Design for conflict resolution and empathy in Facebook

The overall process Facebook used for the project.
As discussed in a previous post, Social Media and Wellbeing - the Role of Design, the effects of social networking on wellbeing come in positive and negative forms and receive significant media attention.  We argue that a software environment as complex and multifaceted as a social media environment will always represent a complex web of positive and negative effects influenced by user characteristics, duration of use, and other use-case elements, and that it is the responsibility of designers and researchers to tease out wellbeing impact factors and work to design for the positive.

 Companies that take into account how their software can support wellbeing have an excellent business case for doing so: online communities in which people thrive attract more users and keep them engaged.  Facebook is among the largest companies that stands to benefit from designing systems that support positive social interaction, so it may not come as a surprise that since 2011 they have held an annual 'Compassion Research Day' (CRD) aimed at bringing together experts who can inform the way design of Facebook with the latest research on prosocial interaction.

A browse through their discussions and videos is instructive. The first CRD was held in December 2011 and brought Dacher Keltner and Emiliana Simon-Thomas, social psychologists from Berkeley's Center for the Greater Good.  Dachner who studies the social functions of emotions, including compassion and how it has evolved, described a project using Darwin’s ideas on emotional expressions to inform the design of Facebook features. Their idea is that improving the way we communicate emotions in Facebook can help us develop as social beings. Dachner presented stats of usage behaviours from people  around world--very interesting results that I hope to see published.

One of the talks I found most interesting was given by four Facebook Engineers who described a 6-12 month-long project that impelented a conflict resolution system intended to improve the process of reporting anti-social content.

Rather than porn or violence, most content users report have to do with images relating to them personally or opinions (see cloud map above).

Combining this data with data about users' preferred follow-up action,  the team implemented a framework that allowed for a more nuanced approach than just "remove content". (see cloud map below).  The approach recognised the limitations of ecological validity and explored new methodologies including real-time surveys with open and closed questions.

In one of the experiments they tested an interface design for people who receive a request to remove a photo. In particular they tested if reminding people about their relationship to or  commonalities with the person making the request improved outcomes. The basic tenet is that understanding where the other is coming from is a first step to conflict resolution.  The feature increased a feeling of similarity to the other person (ie. empathy) by 12%.

Metrics used included:
  • Messaging success and accuracy: Did the person remove the content or not, and was the reason for reporting appropriate or not.
  • Drop-off from a workflow (a measure based on behaviour) and 
  • User satisfaction
  • Volume of use was acknowledged as a problematic measure because more people going through the process might not be good, while fewer might mean the system is not working.
Very interesting work! We appreciate that these events are made public online.