On May 8, 2023, a Twitter user expressed sadness over the loss of a dead loved one’s Twitter account: “My sister died 10 years ago, and her Twitter hasn’t been touched since then. It’s now gone because of Elon Musk’s newest farce of a policy. Fuck you @elonmusk, your nonsense has taken away a monument to my sister’s mark on this earth.”
Soon after Twitter’s new deletion policy took hold, Google made an announcement of its own. A May 16, 2023, blog post stated that Google would start deleting inactive personal accounts: “if a Google Account has not been used or signed into for at least 2 years, we may delete the account and its contents.” Much like Twitter, Google blamed security issues as the company’s main concern. (Long-inactive accounts are less likely than regularly accessed ones to have two-factor authentication and may become compromised, spewing spam or other unpleasant content out into the world.) However, this policy change invalidates Google’s earlier promise to store your data forever for free. For example, Google Photos claimed in 2015:
Google Photos gives you a single, private place to keep a lifetime of memories, and access them from any device. They’re automatically backed up and synced, so you can have peace of mind that your photos are safe, available across all your devices.
And when we say a lifetime of memories, we really mean it. With Google Photos, you can now backup and store unlimited, high-quality photos and videos, for free.
Google Photos ended its free unlimited storage in 2021.
Tech titans gained power and wealth from the accumulation of data, but that doesn’t mean they are equipped to be long-term stewards of personal and collective memories. Even the longest-lived social media platforms have undergone tremendous changes, and some, like Twitter (now X), teeter on the precipice of oblivion. And many companies, it seems, would rather eschew their responsibilities as digital caregivers. They gobbled up massive amounts of user data for model building and to attract advertisers, and they can just as easily decide to free themselves of their obligations to preserve such data.
For ordinary users, personal data may seem permanent, something that can follow them across the life cycle. Yet such permanence doesn’t always align with corporate interests in and interpretations of data. Today, Big Tech companies are no longer willing to maintain data in perpetuity. We are perhaps reaching the limits of what the cloud can afford.
Tech companies, whether fledgling digital estate-planning startups or massive multinational corporations, are ill equipped to broker the intergenerational transfer of digital remains because of their short attention spans. Moreover, corporations often propose the deactivation or deletion of dormant accounts to avoid liability for any security issues that might arise from keeping them online. Twitter has repeatedly planned to deactivate such accounts, but up until this latest policy shift, user pushback and press attention prevented it from becoming a reality. Under Elon Musk’s chaotic ownership, this time the plan was carried out, at least in part. One petty billionaire had the power to delete long-standing memorials to the dead. Such deletions can also carry their own political implications, such as freeing up handles for right-wing politicians, one possible incentive for Musk’s decision, although simultaneously upsetting the loved ones of dead users.
Despite Big Tech’s tendency to ignore the dead, however, death seems to haunt data infrastructures. In my book, Death Glitch: How Techno-Solutionism Fails Us in This Life and Beyond, I discuss the thorny problem of maintaining the data of the dead, which requires enacting care both at scale and over time.
Here I explore the politics and ethics of endless posthumous data storage, especially at a time when the climate impact of the proverbial cloud is a pressing concern amid the rise of generative AI and other high-energy workloads.
Over the decades, platforms have grappled with the problem of retaining and caring for the data of the dead. Digital remains are complex inheritances, because they depend on the longevity and commercial viability of corporate platforms and proprietary systems. Consider how the remains of the dead might well encompass everything from email, blog, and social media accounts to the ambient forms of metadata that track individuals and their networks. All this—when users die or platform infrastructures break down—becomes digital remains.
Commercial platforms can provide the scaffolding for sacred communion with the dead. But such relationships depend on the whims of platform owners and the design decisions of technologists.
For ordinary users, personal data may seem permanent. Yet such permanence doesn’t always align with corporate interests in and interpretations of data.
For mourners, losing access to a dead loved one’s accounts can be devastating. And such losses can be emotionally significant to entire online communities, even aside from an individual’s close friends and family.
