The Right to Be Forgotten and the Pitfalls of Averting
the Digital Dark Age

When the Internet Archive first became known, many observers were alarmed. What about the privacy implications? In July 1996, David Berreby encouraged his Slate readers to consider the “most embarrassing e-mail you ever wrote, available to anyone curious enough to go looking . . . As we’re encouraged to exult over the vast new volumes of information that are becoming easier and easier to capture, remember that the art of losing is also important to master.”16 Dan Gillmor similarly worried in September 1996 about the prospect of “every dumb thing [that] I’ve said on-line” being saved.17 A year later in March 1997, John Markoff worried that the bigger the web archive, the bigger the Big Brother problem would be.18 When the Wayback Machine launched in 2001, it provided relatively complete access to the Internet Archive collection. However, as a user needed a URL until the recent advent of limited keyword search, the Wayback Machine provided privacy through obscurity. The apocalyptic visions of 1996 did not come to pass. Both public and private people alike have been affected by the Internet Archive, and things may have been remembered that they wish were not.

Yet by 2010, there was also increasing concern about the impact of old, decontextualized information. Viktor Mayer-Schönberger’s 2009 Delete traced in part the collision between Web 2.0 (by 2001, as he notes, “users began realizing that the Internet wasn’t just a network to receive information, but one where you could produce and share information with your peers”)19 and the accessibility of web-based personal information on the web. At times Mayer-Schönberger overstates his points—the argument that “forgetting has become costly and difficult, while remembering is inexpensive and easy” is not true.20 However, his argument that “comprehensive digital memory represents an even more pernicious version of the digital panopticon. As much of what we say and do is stored and accessible through digital memory, our words and deeds may be judged not only by our present peers, but also by all our future ones” is profound.21 Reactions to this emerging trend have taken different shapes around the world.

The “right to be forgotten,” as a legal precedent, stemmed in part from a Court of Justice of the European Union ruling in the 2014 Google Spain decision. The right drew on deeper histories, concepts, and precedents. The case involved a Spanish man who was unhappy that a Google search for his name surfaced details of a government auction of his property. He felt this resolved situation was no longer relevant and impugned his reputation.22 The Court of Justice ruled in the man’s favor, holding that Google had a responsibility to balance its rights with those of users, giving the plaintiff the right of erasure.

This “right to be forgotten” influenced the European Union’s subsequent 2018 General Data Protection Regulation. GDPR enshrined the right of individuals to request that their data be erased for specific reasons (which then needed to be balanced against free speech rights). As of 2020, approximately 45,000 requests were lodged with regulators to have information delisted from search engines in Europe, of which just under half (43%) were deemed to be valid.23 Reflecting different legal and cultural traditions, many North American commentators reacted with surprise to these decisions and regulations. Would the “right to be forgotten” not limit freedom of expression?24 The leaders of the Wikimedia Foundation (which oversees Wikipedia) worried about these rulings creating the prospect of “an internet riddled with memory holes—places where inconvenient information simply disappears.”25 Would all the hard work of web archiving be undone if inconvenient documents could disappear through legal fiat?

Legal scholars Melanie Dulong de Rosnay and Andres Guadamuz argued that ultimately the legal requirements would have little to no impact on web archives, given the already existing policies and approaches to privacy.26 The Oakland Archive Policy already implicitly gave most people a right to be forgotten. Conversely, as much of the right to be forgotten involves delisting content, the combination of the lack of full-text search in most web archives as well as the relegation of national library legal deposit collections to on-site only access had already established a balance between privacy and access. As one does not need to delete content to conform with the right, only to make it less accessible, web archives were compliant.

Improving computational access may unsettle this balance. Some of the original ideas around access, such as treating archives as akin to census data, had tried to strike a balance between privacy and access. Since the Wayback Machine’s launch, most models have implicitly adopted a privacy-by-obscurity approach. In the case of national libraries, some have been required to restrict access to on-site reading rooms. On-site access, which often strikes observers as ludicrous—forcing a researcher to physically travel to sit at a computer to view networked resources—is a reasonable compromise between access and privacy. Sitting in a reading room on a special terminal, few readers would mistake archived webpages for live content. Context collapse can be avoided. Parallels can be drawn with the digitization of nondigital archival sources, which also raise ethical access questions.27 Giving access to researchers in a reading room is different from unfettered decontextualized access to anybody with a Google search query, even if both collections are, in theory, open. Conscious of the burden that travel places on many researchers, particularly those with limited travel funding or caregiver responsibilities, perhaps a similar outcome could be achieved through a Virtual Private Network (VPN). Some degree of friction is not a bad thing.

One fear of unfettered digital access is the prospect of an archived website or social media post being consumed out of context and used to shame somebody. For example, in my earlier archival work on 1960s student activists, I encountered documents that raised questions about how my reaction might have been different had these activists tweeted or even had all of their material digitized rather than having letters that ended up in archival boxes.28 One letter, from the McMaster University Archives and Special Collections, contained 1962 musings by a student activist in their late 20s who thought it would be “fantastic news” if nuclear-tipped Bomarc missiles (which were to be used over Canada in an anti-­aircraft defense strategy) could be redirected by the Soviet Union after launch to turn around and destroy their launch bases. This letter helped me understand the context of the New Left, presenting a vivid example of how the New Left was very different from the mainstream Canadian social democratic left. Yet the thought process of that letter in isolation is horrifying on the surface. It effectively mused about the deaths of Canadian military personnel and those nearby. But as part of a broader collection of correspondence, it illustrated the ways in which New Leftist intellectual thought was developing.

As a researcher sitting in an archive, I found the letter to be unremarkable. Yet if a keyword search for the author would surface it, readers might be less charitable sans context. Suddenly the need to travel to another city to visit an archive or log into a cumbersome VPN is less of an inconvenience but perhaps part of thoughtful, historical research. What if that letter had been a tweet or a blog post in a web archive? A student writing that today could become the target of an outrage campaign.

