Abstract
Throughout Delete, Viktor Mayer-Schönberger provides us with examples of how our digitized thoughts and experiences can betray us long after organic memories of these episodes have faded. We are aware of these horror-stories, yet we continue to sign away our personal information to search algorithms and redundantly backed-up data-centres, leaving us vulnerable to the internet’s acontextually perfect memory. Why do we do this? Because we appreciate that social networks and the broader networks that comprise the internet hinge on easy information sharing.
Google’s recent actions illustrate well our paradoxical relationship with online privacy. In June 2011 Google+ was launched and touted as a social network with user-determined privacy at its core. Google’s latest attempt at social networking sought to exploit the erosion of trust in Facebook’s ever-changing sharing settings. But only 9 months later, Google changed its terms of service to enable greater information-sharing between its own services. Servers in Palo Alto can now amalgamate our search histories with the sites we subscribe to, the videos we watch and even our GPS coordinates, to ‘create a beautifully simple, intuitive user experience’ (Google, 2012). On the one hand, many of us rail against the unnecessary sharing of personal information, but on the other, we all find it easiest to use online services that have been highly customized to our needs with information we have (unwittingly) provided.
The majority of Delete focuses on building Mayer-Schönberger’s case for his solution. The characters who illustrate particularly interesting facets of memory function will be familiar to those who have read popular science memory texts such as Joshua Foer’s Moonwalking with Einstein (2011). For example, we are presented with Graham Bell, a man who is chronicling his life using SenseCam but who nevertheless keeps his meticulously catalogued information private. We also encounter hypermnesics, real and fictional, who struggle to make sense of the world without the smoothing of inconsistent memory into manageable coherence. Of the more technologically focused cases, we are presented with people who have been adversely affected by the online publication of photographs, archiving of emails, and even the digitization of a scholarly manuscript. All of which, Mayer-Schönberger believes, should lead us to the conclusion that something akin to the human model of memory is the model we should adopt for our digital data. In the final chapters, he gives us his solution: when we provide personal information, we should provide a use-by date, after which that information should be ‘forgotten’.
Mayer-Schönberger’s solution is the culmination of his extended and well-argued discussion of online privacy. It is therefore worth exploring in greater detail than the rest of the book.
This proposal, to introduce user-determined digital forgetting, is touted as straightforward but I suspect it would be impossible to implement. One of the greatest barriers would be user compliance. We don’t take kindly to the interruption of everyday computing tasks, as Microsoft discovered when they introduced User Account Control (UAC) to Windows Vista. This compulsory feature dimmed the screen whenever applications were about to be installed, alerting users to the potential dangers of such an operation and making them confirm that they wished to proceed. It speaks volumes that Microsoft’s next operating system Windows 7 made UAC optional – if a safeguard that requires only one additional mouse click per application installation is too cumbersome to justify the increase in security awareness, how would we cope with a safeguard that requires the specification of a date or time period every time we enter information accessed by an online server?
Integral to this compliance barrier is the nature of the proposed forgetting. Human forgetting is as frustrating as it is liberating, precisely because it is involuntary. Mayer-Schönberger is right to shine the spotlight on its oft-ignored benefits, but risks ignoring how our acquisitive tendencies shape our behaviours. Once again, Google provide us with an excellent barometer of public opinion. Gmail changed the way we deal with email, encouraging us to ‘archive instead of delete’ and enabled this by providing an unprecedented amount of storage for free. It is difficult to see how the service’s 350 million users would take to a system that encourages voluntary memory deletion in spite of Google’s continued pride in its provision of storage capacity that negates the need to delete at all – a counter just shy of 8 GB on Gmail’s login screen ticks upwards as users login to access their email. Should this solution be implemented, it is likely that many users would set the automatic deletion date to its upper limit and attempt to circumvent deletion nearer the time (by printing onto paper?). After all, we are used to measuring ourselves by our wealth. This wealth can take many forms, money, collections, health or, as any student or academic will tell you, acquired knowledge. It is hard to envisage giving up our memories, little pieces of our selves, to solve a problem few recognize.
There is a much larger debate to be had on the suggestion that digital memory could even ‘mimic human memory.’ Mayer-Schönberger suggests a rusting memory that undergoes analogue degradation, with sections of documents and pieces of information losing fidelity with time. But is the author’s call for memory mimicry intended for only the errors of omission made by our memories? Misattribution, suggestibility and bias, errors of commission, comprise three of Daniel Schacter’s (2001) seven sins of memory, and to be successfully implemented, would require not just the degradation of data, but their modification. These errors would lead to much of our digital data being inaccurate and volatile, returning incorrect or inconsistent information depending on the manner in which they were recovered. Memory errors of omission and commission work together to blunt our painful memories and enhance the consistency of those memories we do retain, end-results that Mayer-Schönberger is willing to suggest would result from a more human digital memory. Thus, it would seem appropriate that they be incorporated into a human-like digital memory system. But would this sort of digital memory take us too far from the accuracy and constancy of the digital memory systems we currently rely on?
In the Afterword, Mayer-Schönberger states that ‘[i]f Delete facilitates the discussion it has achieved its goal’. If aiding discussion of the problem was what the author intended, then he will be very pleased with his work. It provides some fascinating historical context for today’s reliance on digital memory, and perhaps most importantly, gives the reader an insight into how difficult it is to suggest a viable solution to the ever-growing invasion of digital privacy. But, if the author intended discussion of his solution with a view to its adoption, then there is much more to be done. Beyond settling on any one solution, I suspect the greatest challenge lies in convincing the general public that a problem, never mind a solution, even exists. Collective awareness may only reach critical mass when we see a large-scale invasion of public privacy with unambiguously bad consequences for those affected. By then, for many, it will be too late.