There is also the question of collective historical memory and the politics of the archive. As historians of computing have argued, the web itself is difficult to preserve. Think about experiences with the early web. What traces are there? What isn’t captured by the Internet Archive’s Wayback Machine or other digital archives? Remarkably, the first web page, created by Tim Berners-Lee in 1990, is not preserved.
In general, not everything attached to the web can be preserved, including unspoken practices or the feeling of using a particular feature. This means that users have to take responsibility for their own preservation and must take screenshots and maintain hard drives across obsolete systems. This requires a great deal of maintenance labor. But even that labor cannot ensure that data will be saved and accessible as formats change or devices and their permissions stop working.
Such issues reveal the collective nature of digital production, which requires the reproduction of specific practices and expertise. A single person’s emails, tweets, or posts in early web forums are part of an entire ecosystem. And care for these remains can be emotionally taxing for others.
As I found by interviewing digital caretakers for my book, there is often a disconnect between what people think they want others to do after they die and what mourners really want to do. Physical caregiving for a loved one intersects with digital housekeeping, from the deletion of spam emails to payments for domain names, or, in some cases, uploading a final goodbye post on behalf of a terminally ill blogger spouse before they take their last breath.
Death reveals other tensions, which can play out in the control of digital remains. When the prolific blogger Mac Tonnies died in 2009 at the age of 34, his parents were surprised to find a stream of condolence messages on his blog, which had become a gathering place for an entire online community. His very offline parents did not think about his blog or Flickr accounts as they sorted through his belongings, so it was up to his online friends to preserve Posthuman Blues on a hard drive. Thanks to their efforts, that blog is still online today, as if frozen in time.
The networked labor required to maintain the data of the dead becomes all the more complicated when companies themselves change policies or go offline altogether. As I recount in my book, companies like Google, Twitter, Facebook, and Yahoo all have different policies regarding digital remains, and those policies fluctuate over time to accommodate changes in technology and culture. Consequently, there is not a streamlined protocol for dealing with dead users.
In a statement in late 2019, Twitter said the company would delete inactive profiles. From Twitter’s perspective, this would encourage more engagement while also protecting against bots, impersonators, or other shadow accounts. In theory, it would make the platform more trustworthy while freeing up underutilized handles. To Twitter’s surprise, many people responded in anger: the company had forgotten about dead users. Craig Jenkins, a music critic for New York Magazine, tweeted, “Please do not scrub the dead homies’ accounts.” Twitter representatives backtracked and apologized, saying they would delete inactive accounts only after they had found a way to properly memorialize the accounts of the dead.
Google was one of the first major companies to institute a formal policy, launching the Inactive Account Manager in April 2013. A product manager, Andreas Tuerk, wrote about this new feature: “Not many of us like thinking about death—especially our own. But making plans for what happens after you’re gone is really important for the people you leave behind. So today, we’re launching a new feature that makes it easy to tell Google what you want done with your digital assets when you die or can no longer use your account.” With this novel service, users could select up to 10 people to receive designated parts of their Google account data. The individual guardians could then download the content, although they would not receive access to the account itself. But, as with Facebook legacy contact and other attempts at managing digital death through design features, a large percentage of users never signed up for the service in the first place.
It’s funny, because digital estate-planning startup founders I interviewed back in 2012 or 2013 saw the coming tensions between data as power and data as liability. Although data collection helped make tech companies powerful, it is expensive and energy intensive to maintain vast stores of data for eternity. So it is actually in companies’ best interest to eventually scale down their preservation efforts, particularly when it comes to long-term storage.
When Google released their Inactive Account Manager, one savvy digital estate-planning startup founder was cynical about why Google was doing it. She told me that companies don’t want to have to maintain people’s data forever. It isn’t worth the cost, or the potential security and legal risks. Even as companies began to recognize the monetary value of data—including the data of the dead, which can be valuable in its own right for creating models of user and consumer behavior while keeping living users invested in aging platforms—they began to weigh the consequences of preservation at scale.
