delete: The Virtue of Forgetting in the Digital Age - Book Review
Updated: Nov 17, 2020
delete: The Virtue of Forgetting in the Digital Age
By Viktor Mayer-Schonberger
There is an often misquoted phrase from Bill Gates, that one will never need more than 640 kilobytes of Random-Access Memory (RAM). RAM is the active memory of a computer while it is on but purges itself when the computer is turned off. Bill Gates never said this, but such a phrase is a statement of the potential thinking at the time. Machines were a lot simpler than in terms of its practical applications but also resource demands; no one could have imagined the exponential growth from then to now.
For example, today, the average computer probably has around 8192 kilobytes of RAM, which is over a thousand-fold increase over 640 kilobytes. Readily available, cost-effective technology upgrades and continual processing upgrades, in RAM and also permanent storage drives and capabilities as well, are essential to Viktor Mayer-Shoneberger's argument.
He argues we have lost one of the more vital flaws but even perhaps our most significant strengths of human cognition, the ability to forget as we transition from analog memory to digital. And with that loss and the permanent archival of data, our actions and thinking patterns are altered but also have negative consequences--a panopticon, of sorts, where seemingly trivial messages and responses come to haunt us later due to perfect, digital memory.
Chapter one starts with two examples of the deleterious effects of permanent memory. The first is a prior schoolteacher, who, upon starting her career, found herself terminated because of a possibly suggestive picture involving alcohol on a personal social media website. The second was a psychotherapist who was not allowed into the country because of an academic article where he admitted to ingesting a psychoactive drug. Mayer-Schönberger uses these two examples to set the tone for the rest of the book. But he also expands on the idea of perfect and absolute memory. Regardless of how one might stand on the morality of these two individual choices, Mayer-Schönberger elucidates that once those possible mistakes find themselves incorporated into the digital memory, that inscription will remain, possibly forever—which affects the users' decisions and thinking.
Chapter 2 and 3
In chapters two and three, Mayer-Schönberger argues that with the analog way of thinking, forgetting was something we fought. The written letter, scribes, speech, and oral traditions served as tools to preserve the forgetfulness of our minds. Therefore, our default mode, however, was to remember what we tried to remember. So undoubtedly, things that we did not want to remember were forgotten, giving us a second chance to learn and grow without long-term stigmatization. However, as the world transitioned to the digital, perfect, and all remembering memory, such positive net benefits were lost.
Further, he argues that with long term data collection, more powerful inferences may find unintended uses. For example, in the Information Security domain, collecting logged information is vital, but only if you have many sources of information. Using databases and other technologies, one could generate reports aggregated volumetric data to spot security weaknesses. Still, this analogy could find use in Mayer-Schönberger applications as well in terms of negative use cases. Such as a citizen score that takes into account how we behave in society.
Chapter 4 and 5
In chapter four, he starts to develop and expand on the power of data collection over time. One piece of information is insignificant. But aggregated data could be analyzed to suss out parallels and inferences. However, in this chapter, Mayer-Schönberger explains what the real danger in all of this is, and he identifies two points. First, it is a loss of power to the ones generating all this data. This power loss manifests in several ways, but chief among them is that digital information copies easily and potentially used out of context. Second, and perhaps his most poignant thought, with total recall of information, there is a real threat to our thought processes, including reason and abstract thinking (Mayer-Schönberger, 2009, p. 123). Before wrapping up his arguments and final closing thoughts in chapter five, he argues that a solution to the problems he has much elucidated is to use a time-based expiration mechanism. Some data would be short-lived, while others like family photos might persist for a very long time.
There are two main sticky points of Mayer-Schnoberger's argument. The first is that the assumptions of that vast amount of data gathered are easily accessible to do the ills that he suggests that mass data collection is capable. For example, the credit reporting agencies and Facebook have a tremendous amount of data on American citizens. However, to common knowledge, those repositories are not linked together in any practical way. So, in other words, most of the information is siloed to each respective platform. Even with the existence of data clearinghouses, there remains to be a single pipeline of generative data that would make what he argues possible, at least at mass scale.
The other point of contention is his proposed solution of using expiry dates on the information. It sounds good on paper, but the application would be terrifically difficult. For example, as data passes through a network, at most stages, the data is undiscernible; it is only later as the information is reassembled and inspected are the actual contents discernable. So, such a system would rely heavily on the user to specify the expiry time. Such granular control may be needed from time to time, for everything else, it would be very cumbersome. The user would eventually bypass it for the sake of convenience. For this idea to work, you would need an analogous pipe of data where one could add 'tags' to pieces of or sets of data. The bitcoin blockchain may have been worth investigating by the author as a means to carry out his suggestion. Bitcoin and its blockchain technology already have implementations for Smart Contracts, which could be repurposed or expanded upon to fit the author's idea of expiration dates on digital data.
What the author does exceptionally well is spurring the discussion. For example, since the time of the book's writing, in 2009, California has passed its landmark privacy bill, the CCPA, which gives the user more control over the data they generate. Before California, the European Union passed its landmark law, the GPDR, that aims at the same thing. Both enshrined in 2018. So, the needle is moving in the right direction. And Mayer-Schonberger has helped. It's entirely plausible that in the future, the United States will have a patchwork of state privacy laws or national regulation. However, what remains to materialize is if such laws will stem the tide of the gluttonous data collection. Overall, a very worthwhile read.