Military Families Learning Network

Wiertz Sebastien – Privacy used under Creative Commons license  http://www.flickr.com/photos/wiertz/6092000030/

When I wrote last year about Online privacy, the message was that our perceptions of online privacy revolve around the use of information we consider private or personal in a context we weren’t expecting. This post will take a look at how our offline behavior and information is being used to influence the advertising we see online – in particular, ads on Facebook.

Your first reaction may be that there’s no way that your Facebook identity can be connected to your offline activities, but that’s not correct. There are a large number of companies, called data brokers, that gather information from a variety of sources, and link that information to create a profile of an individual. Once these data brokers have an email address linked to an individual, they can use that to create targeted advertising campaigns through Facebook.

What kinds of information do data brokers collect?

To understand the kinds of information that data brokers collect, it’s instructive to take a look at the company Acxiom.  According to their document, Understanding Acxiom’s Marketing Products, Acxiom has both household and individual data, including name, address, telephone, email, gender, education level, occupation, voter party, date of birth, marital status, number of children in household, children’s age ranges, household interests, home owner status, home purchase date, home loan amount, home market value, and much more. This data comes from a variety of sources, including public records (marriage licenses, property transfer and tax records, etc.), self-reported survey information, purchase information, etc. Axciom then uses this information to provide services to its customers including targeted or addressable advertising. Axciom states that they don’t share sensitive data, that any individual record contains only a subset of data that they collect, and that data may be combined to create “inferred elements.”

How can this information be connected with my Facebook profile?

The Electronic Frontier Foundation EFF) recently wrote about the details of how data brokers are able to partner with Facebook to show you targeted ads. In brief, data brokers provide Facebook with a hash (a hash is derived summary of an original value that is not reversible, so the original value is obscured) of an email address for each user they’d like to see a particular ad. Facebook then compares that hash to the hashes of the email address of each Facebook user. When there is a match, the two parties can be confident that it is the same person, even though they didn’t share the actual email address with each other. Facebook is then able to present the purchased ad to the user. In turn, Facebook provides information back to the broker about the success of the ad and aggregate demographic information about the viewers.

A simple, contrived example

It may be easiest to get a sense of what’s happening through a fictitious example: A data broker would like to advertise dog food on Facebook, but only display that information to dog owners. In their dataset, the broker has stored publicly available dog license information and associated that with particular individuals whose email addresses they have also determined (through surveys or commercial entities.) The broker gives Facebook a list of hashed email addresses (they don’t share the actual email address) and Facebook compares that list to their own list of all hashed email addresses associated with Facebook accounts. The dog food ad is displayed to each user who is on both lists. So, even if a user has never posted about their dog on Facebook, they could see ads that are targeted to them based on offline information.

So what?

Is this a privacy violation? It likely depends on your perspective. Data brokers would contend that the information they gather is publicly available or shared by the individual. Facebook would contend that the resultant ads have greater relevance to the user, and are more desirable than displaying random ads to each user. The individual may find it creepy that Facebook appears to “know” about things that they did offline and did not intend to share with Facebook. The user may not have thought that licensing their pet would lead to them getting pet supply related ads on Facebook.

As technology makes the sharing and combining of this sort of data easier, we can expect to see more examples like this. I remember a conversation from 15 years ago with a friend that sold life insurance. He would hire a college student to go to the county records office and get the information on marriages and births, so he could send the people letters offering his services.  Now that large companies are combing through and digitizing these records, they are public in a way we may not be thinking.

How can I stop it?

The EFF article has information on how to opt out and what that really means. Unfortunately, there is no central clearinghouse where you can opt out from all data brokers at once, and opting out does not mean that data brokers will stop collecting your data. Opting out only affects how the data broker will use your data.

Sign of the times

Many people have had the experience with physical junk mail, of suddenly receiving a flood of mail related to something they’ve done, like receiving extended warranty offers after purchasing a car. It appears that online ads have become the analog of junk mail, targeting you based on information gleaned elsewhere. As long ago as 1999, Scott McNealy, then CEO of Sun Microsystems, was quoted, “You have zero privacy anyway…Get over it!”

The only things that are truly private are those things known only to you. Once others know our actions, behaviors, or information, it is no longer private and we are confronted with how that information is used and in what context. Controlling our expectations and how others use the information we leave scattered in our wake is a challenge we will continue to face.

- An excellent explanation of the technical aspects of the data broker / Facebook relationship – Security Now podcast Epsiode 404 with Steve Gibson from the TWIT Network.

 This work is licensed under a Creative Commons Attribution 3.0 Unported License.

Disco ball - used under Creative Commons license from http://www.flickr.com/photos/30946603@N04

Headlines about privacy breaches, companies gathering data without informing users, and companies changing privacy policies seem to be a daily occurrence. While future blog posts will look at what you, as a user of online tools, can do to protect your privacy, I wanted to step back and examine what we mean by privacy. It’s important to know what privacy means and how we perceive the privacy of certain information to understand how we want to handle such information.

