Category Archives: Big Data

Does Disney screen for sex offenders? Should they?

I’m sitting here with a 4 day park pass from Disney I purchased in 2000.  It has two days used and two days remaining. I’ve been reluctant to use it because, I know Disney will want me to convert it into one of their less anonymous park passes, requiring either biometric identification or my name.  Disney has been marching towards full scale identification of it’s park visitors for years. I still remember the days in the 70’s when park tickets consisted of coupons for rides in the various theme sections of the park. In 1977 they began issuing 2 day park passes to resort guest (and this was our family vacation so that’s how we rolled) and shortly thereafter began selling all access passports. This quickly became the norm, eliminating the per ride coupon books by the end of 1981, and thus began their need to track patrons for fraud purposes. I call this the “anti-fraud surveillance” business model.

I remember the last time I used this ticket I have in 2001, the ticket taker at the entrance suggested I take the ticket to customer service after I entered to have it upgraded. I politely ignored them. It’s not that I haven’t been to Disney since then but I haven’t had an opportunity to use this particular ticket.  I’ve been wanting to go again recently, if nothing else to get use of this ticket which I paid for so many years ago and have been holding onto.  However, if I go, it may be my last. Disney is getting too creepy for me.  I did get to thinking though, if Disney continues on the road to identifying and tracking guests, will they start screening for sex offenders?

To date I have no knowledge that they do so for their guests, though they do for their employees. Disney does have a problem with sex offenders on their property.  However, the problem isn’t that Disney is overrun with offenders, quite the contrary.  They have a public perception problem. Because they are geared towards children, every incident becomes a public relations fiasco.

Screening for sex offenders is difficult. There are lots of false positives and many more false negatives as registrants find ways of skirting the system. Even given the heightened scrutiny that Disney is under, I think they would be reluctant to embark on such an offensive. However, as they collect more data about their visitors, they may be inclined to use correlation data to screen and monitor guests. Single male spending too much time around It’s a Small World?  Group of teens going from shop to shop but not riding the rides? False profiling is something that is real and problematic. Of course, it’s something I know nothing about.

 

Wal-Mart does not “know what’s up”

Wal-Mart knows what's up meme.  A friend recently shared this image on Facebook.  The image which appeared to have quite a few shares is meant to imply that Wal-Mart is aware enough to know that Beer Pong enthusiasts will be buying lots of ping pong balls to go with their red Silo cups (wait, where are the kegs, Wal-Mart?).  The truth is Wal-Mart probably has know idea why people who by red Silo cups at their store also buy ping pongs. The do know that it happens though because Big Data analysis tells them so.  So what does Wal-Mart do in response? They put the two oft purchased items together to increase the sales.  They assume that many people want to buy these items together and so by placing them together they will increase sales.

What I love about this example, is it is a good use of Big Data which doesn’t necessarily implicate privacy issues.  They don’t need to track individual purchases or purchasers, they only need to know that there is a correlation between cups and balls.  They don’t need to pry into people’s lives as to why they purchase these together, but simply that they do.  However, some risk remains if Big Data turns into Big Brother.  If they did make the connection, could they require ID of purchases ping pong balls along with Silo cups? Sorry you must be 21 to purchase these items together.

Maybe in retrospect Wal-Mart does know what is up (the correlation of purchases) but not why (causation). For more information on Big Data, I suggest the Mayer-Schonberger and Cukier book of the same name.

 

 

Facebook and Real Names and social circle segmentation

At the IAPP Global Summit in Washington, D.C., Jules Polonetsky (@JulesPolonetsky) conducted a public discussion with Facebook Chief Privacy 542842_10102843702374493_485191411_nOfficer Erin Egan.  During the audience Q&A portion of the discussion, I posed two question: essentially what does Facebook do to ensure its developers are assuring that contextual clues help the Facebook audience know what information is being shared and with whom and secondly, why does Facebook insist on a real names policy despite the fact that there exist a clear minority of it’s audience that reject the idea.

I’ll save you an analysis of the response to the first question which essentially amounted to context is important and our developers know that. The response to the second question, though, bears further investigation. Erin answered, essentially, that its a means to encourage good community standards; that being anonymous or pseudononymous on the Internet leads (or allows) people to engage in behavior that they, shall we say, wouldn’t want their mother to see them doing. Jules chimed in that at AOL they saw rampant disregard for social norms due to the pseudononymous nature of that forum. I later approached Jules and suggested that, while the pseudononyms may play a role, another factor may have contributed more to AOL’s raucous nature. Unlike Facebook, AOL was primarily based on public forums (chat rooms, bulletin boards, groups). Facebook, though it has some of those features, is primarily based on private forums: private messages, postings on friend’s walls. The public forums do exists but they are largely an ancillary service to Facebook’s primary use (sharing old high school photos).  I would put out the hypothesis that this is the major contributing factor to people being on their best behavior.  If they are obnoxious, rude, crude, or otherwise inappropriate, users have the ability to ban those people from their private spaces (ignore their posts, unfriend them or block them).  Even the public spaces generally have moderators that can remove unwanted visitors.

