In his novel Dans le cafe de la jeunesse perdue Patrick Modiano (Nobel Prize for literature 2014 ) sketches Louki . Never wanting to attract attention, at Cafe Conde', she always stayed in the background or hid herself among the crowd. But alas! in the rare and random group photos taken in cafe, she stands out because of her photogenic face. When she goes missing , Mr Caislly a detective is put on the job to trace her. At one point Mr Caisley feels morally bound to respect Louki's right to remain untraced. The story is set in the 1960's, when it was easy to remain hidden from public view. Fast forward to modern times. Do we have the privilege to remain anonymous? Can we wish to be among the "Forgotten People"? Well, I may be a total stranger to my next door neighbor but I leave digital footprints all over. Mobile phone, credit cards, browsing habits. everything leaves a mark. Oh! No not just marks, I generate a pattern of my behavioral traits- restaurants I frequent, favorite holiday spots, reading/browsing habits, favorite cab service.......So much so that often I receive helpful prompts : here is a book you may like or We haven't seen you for so long........ Such personalized messages are just one part, I might have inadvertently checked a tiny box somewhere sometime to "be in the loop".
But what about anonymized or ghost data? We are talking about millions of such digital footprints devoid of users' identification marks. Because of ease of data collection, consolidation, filtration and processing such ghost data( or metadata) , constitute rich gold mine for market research, R&D activities and also for meaningful studies in sociology behavioral science, public health, etc. How harmless are such ghost datasets ? Not quite that is what Anthony Tocker found out. Curious brains can mix/ match/superimpose ghost data with other pieces of information freely available in the public domain and breathe life into skeletons. For example when New York City Taxi and Limousine Commission put out rather innocuous details of millions of trips for the year 2013 on the web, (of course without passenger details) it was not at all difficult for Anthony Tockar to couple trip info with freely available public domain data and thus de-anonymize the riders. As simple as putting 2 and 2 together, but in a rather complex way.
But what about anonymized or ghost data? We are talking about millions of such digital footprints devoid of users' identification marks. Because of ease of data collection, consolidation, filtration and processing such ghost data( or metadata) , constitute rich gold mine for market research, R&D activities and also for meaningful studies in sociology behavioral science, public health, etc. How harmless are such ghost datasets ? Not quite that is what Anthony Tocker found out. Curious brains can mix/ match/superimpose ghost data with other pieces of information freely available in the public domain and breathe life into skeletons. For example when New York City Taxi and Limousine Commission put out rather innocuous details of millions of trips for the year 2013 on the web, (of course without passenger details) it was not at all difficult for Anthony Tockar to couple trip info with freely available public domain data and thus de-anonymize the riders. As simple as putting 2 and 2 together, but in a rather complex way.
de Montjoye, at the computational privacy at the Massachusetts Institute of Technology, prefers to describe the process as Correlation Attack. de Montjoye and his team decided to analyze credit card data in the same way. They collected ghost data for credit card transactions over a span of 3 months for 1.1 million people and 10,000 shops sans user name, sans card number , sans shop name and sans time of transaction. Armed with just the details of amount spent, shop type, date stamp. and a code for each person, the team demonstrated that " 4 spatiotemporal points are enough to uniquely reidentify 90% of individuals." and that " women are more identifiable than men in the credit card meta data."
Efforts are on to fix such loopholes. Anonymous search engine duckduckgo doesn't capture user's ip address or store search history. TrackMeNot an add-on to Firefox or Chrome throws the tracker off the track and into a Daedalian labyrinth with no Ariadne around to help. Researchers at Duke University have developed CacheCloak, a program which camouflages mobile location. Are we getting paranoid about our privacy? Not necessarily. In my mobile contact list there are very few without a face; some have uploaded their entire family! More interestingly there are monthly if not weekly updates. Facebook too tells the same story; people are eager to talk about themselves, share their experiences- holidays, parties, workplace, constant updates...... Taking advantage of the millions of authenticated photos in their treasury, Facebook is all set to roll out DeepFace, a face recognition program, which can automatically tag a face with 97.25% accuracy.
Perhaps Digital Age has precipitated our eternal paradox: to be unique and yet ordinary.
Tailpiece:
Efforts are on to fix such loopholes. Anonymous search engine duckduckgo doesn't capture user's ip address or store search history. TrackMeNot an add-on to Firefox or Chrome throws the tracker off the track and into a Daedalian labyrinth with no Ariadne around to help. Researchers at Duke University have developed CacheCloak, a program which camouflages mobile location. Are we getting paranoid about our privacy? Not necessarily. In my mobile contact list there are very few without a face; some have uploaded their entire family! More interestingly there are monthly if not weekly updates. Facebook too tells the same story; people are eager to talk about themselves, share their experiences- holidays, parties, workplace, constant updates...... Taking advantage of the millions of authenticated photos in their treasury, Facebook is all set to roll out DeepFace, a face recognition program, which can automatically tag a face with 97.25% accuracy.
Perhaps Digital Age has precipitated our eternal paradox: to be unique and yet ordinary.
Tailpiece:
Stars, hide your fires;
Let not light see my black and deep desires (Macbeth Act 1, Scene 4)
References:
..
1. Dans le Cafe de la jeunesse perdue : Patrick Modiano, Gallimard 2007
2.Differential Privacy : The basics
3.Riding with the stars passenger privacy in the nyc taxicab dataset
2.Differential Privacy : The basics
3.Riding with the stars passenger privacy in the nyc taxicab dataset
4. Unique in Shopping mall : On the reidentifiability of credit card meta data
de Montjoye et al Science 30 Jan. 2015, Vol. 347, issue 6221, pages 536-539
5. Hiding in plain sight, & Camouflaging searches in a sea of fake queries.
Jia You ibid page 500 & 502
6. Facebook will soon be able to ID you in any photo
6. Deep Face:Closing the gap to human level performance in face verification
5. Hiding in plain sight, & Camouflaging searches in a sea of fake queries.
Jia You ibid page 500 & 502
6. Facebook will soon be able to ID you in any photo
6. Deep Face:Closing the gap to human level performance in face verification
