Quantification, in the form of “Big Data,” is the subject of Viktor Mayer-Schonberger and Kenneth Cukier’s initially more celebratory book, Big Data: A Revolution That Will Transform How We Live, Work and Think. Now that we can collect and analyze vast quantities of data — often, all or nearly all the relevant data rather than just statistical samples — wonderful things can happen. Google predicts the spread of flu in near-real-time by analyzing searches; engineers foresee the failure of engine parts that wirelessly phone home; and Walmart notices that, just before a hurricane hits, sales of Pop Tarts increase. That there is a certain bathos to the progression of these examples is to be expected in an era that does not differentiate too pedantically between what is good for business and what is good for people.
Mayer-Schonberger and Cukier laudably demolish some of the more ludicrous big-data fantasies — for example, the claim by former Wired editor Chris Anderson that big data in science means “the end of theory” — but they also choose not to draw some arguably important distinctions. Is there, perhaps, a difference between “data” and “information” and “knowledge?” Might it be useful to distinguish between which articles a computer program has determined are “popular” on the Internet, and which are actually worth reading?
The dark side of big data, according to the authors, lies in surveillance — in communist East Germany, they point out, the Stasi were aspiring big-data fanatics — and in the alarming prospect of Minority Report-style pre-emptive policing. (According to a study too recent to make it into either book, Facebook “likes” can already be used to accurately predict “sexual orientation, ethnicity, religious and political views, personality traits, intelligence, happiness, use of addictive substances, parental separation, age and gender.”) We could sleepwalk into a “dictatorship of data,” where the algorithms mining the data for actionable recommendations are inscrutable and unaccountable “black boxes.” So, these authors conclude, we need a new cadre of “algorithmists,” people who scrutinize code for its obscured political choices.
Morozov, too, calls for “algorithmic auditors.” (Already, he points out, a single Californian company determines automatically what will count as hate speech and obscenity in the comment systems of thousands of Web sites.) More imaginatively, he also points out many possible consequences of the social engineers’ techno-fixes. “Would self-driving cars,” he wonders pointedly, “result in inferior public transportation as more people took up driving?” If you can measure and upload your health, diet and fitness data to be “shared” with insurance companies, then you’ll get cheaper insurance, say Mayer-Schonberger and Cukier cheerfully. But wait, says Morozov, such individual decisions don’t take place in a vacuum: “If I choose to track and publicize my health, and you choose not to, then sooner or later your decision to do nothing might be seen as tacit acknowledgment that you have something to hide.” These are, then, social and political problems, and ones that the mantra of individual choice cyborgized through shiny new technologies will often answer in ways that harm the already vulnerable.