Where is transparency an urgent and pervasive need in technology?
The recent election, concerns regarding the prevalance of fake news, corrupt personalization of information consumed by users of social media, we decided to research on the problems around credibility assessment during consumption of news media, and the prevalance of the filter bubble among millenials.
Duration: 14 weeks
(Sep - Dec 2017)
Photo Diary Study
I led the research phase of the project: identifying key areas for research, and conducting the user study. I also led the development of the interactive prototype.
Literature review revealed concerns regarding credibility assessment by consumers, the spread of fake news and the problem of filter bubble due to over-personalization.
News articles raised questions to major corporations such as Google and Facebook on the steps that are being taken to curb this. There is a growing urgency in the tech industry, but the solutions are insufficient.    
We consulted two expert researchers at the University of Michigan, to narrow down our scope, better understand user behavior in online spaces, and the challenges involved.
Social media has become a common source for information and news consumption for a large group of people, which can result in a filter bubble.
The source of people’s news consumption can become increasingly polarised based on current events on the political scene. 
How do we help the digital generation resist “corrupt personalization”? Millenials' source of information may result in a harmful filter bubble, and an inability to assess the credibility of the information that they consume.
We need to meet their particular news consumption needs – mobile, shareable, aspirational – in a critically reflective way.
They are digital first consumers and studies have indicated they struggle as much with credibility assessment (and filter bubble effects) as any other group, despite perceptions otherwise. Also they were an accessible population
To help us design a formative user study, we formulated a set of decision and the corresponding research questions. Research questions were classified as behavioral or attitudinal.
Since our questions were primarily behavioral, we identified a photo-diary study which allows the participant to highlight details of their personal lives directly, to be the most appropriate method.
We conducted a 1 week long photo diary study with 6 participants, and received 82 photos accompanied with captions as submissions. To motivate the participants, we compensated them with $15 on study completion (submission of 12+ photos across 4 different days) and an additional $5 for a follow-up interview.
To synthesize findings, we did an affinity analysis using the photos, notes by our participants, and our notes from the follow-up interviews.
Participants’ news media consumption is closely aligned to their social media activities.
Users primarily encounter news via personalized social media feeds. This occurs in a context where self exploration, self affirmation, and self expression is the primary objective.
Credibility assessment is not a top-of-mind activity. Our participants often used heuristics for credibility assessment.
Information is consumed in micro-periods, and engaging with newsmedia can be a passive activity. ("...in the morning (in bed)”, “...over breakfast”, “...between class”, “...before I go to sleep”)
Solution: Leverage social media to present information from outside the user’s social network
Solution: Offer news articles aggregated from different sources - the playlist metaphor
Solution: User Profile to provide details about the information being consumed
Solution: Provide details about the how the system presents articles to the user - opening the blackbox metaphor
Solution: Mobile based design
Pheme organizes your news into Playlists - leveraging ur social media feed to serveup collections of perspectivesyou might have missed.
Share better. One article is only one perspective. Share custom playlists from all angles of an issue.
Transparency matters. Tap the blackbox icon on any page to learn more about our system, a playlist, or the article that you are reading.
Our study participants got most of their news from social media, and it’s a completely personalized experience. Hence, to provde a bigger picture, it was essential to know what our users are not receiving.
Users indicated they trusted aggregated news solutions (from multiple sources) the most, but also found those solutions scarce. We've conceptualized playlists as group of articles organized to serve users with perspectives that they normally do not see. In addition, we don't present just articles on a topic from one source, but several, to allow the user to decide for themselves
In the age of recommender systems, there is a need for system transparency to promote user control and trust. We want the user to understand why and how Pheme presents information it does on each screen.
It is useful to share multiple perspectives on the same topic with one's social media network - rather than limiting the discussion to one point of view. This helps to facilitate discussion on a specific topic (via search) on social media.
As our concept is aiming for behavior change, it needs to enable for reflection, develop a personal connection, and access to system settings. For a user, this allows better understanding of one's media consumption habits, and Pheme’s impact on that over time.
We have translated research findings into design decisions -- do those design decisions, in turn, “translate” into an understandable value proposition and a usable system for our users? We conducted both an in person usability test, and do some lean remote user testing using a Qualtrics survey.
We recruited 4 participants for the in-person usability test. Our survey online received 19 responses over a period of 24 hours. The set of key questions along which we wanted to evaluate our design were primarily regarding the onboarding process and initial reactions to the homepage.
"Playlists" as an organizing principle for news requires additional onboarding support, or in app prompts / in-context onboarding.
Changes to the “Playlist” feature might influence presentation of the “home screen” experience. Users want to directly access the articles.
It is currently not very evident that the users can get daily push notifications or alerts for news. We should change the icon or adding labels to the icons.
Through this project, we wanted to see how we would research and attempt to solve a critical problem that is widely acknowledged to be pervasive in our media consumption. Our extensive literature research, and detailed formative qualitative study, helped us identify some key insights into users' news consumption behavior.
However, I feel that we struggled when it came to translating these findings into design concepts, as evidenced by some of the comments from our evaluation study, as well as the feedback we received from the instructor. We are hoping to revise our designs based on this feedback, and evaluate them further.