists of "Agrippa." As Kirschenbaum (2012) noted in the description for the YouTube video I have embedded below: "Unbeknownst to the event's organizers [an event participant] had slotted a blank video cassette into the camera used for the live feed." Even though a participant at the event deliberately attempted to sabotage the live stream of the poem, another video was taken during the event, which depicts the poem anyway. From about 25:00 to 52:00 in the video, we bear witness to the poem unfolding and folding away to be forgotten. Even when it is not properly archived, media tends to lend itself to reproduction, circulation, and distribution.

Webpages have to be curated and maintained, which is often the responsibility of corporate interests (like YouTube and other widely used social media channels), but also individuals. The interplay between people and these platforms allows information that we might want forgotten to proliferate. The text of "Agrippa" matters enough to some people that it has been saved and stored and maintained on the internet since 1992. We can't say the same thing about everything that exists online, but the aura of some works of art allows them to remain, even in an age of reproduction.

In a 2005 blog post referenced earlier, Kirschenbaum discussed the textual mutation and persistence of Agrippa online. He explained that the text itself has persisted despite the fact that it was built by design to be "uniquely volatile." The fact that the poem has had a rich existence after it was built to be destroyed leaves us with a valuable lesson about digital textual objects and their durability. Kirschenbaum's analysis (and the poem itself, of course) debuted before Snapchat and other forms of deliberately volatile, ephemeral media, but the traces produced within these platforms also often enjoy an afterlife. This is the same for websites that are deleted, texts that are deleted, and images on Twitter.

Kirschenbaum (2005) argued, "'Agrippa' owes its transmission and continuing availability to a complex network of individuals, communities, ideologies, markets, technologies, and motives." Using this poem and its subsequent event as an example, he went on to explain that

the preservation of digital media has a profound social dimension that is at least as important as purely technical considerations. Today, the 404 File Not Found messages that Web browsing readers of "Agrippa" inevitably encounter—the result of the file no longer residing on a local server where it was originally indexed by the search engine—are more than just false leads: they are latent affirmations of the work's original act of erasure that allow the text to stage anew all of its essential points about artifacts, memory, and technology. "Because the struggle for the text is the text." (Kirschenbaum, 2005, quoting McLeod, 1991)

Ironically, the practice of remembering Agrippa helped to forget Agrippa—that is, while we might remember the story of its transmission and the fact that it says something about memory, we do not remember the words of the poem itself.

The visual reminders of what were once textual objects that Kirschenbaum described above call to mind the succession of retweets in the image and events that transpired on Twitter and elsewhere on the internet as I've explained in this webtext. The original text still exists in its deleted representation, and even the absence of text still leaves a trace of what once was.

this is the perfect post. nobody will ever top it https://t.co/lbNxdudiGe

— KT NELSON (@KrangTNelson) October 8, 2018

The "Agrippa" example offers a media archaeological lens through which to understand the social dimension of media distribution, both media that is meant to be destroyed, but also media that is meant to be "permanent." In many cases, the social dimension cannot be stopped, much as we would like it to be, and at the same time "the computer itself in its storage capabilities is far from a static machine of memory" (Parikka, 2012, p. 93). Digital media is social, and here we might extend that sociality not only to humans but also to the inherent sociality of networks. In this way, forgetting also seems to be social when it works effectively; that is, to think of forgetting as simple deletion ignores the social dimension of media. "Agrippa" is an example of an object that was meant to be destroyed but that was dug up and distributed because the way it was deleted served a purpose both to research and human interest. But human interest is also relevant to media produced by private citizens who are not celebrated science fiction authors, and all digital objects have potential to become objects of inquiry if they were of interest to someone, somewhere.

Though Marla Reynolds deleted her original tweet, I have been able to trace the story back through representations and impressions provided by social sharing and archiving. But I have also been able to trace this story back through and because of platforms, algorithms, mechanisms. Even when we create something that is explicitly meant to be destroyed on an ephemeral platform like Snapchat, we generally have the ability to recover that object. And if the recovery and storage of traces on platforms that are meant to be ephemeral is possible, it is even more possible on platforms like Twitter, Facebook, Tumblr, Medium, and many others, that are meant to save. The example of Pieter Hanson is one small example of how we (as humans) might work with this ability to recover traces regardless of the platform in which a media representation was born. I should also note that the means to forgetting could be much different, more labor-intensive, and more challenging for someone who is not an able-bodied white male but a woman or a woman of color, for example. Despite this (but not in spite of it), the critical attention here is in the acknowledgement that the systems of reproduction enabling media storage are also an integral component of forgetting.

This webtext will disappear as a function of its design. But the page will refresh, tweets will be saved with screenshots, and we will lose control of the things we say after making the choice to say them. At best, users will be inconvenienced by the way forgetting has not been built into our systems, and at worst it might have life-altering consequences. And still the act of deletion signals a whole other host of repetitions that make the internet what it is.

step 4: forget, again