Sarah Esther Williams

                                                                Wisdom, Discipline, Courage

LCC 6311 Final

Visualising Privacy in Online Social Networks

 

Update : 1/26/2007

   This project came out of my midterm proposal for this class and it's great to see that some of the ideas therein have been recently implemented by Facebook. My midterm proposal can be found here.

Overview

                Social Networking Websites have been adopted by the masses in recent years at an almost viral rate (See figure to the left courtesy of www.newscientisttech.com). The success of websites like Friendster, Tribe and Facebook has attracted the attention of the media, industry executives and researchers alike.  As individuals of all ages herd online to create profiles, hook up, connect with old friends and even find jobs on these digital networks, the user’s privacy has become an area of increasing concern.

                Recently, Facebook officers got a taste of exactly how sensitive some college students were about the privacy of their information. A few weeks ago, Facebook introduced what they called News Feeds and Mini Feeds. This involved taking snippets from your recent activities and placing that directly on the home page of your friends. What pictures you posted, what part of your profile you changed, your public comments to other friends could be seen by all the persons you were directly connected to.  Facebook users were outraged and genuinely alarmed about their privacy. As a result, Facebook rushed to provide privacy settings that could allow people to control what got placed in and out of their News Feed. Moreover, Facebook started to remind people more aggressively about their overall privacy settings – after all everything in the news feed.

                But what are the privacy settings like? They are a daunting jumble of text, checkboxes and radio buttons that allow you to determine your level of privacy. On some pages you scroll through two or three pages of settings and are sometimes unsure of what the impact would be on your privacy. No wonder most people just keep the default settings: according to a study done by Carnegie Mellon Researchers which analyzed how CMU students used the Facebook privacy settings, only 1.2% of the students (18 female and 45 male) made limited their profile’s searchability to their own institution. Additionally, only a meager 0.06% (3 profiles) left their profiles accessible to non-friends. An overwhelming majority of users kept the facebook defaults that rendered their profiles easily findable. [1]

                Why aren’t people making use of their privacy settings? It could be that the have inspected the defaults and agree with them. Another possible explanation is that they have never thought of looking at current settings or having made an attempt, found the experience too tedious. After the introduction of the newsfeed, I myself took a long look at the privacy page but was greeted by a disheartening heap of cluttered text and controls

                In light of the above compelling information, I would like to develop an interactive application, done perhaps with Flash, processing and/or the new FACEBOOK Developer API toolset) which enables members of the facebook social network to visualize they’re privacy settings. It is my hypothesis that a visual representation of settings will

·         Increase the user’s awareness of privacy issues such as stalking, re-identification and outside digital dossiers.

·         Motivate users to discover, understand and effectively use their privacy settings more often

·         Give users a greater feeling of control which in turn can lead them to feeling safer in their online environments.

·         Give them a better idea of what impact their privacy settings could have on their transparency on the web.

Target Audience

                I am targeting facebook users of all ages. Today, the facebook privacy settings very crudely presented to the user and, I suspect, hastily put together in response to user complains. This project will give users the opportunity to visualize their privacy settings in a way that will motivate and not deter them from examining their privacy settings.

Visualisation Requirements

I consulted two papers on privacy within the facebook space and one paper on social data analysis :

  1.      Information Revelation and Privacy in Online Social Networks ,  ACM Workshop on Privacy in the Electronic Society (WPES), 2005

  2.     Imagined Communities- Awareness, Information Sharing, and Privacy on the Facebook, PET 2006 

  3.     Designing for Social Data Analysis, IEEE Transactions on Visualization and Computer Graphics, Vol 12, No. 4, 2006 

Paper A, which examined the online behaviors of 4000 Carnegie Mellon University students, showed that

  1.     A minimal percentage of users change the highly permeable privacy preferences.

  2.     However, reasons for this behavior were unknown: Users could have inspected settings and found them to be desirable, or users could have changed their settings then changed them back to the default. Additionally, researchers did not know whether or not facebook users' privacy settings remained unchanged because of a lack of knowledge about the settings or simply a lack of interest or concern about privacy. Finally, peer pressure or herding behavior, faith in the community and its members, and lack of knowledge about privacy risks could all influence persons to reveal more and more information.

  3.     The relation between privacy and a person's social network is multi-faceted. In certain situations, persons want to reveal information about themselves to a small circle of close friends and not  to strangers. On the other hand, persons may be wiling to expose certain sides of themselves to strangers while hiding the same from those who know them better.

  4.     Regardless of what the user's offline social circle looks like, one can boast hundreds of 'friends' and "include hundreds of thousands of additional friends within just three degrees of separation." [1]

  5.       According to a cited paper, " [real privacy] resides in selectively revealing private information to certain individuals, but not to others." [2]

  6.       Only 1.2% of the students (18 female and 45 male) made limited their profile’s searchability to their own institution. Additionally, only a meager 0.06% (3 profiles) left their profiles inaccessible to non-friends. An overwhelming majority of users kept the facebook defaults that rendered their profiles easily findable.

Paper B, a follow up study to paper A showed that

  1.     Members have a false impression of the network's actual size and composition and of the extent the their profiles are visible to others.

  2.     Facebook's offers granular and powerful control over profile searchability and visibility but default settings render the profile very permeable.

  3.     There is a marked difference between attitudes and behavior : having a strong concern about a certain privacy issue did not affect how much information most participants revealed. For example, of the participants who were very concerned about a stranger knowing their schedule of classes, and where they lived, 22%, provided at least their address, and almost 40% provided their schedule of classes.