To help get at the ideas behind privacy, let’s use a fictitious example:

Joe

Joe is a mid-career civil engineer and construction manager at a large firm. He earned his degree from a Big 10 university and was active in intramural athletics and his fraternity. He is married to Cathy and has three children.

Joe loves the song “Dancing Queen” by the Swedish band Abba, but worries about what others would think of him if they knew. While he’ll sing the song when he’s alone in the car, he hasn’t shared his feeling with anyone else.

Joe’s love of “Dancing Queen” is private. As long as he doesn’t share that information, it will remain private.

Let’s look at some ways Joe might share that information, along with some possible ramifications.

Joe tells Cathy

Suppose Joe tells his wife Cathy that he loves the song. Joe trusts and loves Cathy, and her knowing about “Dancing Queen” doesn’t change the way she feels for or behaves towards him. Therefore, there’s no direct harm from Cathy knowing his private information. Indeed, he could benefit, because now she may let him know when it’s playing on the radio, buy him an Abba album for his birthday, etc.

Cathy tells a friend

At a party, Cathy hears “Dancing Queen” playing and tells her best friend, Nell, who’s married to one of Joe’s co-workers, that it’s Joe’s favorite song, but not to tell anyone. If Nell doesn’t tell anyone else, then there is still no harm to Joe. However, suppose Nell mentions to her husband that Joe loves “Dancing Queen” and the next day, as a joke, he blasts the song on the loud speaker at the construction site. Joe may feel harmed, because the information he considered private and embarrassing is now known by his co-workers and is being used to poke fun at him.

Joe uses the Internet

Joe wants the lyrics to “Dancing Queen” so he goes to the Google website and searches for “Abba Dancing Queen lyrics.” He finds the lyrics, and also looks at some other sites related to Abba. He does a couple more searches for Abba information, and ends up buying an Abba compilation album from Amazon.com and downloads an MP3 of “Dancing Queen” from iTunes. Then he uses the online service Spotify, to listen to a bunch of Abba’s music.

Here we have moved away from the familiar interpersonal sharing of information, and are instead dealing with corporations and large databases. So what information has Joe shared, and what are the possible effects?

Because he did a bunch of searches for Abba-related information, it’s quite likely that Joe will begin seeing advertisements for Abba-related items on Google sites. The advertising services on the sites Joe visited probably set a cookie in his browser, so that when he goes to other sites they serve, they may present him with Abba-related advertisements. Amazon and iTunes may begin promoting to him items similar to the Abba album he purchased. Spotify will recommend similar music, and if he has connected Facebook and Spotify, his listening to the Abba music may be shared with his Facebook friends. In addition, there are other indirect players that know about his interest, including his Internet Service Provider and credit-card company.

The actions Joe took online have many diverse effects. Some may be viewed as positive, e.g., if Joe ends up seeing an ad for the Blu-Ray release of “Mamma Mia,” which features music by Abba. Other effects are neutral; e.g., the service or company having a record of Joe’s interest in a database somewhere has no measurable impact on Joe. But if his preference ends up being shared on a network like Facebook, Joe may view it as potentially harmful.

Use of information vs. possession

The above stories illustrate that having someone else simply possess information we think of as private is not the real problem. Instead, it is how that information is used. Does it change the possessor’s behavior or feelings? Does the possessor share that information with someone else? Are we even aware of how that information is used or shared? These are the real questions we should should take into account when we decide what we do with our private information.

The love of “Dancing Queen” is a silly example of information one might consider private, but it’s easy to see how this story might apply to other concerns such as health issues, financial information, and other important data. The only truly private information we have is information that we keep only to ourselves. Once it is known by others, we have to trust that they will use our information only in ways we wish.

The concerns about privacy are really concerns about what a person or company will do with our private information. Therefore, when we make a decision to share data we consider personal or private, we must consider how that information will be used. We trust (or don’t trust) other people based on our existing relationships with them and our expectations of what they will do with our information. Sometimes, but not often, we may have a written or spoken agreement about how our information will be used.

With companies, particularly online, we are dealing with entities that are typically, not a single person, and that we may not have dealt with before. We therefore rely on other factors before deciding to share personal data with them: reputation, transparency, written agreements, cost versus benefit, etc.

Consciously or not, each time we share information or browse sites, we weigh the risk that the information we’ve shared gets used beyond our desires compared to the perceived benefit we get from sharing the information or visiting the site.

Future blog posts will examine these issues in more detail, particularly the factors we consider when sharing information in an online space.

(If you’re not familiar with the song, you can view the music video of Dancing Queen on YouTube)

This work is licensed under a Creative Commons Attribution 3.0 Unported License.