The paper for this session was “Is This What You Meant? Promoting Listening on the Web with Reflect,” a work from University of Washington, Seattle researchers Kriplean et al. In this work, they set out to design a system to improve online discussions by enabling active listening.
Active listening is a crucial part of in-person conversation. Body language, short utterances, and a give-and-take pattern of speech allow for mutual understanding. This is a process known as grounding. These techniques are all difficult to perform in an online setting. As anyone who has engaged in an online discussion can confirm, these environments can often turn toxic very quickly. The authors of this paper hope to improve this through the design of their system. Reflect, the result of the authors’ work, enables listening through restatement. The unique system devised overcomes some of the difficulties present in current features of online discussion: threaded forums and “like” buttons. While both of these features could possibly enable active listening, they also can add complexity that conflates issues. Threaded replies are often expected to add new content to the discussion in addition to providing evidence of active listening, and “like” buttons add judgment as well as listening confirmation. This is why the Reflect system is useful.
The paper does an excellent job of walking through a use case of the system (see Figure 2), so I will merely summarize the main workflow of the system. In reply to a post, other users can post what they hear the original author saying. This reply is linked to sentences in the original post that support this summarization. The original author can then verify these comments, posting any necessary classificatory statements whose necessity was revealed by the comment.
This system was shown to be rather effective toward the goal of enabling active listening. In the testing of the system in Slashdot discussions, an environment described by the authors as “often vitriolic,” participants wrote comments that restated points in a largely neutral fashion. As the authors explain, this is a good indication that their system improves online discussions in the way that they intended. In addition, Reflect was found to cause participants to both consider more deeply the meanings of comments read and write comments that succinctly summarized long passages. This clearly fills a gap in the current implementation of online discussions.
I found this paper to be fascinating. While internet discussions usually devolve horrifically and quickly, I have personally always been impressed by the comments section on Reddit. With the threaded replies and moderation through “upvotes,” the comments can often provide fascinating insights. I had never thought about the additional implications that these techniques impose on online discussions. I would be very interested to try Reflect and see how it realizes the incredible potential of online discussion.