In 2016, my colleague Nick Ruest and I were bullish on the prospects on what social media meant for future historical research. “Consider what the scale of this dataset means,” we wrote. “Social and cultural historians will have access to the thoughts, behaviours, and activities of everyday people, the sorts of which are not generally preserved in the record.”29 I still stand behind this. Historians, professionally trained and responsible ones, conscientiously grappling with the past’s complexity, will try to understand the proper context of archival documents. When I discovered the 1962 Bomarc missile letter, I did not turn to a conservative Canadian media outlet, claiming to possess the smoking gun of the New Left’s moral degeneracy. Context matters: private correspondence, part of an ongoing debate around the adoption of nuclear weapons in Canada, by a young student exploring new ideas in a climate of intellectual exploration. I suspect few professional historians would think differently.

On balance, I believe the value of a society-wide “right to be remembered” outweighs in general the value of an individual’s right to be forgotten. This needs to be considered in a context of archival ethics and care. The compromise position of complicated on-site or remote access, requiring researcher registration, compels researchers to view and think about documents in context. By doing so, readers know that web archives are not just websites like any other on the live web. Furthermore, researchers cannot easily share the “gotcha” moment, flattening time as if a blog post from 2001 is the equivalent of one in 2021.

Historical processes unfold by virtue of human choices and decisions. The current situation when it comes to digital access was not inevitable. The international library community has generally taken a conscious choice toward expansive selection. In the case of the Internet Archive, it ultimately adopted relatively open access policies, but it handled privacy through obscurity. As institutions develop next-generation search and retrieval systems, the lessons of the past serve as concrete inspiration about alternate visions of web archival access.

The Little Digital Dark Age, 1991–1996

The direst predictions of a digital dark age were averted by the beginning of active web preservation in 1996. Yet there was a short dark age after all, between the advent of the web in 1991 and the development of memory institutions in 1996. In some ways, this ultimately formed the dreaded “digital gap” that commentators such as Danny Hillis had worried about. Seeing what we have lost can help us gain a better appreciation of what has been saved.

Not everything published on the web before late 1996 has been lost. Magazine articles, newspaper accounts, oral interviews, journal articles all help to reconstruct the earliest web. Digital forensics can as well with considerable effort. Tim Berners-Lee’s first website, for example, had been launched in December 1990 at CERN. By the time web archiving began in 1996, it had been converted into a museum site. An effort to reconstruct it drew not only on preserved files but also an array of other contemporary sources to recreate the browser experience as it might have been in 1990.30 Similarly, the first American webpage, the 1991 homepage of the SLAC National Accelerator Laboratory was restored from backup. Researchers needed to undertake a fairly involved process to convert an “original list of scattered files into an accessible and browsable website.”31 Apart from these rare exceptions, however, most of the actual websites from this period are lost. The labor needed to reconstruct sites like this does not scale. We can reconstruct a few significant sites with great effort and the fortune of either backups or contemporary documentation. Most sites are beyond this.

What have we lost? We can look to events that took place in 1995 and early 1996 to understand the gap. Before 1995, few corporations or people were on the web. Accordingly, 1995 to 1996 was the earliest period when the web was part of many people’s lives.

Political examples are perhaps the most obvious. Historians can look to significant events to then in turn see what remains. In October 1995, the Canadian province of Quebec held a referendum on whether the province should pursue sovereignty. Digital historian Ryan Deschamps conducted preliminary research into what a digital history of 1990s Canada would look like. He found that despite evidence from print media and Usenet discussion boards that both the pro-­independence and pro-separatist sides had important web presences, they were largely not preserved. Given the importance of the Quebec referendum to an understanding of modern Canadian history, this is a major loss. As Deschamps notes, “it was not clear to people in 1995 that web pages were historical documents that people might want to use for research in 2017 [or beyond].”32 While the Bibliothèque et Archives nationales du Québec does have webpages from 1995, the sovereignty campaign pages were lost.33

It is distressing to think of the material that was lost from the early web that we do not know about. Early cultural sites? Homepages? Jokes? Academic servers? Wikipedia’s “List of websites founded before 1995” is an overview of some of the kinds of early websites. As a crowd-sourced document, it is well positioned to draw on the collective memory of early web users.34 Science museums, web comics, religious movements, campus newsletters, business websites, all feature on this list. The Exploratorium science museum in San Francisco, for example, opened its website in 1992, but the earliest snapshot we can access today dates from January 1997.35 The Economist’s first website, launched in March 1994 by one of its correspondents, was reconfigured after eighteen months. As an Economist author bemoaned, “[a]ll records of the original website were subsequently lost. So much for the idea that the internet never forgets. It does.”36 Websites that lasted into 1996 or 1997 may have been preserved. Those that did not last that long, or were small enough that they eluded detection by web archives in their smaller early years, have not. If it was not for the people and institutions discussed in this book, we would have more stories like this. The gap would have been a lot longer. Ultimately, the Little Digital Dark Age of 1991–1996 illustrates how fortunate we are today.

Web archiving and digital preservation has come a long way between the early 1990s and 2001 and continues to develop today. The implications of this are still making themselves clear. As historians move into the web age of history, it will be important for them to know how the archives they use have been constructed and how they came into being. None of the web’s memory is natural or intrinsic to the platform itself. It has been painstakingly constructed, the product of countless conscious decisions across Silicon Valley, national capitals, and academic institutions. Today, the Internet Archive is increasingly part of the internet’s core infrastructure. As research libraries and national institutions preserve swaths of our cultural digital heritage online, it is critical to remember that none of this was inevitable. The web naturally forgets, and it is up to us to help it remember.