Have we reached the limit of the cloud as a computing metaphor? Will companies rethink their earlier promises of digital eternity?
What the media theorist McKenzie Wark calls the vectorialist class took power by gathering and controlling data, but now that form of domination might be more of a curse. Along with privacy headaches and security concerns, there is also the matter of financial cost and energy consumption. Massive amounts of resources are needed to create and preserve data at scale. Data centers suck up water in drought-stricken countries, which is why Google’s plan to build one in Uruguay is being met with resistance. According to VentureBeat, in 2020, data centers created roughly 1 percent of the global electricity demand and 0.3 percent of all CO2 emissions worldwide. MIT anthropologist Steven Gonzalez Monserrate has shown how social and technological factors contribute to the ecological cost of data centers, which emit more carbon than the airline industry. In a recent special issue of Culture Machine, coeditors Mél Hogan and Asta Vonderau also point to the massive environmental footprint of data centers and their dependency on natural resources.
Water and energy are used to cool servers, which can easily overheat. The media scholar Jeffrey Moro discusses the “total heat death that awaits all data and the infrastructural technologies data centers employ to delay it, although they can never eliminate it entirely.” Anthropogenic climate change poses an existential threat to data storage, which ironically requires intensive resources to keep cool, thereby increasing the emissions that cause climate change in the first place. Other theorists, like Tung-Hui Hu, have written about the strong security measures, militaristic metaphors, and physical architectures that protect data, from kill switch to data bunkers.
Tech titans often have more access to and control over the fate of data than most other people. Brian Michael Murphy’s We the Dead: Preserving Data at the End of the World opens with the author’s visit to Iron Mountain’s National Data Center. One of the vaults contains massive amounts of priceless images owned by Bill Gates, who planned to digitize them all before he realized how expensive and time-consuming that process would be. The images are in a deep freeze to prevent them from decaying. Murphy uses this anecdote of a tech billionaire’s desire to preserve information to theorize what he calls the “data complex,” formed as corporations took on the work of creating, mining, and storing vast amounts. Preserving the data of the dead requires massive physical infrastructures and the extraction of natural resources. And it is often the wealthiest techie elite who have access to these.
Big Tech companies may be rethinking their promises of endless data storage and eternal memory keeping. Yet they are certainly investing in energy-intensive technologies like generative AI.
The large language models (LLMs) being released by companies including Microsoft, Google, and OpenAI, as Hugging Face researcher Sasha Luccioni describes, require massive amounts of planetary resources. These include “rare metals for manufacturing GPUs, water to cool huge data centers, energy to keep those data centers running 24/7 on a planetary scale”; moreover, producing them requires associated exploitative labor practices. Training powerful LLMs produces high levels of carbon emissions and also consumes other resources, including water. What kinds of resources will it take to maintain such systems over time, especially if generative AI is used to produce chatbots of the dead?
Some advocates and activists from within the tech industry have called for the formation of green software, or energy-efficient and carbon-aware software that lessens the environmental impact of data production and retention. Data deletion, in this context, can be seen as an ethical choice. Individual developers and end users can choose to delete and consolidate old files, but large companies have far more control over how resources are operationalized in the production and storage of data. There are ways of making software and data storage greener by using renewable energy sources, cleaning up old code, and deleting redundant files, and by timing energy-intensive activities, like training LLMs, for times of day and geographic locations when and where there is more renewable energy available.
Still, a handful of companies have direct control over the way most people’s data are maintained. There is no such thing as automated digital permanence, because there will always be maintenance and upkeep involved. And there is no heavenly cloud, even for the digital dead.
To some extent, there is a lingering fantasy that data are infinite and digitization affords endless storage. But data rely on resources, like water and energy, to exist and persist in perpetuity. Techno-solutions to impermanence are not viable, but perhaps there are ways of making the cloud—and, with it, the data of the dead—more sustainable.