User Research :

DebateCafe is an online deliberation platform that is designed to facilitate and scalably analyse the arguments of both sides of a complex issue. DebateCafe collects quantitative feedback on background questions related to the issue, encourages participants to enter strong arguments on both sides and rate how persuasive the other participant’s arguments are. It provides instant feedback by displaying histograms of quantitative responses and a balance score measuring the two arguments supporting contradicting views. The first exploration with DebateCafe was “Apple vs. FBI” issue asking if Apple should cooperate with FBI requests for personal data.

Designing the user interface :

DebateCafe consists of three stages: Assessment, Discussion, and Ideation. The first stage of the system helps gauge participant's initial position and open-mindedness to the sides of the argument; participants assign ratings to preliminary topics about privacy and security. The second and third stages of the system are a discussion space where participants evaluate the persuasiveness of peer-submitted arguments both for and against Apple's compliance with future FBI requests. Finally, each participant is encouraged to formulate two arguments of their own: one for each side of the argument. The overall quality of participant's contributions is based on the persuasiveness of both of their arguments.

Assessment :

To begin, participants assess their current security/privacy beliefs by rating three quantitative assessment topics (QATs) on a 10-point scale from 0 (Strongly Disagree) to 9 (Strongly Agree).

1. I am willing to give up some privacy for increased security.
2. Personal privacy should be guaranteed by the US Constitution.
3. There are reasonable arguments on both sides of the security and privacy debate.

These statements were chosen as they succinctly capture participants' initial stance on digital privacy, as well as the degree of open-mindedness to opposing arguments. Participants have the option to skip any question they choose not to answer by either hitting the skip button or leaving the response blank.

Once participants finish the assessment phase, they are asked to provide their zip code. We find zip code to be an informative demographic statistic, while not being so intrusive as to hinder further participation in the system.

Discussion Phase :

After submitting a zip code, participants enter the discussion space, where the central discussion question is: “In the future, should Apple cooperate with FBI requests for personal data?”.

Each participant’s response is represented with a sphere in a 2D space. The spheres are placed using Principal Component Analysis (PCA) applied to the grades of the initial three issues. Skipped issues are assigned the mean grade for that issue.

Participants click on the objects and open a rating interface with consistent design as in the assessment phase.

Participants evaluated these peer-submitted statements on the question "How persuasive is this argument?" again using a scale from 0 to 9.

After rating two arguments, participants are then encouraged to propose their own arguments.

Ideation :

When participants click on their own sphere, they are prompted to formulate their own arguments. Compared with previous CAFE instances, DebateCAFE is novel in that participants are prompted to enter two arguments: one for each side. To help bootstrap participant's writing, we pre-populate the textboxes with "Apple should cooperate because", and "Apple should not cooperate because." The interface strongly encourages, but does not require that a participant fill in both arguments. After entering the arguments, the platform instantly calculates the “balance score” of the participant. DebateCAFE also requests participants to supply their email address so that we can update participants with true peer-rating scores on their arguments.

Beta version testing :

To test our first version of the platform, we got students from UC Berkeley to use the platform and the team made observations on how the people interacted with the interface. We got data from 300 people which we then analysed to see the trends in the user arguments on the case. We found a lot of repetitive arguments made by students. In addition, most of the students had strong arguments for only 1 side of the case and hence they could not balance the arguments on both sides. Therefore, to check the working of the platform in optimal conditions, we decided to test DebateCafe amongst debate teams. We contacted debate team at UC Berkeley and high school debate teams in the Bay Area as we believed that they could give strong arguments for both sides of the case.

Brief :

DebateCAFE is a web platform encouraging citizen participation to form opinions, leading to idea generation and crowdsourcing solutions.

Role :

UX Design Researcher.

Purpose :

Worked under Professor Ken Goldberg for 10 months.

Design Process :

Researched different UX/UI website models to develop our modular components such as proposal making, deliberation and versioning and developed data visualizing opinion spaces.

Team :

Ken Goldberg, Camille Crittenden and Brandie Nonnecke.