« Back to Projects

Facebook Roles

UW HCDE 505 Computer Mediated Communication


With a small team in UW's HCDE 505, I conducted extensive academic literature research to understand the problem space of social media context collapse, then proposed a role-based solution accompanied by a detailed paper justifying its design.

A mock-up of an interface designed to allow users to efficiently sort their contacts by the user's own roles.


The class requirements asked us to come up with a project that involved computer-mediated communication (CMC) to some extent. I proposed to the class a project to investigate the sharing habits of users on Facebook – how users decide who to share posts with – and was approached my classmates who were interested in forming a group. Our assignment was to conduct thorough academic research in the problem space, then develop a design proposal that leveraged CMC principles.


The first step was for each team member to read several academic research articles. I personally found the ACM Digital Library to be a fantastic resource for finding CMC research articles. In particular, we focused on identifying the concerns that social media users have with online privacy, specifically scoped to voluntary disclosure (as opposed to compromised back-end systems). We each wrote an annotated bibliography for at least 8 articles, summarizing the main findings of the articles and relating them to our project. As a team, we synthesized the lessons from the articles and identified main themes to address in our project.

In our literature review, we found that many users faced the problem of context collapse, a situation in which users' traditionally-separate social contexts (work, family, separate friend groups, etc.) are aggregated together in a single place (Marwick & boyd, 2011). This leads to individuals having difficult using their offline social techniques for dealing with multiple audiences. This has implications like accidentally disclosing private information to the wrong people or experiencing their separate contacts interacting – their worlds collide.

We read a lot of amazing research articles.
A sample annotated bibliography entry for one research article.

Once we better understood the problem space, we brainstormed ideas for how to alleviate some of the pressures of context collapse. We noted the dramatic difference between LinkedIn and Facebook interactions, wondering how Facebook might be able to capture the same kind of professional interactions. We also noted that some users even set up separate Facebook profiles to keep their contacts separate, a practice that Facebook discourages.

I proposed the idea of having separate identities or "roles" united under a single profile. Under this model, users could have a single Facebook profile, built specifically with very defined boundaries to the extent that the roles felt like entirely separate profiles, just as users might currently imagine their LinkedIn and Facebook profiles. We felt that this solution would help resolve the issue of context collapse while also fitting in with Facebook's business model by theoretically driving more interactions on Facebook (instead of on a separate social network).

Though Facebook currently has Friend Lists which could be used for the same purpose, academic research showed that the feature was underutilized; we believed this was because of its positioning as a secondary feature rather than a core organizational tool.

We envisioned a system by which users could mirror their offline roles in the digital environment of Facebook, separately articulating the details they would share in each respective context. For example, while an individual might share their religion with their family, they would withhold it from their co-workers; in this proposed system, a user could more easily say "my family sees this version of me, while my co-workers see this version."

The "All Friends" model of friends. Most users don't use any kind of organization of their friends, so all their friends are mixed together.
The "Friends List" model. Users can currently organize their friends into lists. Unfortunately, this model requires the user to think about their friends in certain groups.
The proposed "Roles" model. Users think of the roles they play themselves; we posit that it's easier to imagine one's own roles and then think about the versions of themselves that other individuals know.

We noted in our research that many users found the overhead of assigning individuals to Friend Lists to be too laborious. In order for our proposed solution to work, we wanted to make sure that the act of defining which contacts know which role would be as effortless (in terms of time and cognitive load) as possible.

With these ideas in mind, we began sketching overall feature flows and UIs. The main features that we designed were:

  1. Setup and management of roles, including providing profile information exclusive to a single role.
  2. A friend sorting UI in which users could drag identity 'cards' to their respective associated roles. This approach was inspired by a study that found a card-sorting method to be fastest for sorting friends (Kelley et al., 2011). Since we wanted to minimize the overhead of setting up roles, we felt that this was an appropriate and fun method.
  3. A method for gathering what we called "cookie-crumb data," or bits of data that are small but valuable and gathered opportunistically. After users shared a post, for example, we could include a question about a contact's identity that the user could answer in a couple seconds. We felt that over time, this method would collect data that would strengthen the role paradigm.

Ultimately, each of the user's roles would be represented as an entirely separate newsfeed. We made efforts to set the context of each newsfeed, such as by including images of that role's context and a subtle user-chosen background that brings to mind the offline environment (e.g. downtown buildings to suggest the user's work environment).

Whiteboarding the role setup experience.
Whiteboarding the Friend sorting tool.


Ultimately, we produced a series of medium-fidelity designs that walked through each of the features key to this role-based model. We also prepared a presentation to share the research and project with the class.

Most importantly, we wrote a 27-page paper detailing the entire process, from our initial academic research to the rationale for our design decisions.

A mock-up of an interface for defining multiple roles on Facebook.
A mock-up of an interface designed to allow users to efficiently sort their contacts by the user's own roles.
A single-role newsfeed, designed to filter content and provide visual cues that strengthen the context.
An interface designed to allow quick and effortless gathering of small bits of information – "cookie-crumb data" – to enhance the user's ultimate experience.

« Next: Wag-n-Purr Pet Care

Previous: Fulcrum CATS Mobility Refresh »