Facebook is perhaps the ultimate big data company. I would suggest Facebook researchers (they have those right?) do some data analysis on how many adverse reports they get of people in public spaces versus private spaces.  Do users mostly avail themselves of self help (unfriending) or resort to reporting to Facebook? Of those complaints, how many of the users appear to be using pseudonyms and how many appear to be using real names? Inquiring minds want to know. If, as I suspect, the public spaces are much more rife with complaints and pseudononymous users, then perhaps Facebook could require real names for access to public content as opposed to the private spaces.

Many people have justifiable reasons not to use their real names. A one size fits all policy is not appropriate for a space of 1 billion users (*cough cough*).  In the real world, while we use our real names, people engage in social circle segmentation. What I tell my doctor I don’t tell my neighbor. What information I give to my boss may be different than the picture I paint and my kid’s little league game. In those environments, context plays a role in allowing us to socially segment our acquaintances into circles of what we share. While concepts like Google Circles and Facebook Smart Lists allow people to segment their audiences in those platforms, this is often difficult and mentally taxing for people to do. Easier is to segment their friends either on different platform (Facebook for school friends, LinkedIn for professional contacts, Google for online friends, Twitter for ….well it varies by the person).  Pseudonyms on platforms allows for a quick brain response of who am I right now and who is my audience. I don’t have to worry about my boss seeing the picture of my with the lampshade on my head at a party.  Each of my social circles is in a nice distinct bucket. Just some food for thought, Facebook.

 

Predictive policing

At 7:15 this morning I was rudely awaken by a police SWAT team banging on the door. I’m currently in a cold northeastern city visiting a friend (whom I happened to take to the airport last night to fly to my home state of Florida). He offered to let me stay here for a few days until I return to D.C. It’s a great savings of a few hundred dollars in hotel nights and the solitude has given me an opportunity to concentrate on some much needed work. Of course, solitude is not exactly what I had this morning. First there was a knock. As I peered bleary eyed out the window to see if it was an obnoxious solicitor, the knock grew furious. “Police, Open Up” was the shout. I scurried towards the door in only underwear and a tshirt. I opened it to approximately 10-15 police officers in full gear (bullet proof vest, helmets, guns). I stated to the officer at the door (who clearly recognized that I wasn’t whom they were looking for) that I was a house guest. He showed me a picture and asked if I recognized the man and I said no. He gave me his card and ask me to have the resident (my friend) call him.

I passed the information on to my friend who called the detective and spoke at length. Apparently, this is not the first time his house had been visited by the police. The detective explained that the suspect, wanted in connection with a shooting, and his family were listing this address as theirs. My friend explained that he had been there for 3 months and the owner of the house, who previously lived there, had been there many years. The detective offered to email my friend the picture of the suspect and asked to be contacted if he saw him in the neighborhood.

My friend called me back to discuss the incident and we discussed in light of the book I had been reading the previous day while my friend was at the house. That book was Big Data by Viktor Mayer-Schonberger and Kenneth Cukier. Predictive policing, not quite like Minority Report, is the use of big data style analysis for policing. The concept is fairly straight-forword, amalgamate large amounts of information relevant to criminal behavior and find connections that were heretofore unidentifiable. While arrests won’t be made as a result of predictive policing, suspicious actors could be uncovered and scrutinized thereby improving the efficiency of the police department. The risk, however, is having innocent associations place certain members of the population under enhanced scrutiny while others commit crimes. In the old days, this was called profiling and while dispassionate data analysis could be beneficial in removing stereotypes and biases from policing, the risks remains of being caught in a associative bucket of bad guys. My friend, who innocently occupies an address picked by criminals, now potentially will be forever associated with them. Will his car be pulled over more often then not, as police hope to catch him in the act? What other subtle things will threaten his peaceable right to be let alone now? Will credit reporting agencies ding his credit score because he shared an address with a family of criminals?

My ex-girlfriend used to carry her social security card in her wallet, much to my dismay. I pleaded with her not to but her retort was that she had no credit history worthy of stealing so what was the risk? She had a somewhat legitimate need as her drivers license had a different name that her birth certificate, due to custody battle and judges decree when she was just a toddler. She used the SSN as an alternative proof of her name, when her license didn’t match. It is an unfortunately byproduct of living in a society that is hellbent on using identity as a means of security. But the risk to her, were clear. What happens when her identity is stolen for criminal purposes? Or when a criminal uses her identity to commit a violent crime and her name is now tied as an alias to that criminal? While law enforcement making contact with John Smith may do a double take before arresting him on an outstanding warrant, her unique name would not be so lucky.

While efficiencies in the competitive industry of ferreting out criminals is a goal worth pursuing, appropriate safeguards must be in place to not make unwarranted connections. Further, oppression, warrantless searches, identity tattoos (ala WWII germany) make policing efficient but that doesn’t make them ethical. Society must weigh the political repercussions before embarking on the use of big data in this realm.