Here comes the 3rd article of the 5-episode blog post series on Polis, written by Guido Berhörster, member of staff at my company Fre(i)e Software GmbH.
Enjoy also this read on Guido's work on Polis,
Mike
Table of Contents of the Blog Post Series
- Introduction
- Initial evaluation and adaptation
- Issues extending Polis and adjusting our goals (this article)
- Creating (a) new frontend(s) for Polis
- Current status and roadmap
Polis - Issues extending Polis and adjusting our Goals
After the initial implementation of limited branding support, user feedback and the involvement of an UX designer lead to the conclusion that we needed more far-reaching changes to the user interface in order to reduce visual clutter, rearrange and improve UI elements, and provide better integration with the websites in which conversations are embedded.
Challenges when visualizing Data in Polis
Polis visualizes groups using a spatial projection of users based on similarities in voting behavior and places them in two to five groups using a clustering algorithm. During our testing and evaluation users were rarely able to interpret the visualization and often intuitively made incorrect assumptions e.g. by associating the filled area of a group with its significance or size. After consultation with a member of the Multi-Agent Systems (MAS) Group at the University of Groningen we chose to temporarily replace the visualization offered by Polis with simple bar charts representing agreement or disagreement with statements of a group or the majority. We intend to revisit this and explore different forms of visualization at a later point in time.
The different factors playing into the weight attached to statements which determine the pseuodo-random order in which they are presented for voting (“comment routing”) proved difficult to explain to stakeholders and users and the admission of the ad-hoc and heuristic nature of the used algorithm1 by Polis’ authors lead to the decision to temporarily remove this feature. Instead, statements should be placed into three groups, namely
- metadata questions,
- seed statements,
- and participant statements
Statements should then be sorted by group but in a fully randomized order within the group so that metadata questions would be presented before seed statements which would be presented before participant’s statements. This simpler method was deemed sufficient for the scale of our pilot projects, however we intend to revisit this decision and explore different methods of “comment routing” in cooperation with our scientific partners at a later point in time.
An evaluation of the requirements for implementing mandatory authentication and adding support for additional authentication methods to Polis showed that significant changes to both the administration and participation frontend were needed due to a lack of an abstraction layer or extension mechanism and the current authentication providers being hardcoded in many parts of the code base.
A New Frontend is born: Particiapp
Based on the implementation details of the participation frontend, the invasive nature of the changes required, and the overhead of keeping up with active upstream development it became clear that a different, more flexible approach to development was needed. This ultimately lead to the creation of Particiapp, a new Open Source project providing the building blocks and necessary abstraction layers for rapid protoyping and experimentation with different fontends which are compatible with but independent from Polis.
-
Small, Christopher T., Bjorkegren, Michael, Erkkilä, Timo, Shaw, Lynette and Megill, Colin (2021). Polis: Scaling deliberation by mapping high dimensional opinion spaces. Recerca. Revista de Pensament i Anàlisi, 26(2), pp. 1-26. ↩