  4.     About 1/4 of facebook users incorrectly assumed that nobody on facebook could search for their profile.

  5.     More than half of the persons who didn't change their default privacy settings guessed that tend of thousands of users could search for their profile when in fact millions can. 

  6.     Individuals are not  in general concerned about the information itself but  trust in their ability to control that information and who can access it.

I consulted Paper C because it  presented a visualization that not only successfully drew in its target audience (parents looking to name their babies) but also engaged persons not interested in the topic of the visualization : baby names.

After consulting Paper C, I was able to determine a few requirements and NON requirements of the interface:

  1.     According to the paper, four reasons why the Name Voyager interface was so successful were a) it's public, b) strongly social nature, the fact that c) it afforded extended exploration of the data and finally that d) it featured a "game-like sense of fun".
    This let me to believe that, to achieve similar results, my visualisation would have to support c) and d). The interface should support exploration and the discovery of useful information that one may not have necessarily been hunting for. Because of the confidential nature of one's privacy settings, a) and b) may not be desirable side effects thought b) may be conceivable.

  2.     Ben Shneiderman's famous mantra of "overview first, zoom and filter, details on demand" should be followed.

  3.      The NameVoyager's interface featured smooth transitions between states on the screen and engaging animation which enhanced user experience. My visualization should, if necessary included meaningful animations. However animations included should not be excessively distracting to the main task of understanding a user's network, profile visibility and searchability.

  4.      One important component of social data analysis is its ability to foster conversation. I am hoping that my visualization will be able to foster healthy discourse on the topic of privacy in online communities.

  5.      Paper C shows that Richard Bartle's popular classification of user roles in online multiplayer environments as achievers, socializers, explorers and killers fits map almost directly onto roles users of the name voyager seem to be taking on. It is conceivable that persons may use the proposed visualizations in the socializer and killer roles, use in the achievers and explorer roles are more desirable. (See Paper C for more in depth role descriptions)

  6.      According to the paper, social data analysis is achieved by giving users unique perspectives on some common ground, supporting spectator friendly interfaces, and allowing for transfer of your discoveries. Ideally, the proposed visualization would support giving users a unique perspective on their social network, after all, no two friend networks are the same. However, the designing for the spectator is not the goal of this exercise. Finally, discovery transfer  would be a useful feature of the visualization, however, it would have to be done in a way that keeps the user's privacy settings confidential.

In light of the above requirements and background studies, what should be visualized?

  1.     The number of persons two degrees away from you (your friends of friends) and what their primary networks are. (Though only some friends of friends can see your profile). While the number of your friends grows linearly, your friend of friend network grows at a much much larger rate. Quantifying this value can help users make better decisions about their privacy.

  2.     The total number of persons who can view your profile and how this number changes over time.

  3.    The number of profile elements exposed to whom. Since users seem to rely on the fact that they can control was is revealed, the interface should support making changes to a persons privacy settings as information is displayed to them.

Range of Supported Tasks

  1.     The viewing of the number of people that can see your profile and how fast that number grows.

  2.     Changing your exposure levels to friends, friends of friends and everyone.

Possible Designs

  1.     The design of a game in which users are allowed to make changes to their profile and based on how many items are exposed and on how many people can see their profile, a graphic image changes from full clothed (less visible /searchable profile) to half naked (very open profile) 

  2.      A Visualization in which all of a users networks are displayed as visual bubbles that grow in proportion to the network size.

  3.     Use of real time Bar graphs and to show the history of the growth of the network. 

Visualization

         My first attempt at an engaging visualization was Idea number 2 above and was tested on a few college students. The general consensus was "Yes, I know the network is growing, but this does not compel me to change my settings". Though, I never thought of actually implementing idea 1, I decided to ask users how they felt about it anyway. They unanimously responded favorably to the idea of having a visual metaphor to communicate their exposure. Whether that metaphor is a the analogy to a full clothed versus naked person or whether it pertains to a peeled orange, users found that to be the quickest and most intuitive way to get their attention. Future work could include a mixture of the numbers and network growth visualization from idea 2 and the metaphorical avatars from idea 1.

Visualisation Files are here.

Possible Negative Impact of Proposed Visualization

  1.     If people share screen shots of their privacy visualization, others may be able to infer whether or not they are within that person's trusted circle and thus become offended.

Suggestions for Facebook

  1.     Privacy settings should be more restrictive rather than permissive as a default.  This forces users to at least inspect the settings before opening up their profile in ways they may not fully understand.

  2.      Facebook's new News Feed feature may have caused an uproar, but it can be used to publicize changes in Facebook's privacy policy. Currently the News Feed, features updates that other friends have made to their profile or snippets about other Facebook activity. Very seldomly, ads are slipped into these updates. Facebook can insert information into the feed regarding updates to the privacy policy so that users are continually reminded about the policy that governs Facebook Privacy and how it affects them.



[1] Acquisiti, Allessandro and Gross, Ralph, Information Revelation and Privacy in Online Social Networks (The Facebook Case). In Proceedings of the ACM Workshop on Privacy in the Electronic Society (WPES), 2005

 [2] S. Gernstein. Intimacy and privacy. In F. G. Schoeman, editor, Philosophical Dimensions of Privacy: An Anthology. Cambridge University Press, Cambridge, UK, 1984