Chaiwut Chaianuchittrakul

Master's Thesis

December 2013

In this decade, Internet users already make up one-third of the world's population. Numerous web services collect, use and share users' personal information, such as location data and search history. Personal information is used to track and identify unique users in order to customize services for each user. Currently, privacy policies are the only tools that inform users about the flow of data and help users make privacy practice decisions. However, past research shows that privacy policies tend to be difficult and time-consuming to read. Building on the previous research regarding approaches that facilitate user comprehension of privacy policies, including Platform for Privacy Preferences (P3P) and Privacy Nutrition Labels, this project seeks to develop a privacy policy interface that is more concise and user-friendly. We explore the potential of using crowdsourcing techniques to help improve usability and simplify complexity of privacy policies. To evaluate small segments of privacy policies, our research uses Amazon Mechanical Turk, a crowdsourcing online marketplace. Through asking users about the comfort, difficulty and importance of the individual segments of privacy policies, we can identify surprising, difficult and important segments in each privacy policy. In total, the experiment resulted in eight trials with very different privacy policies. Unlike previous findings, which show that the privacy policies are difficult and confusing, our results suggest that a majority of users think almost all segments are understandable and important but do not raise any privacy concerns. This result shows that users think the context of a privacy policy is important. The result also suggests surprising facts that users can understand privacy policies and the privacy policies we evaluated did not raise any concerns.

crowdsourcing, privacy, usable privacy and security