The trade-off of data mining

 

   Minneapolis, Minnesota. It’s been about four years since Charles Duhitt wrote on the NY Times one of the cold Midwest hottest stories: a proud Minnesotan father discovered that his high-school-aged daughter was pregnant. On that, no big news. Only, he got to learn it from the targeted ads for baby products he received from a supermarket chain – quite ironically- named Target, an early success of data mining. Data mining is the practice of extracting information and models from enormous quantities of stored information. It’s not a novelty that companies follow and analyse consumer’s behavior patterns. Their models are accurate, as the example of this teen mom shows: a couple day after furiously storming into the supermarket, accusing its managers of providing incentives for precocious motherhood by sending baby products ads in a teen’s mailbox, the father called back apologetically: he was about to become a grand-father.

   One could sensibly argue that concerns of data mining have nothing to do with a family which seems in need of parental counseling or a couple sex-talks. True. Thing is, though, we’re just as predictable as that Minnesotan girl, and given the enormous amount of information we share everyday – knowingly or unknowingly- we are increasingly exposed to this evolution of this market practice.

« Mass data collection doesn’t stop at supermarket chains. »

   Mass data collection doesn’t stop at supermarket chains. Recently, Volvo trucks launched a similar massive, data gathering program aimed at predicting driving accidents. With fancy names such as Forward collision warning or Lane keeping support, the company initiated an enormous system with variety of ultraperforming sensors simultaneously keeping track of the state of traffic, the truck’s parameters the conductors’s patterns of driving. If functioning well, softwares should be able to deduce situations in which it’s very likely that the truck is going to crash. You could hear in 50 years time BEEP. Please stop the car, you’re about to have an accident. Your body parameters indicate that you’re likely to fall asleep behind the wheel. Great, Volvo Motors saved your life tonight.

   Statistically, of course. You haven’t actually had your accident. But the analysis of that 70 billion megas made it doable to predict your crash. Just as your daughter’s pregnancy was, after her storing credit card transactions, purchases over decades of shopping lists, as well as email exchanges, browsing history and all the rest.

   Great, isn’t it? Metadata saves your life and reveals the secrets your parental instinct had surrendered to the dark meanders of repressed and subconscious. Yes and no. Without falling into false alarmism, let’s briefly linger on one point. Programs simply gather, store, sort information, were it traffic situation, your favorite toothpaste formula, physical attributes of your three ex-boyfriends or the books you’ve read in the past ten years… and produce predictive models. Possibly, a designed program could suggest your next partner and choose more wisely and precisely than you. Then continue with wedding companies before you’ve proposed, eventually lawyers before your lover has asked for divorce. Let’s face it, predictability applies to our most private spheres. Similarly, given the newspaper articles you access, the petitions you sign, the political campaigns you support, you’re statistically politically modelisable. It can be determined which candidate you vote for, just as whether you’re a potential dissident. Of course, if you read this today, you’re living in a democracy, you don’t need to be scared. But one’s never too cautious with that.

« Good or bad, one must be aware that programs at such scale are not necessarily helpful nor innocuous. »

   Mass data programs can save us the hardships of shopping dilemmas and will soon be able to save us from car crashes and heartbreaks too. But they can also facilitate irregular market practices, derive in intrusive procedures or deeply affect the political debate in a civil society. Good or bad, one must be aware that programs at such scale are not necessarily helpful nor innocuous. After all, it’s all about the beneficts we could obtain and the freedom we’re willing to sacrifice.

Maud Barret Bertelloni

Laisser un commentaire

Votre adresse de messagerie ne sera pas publiée. Les champs obligatoires sont indiqués avec *