Paper: C. Ailie Fraser, Mira Dontcheva, Holger Winnemöller, Sheryl Ehrlich, and Scott Klemmer. 2016. DiscoverySpace: Suggesting Actions in Complex Software. In Proceedings of the 2016 ACM Conference on Designing Interactive Systems (DIS ’16), 1221–1232. https://doi.org/10.1145/2901790.2901849
Summary: In this paper, the authors introduce an extension to Adobe Photoshop, called DiscoverySpace, that provides high-level suggestions based on visual features of an image to help onboard new users. These suggestions/actions are drawn from an online user community. There are several problems users face when using a complex system for the first time, such as unfamiliarity with jargon, ease of use of tutorials, and the fact that many tasks can be accomplished through different routes. User studies showed that DiscoverSpace reduce the overhead in introducing new users to Photoshop, and suggest steps that can be replaced with advanced algorithms to speed up this process in the future.
Reflection: The paper tackles a crucial problem for novice users: providing an easy-to-use on-boarding process that simultaneously teaches them how to use the system without scaring them away with to much information. DiscoverySpace is an initial attempt to address this problem by learning from previous users and shows how we can combine AI and user interface design to build effective on-boarding tools for complex systems.
While this system was used specifically for Photoshop, I wonder how a similar approach can be used for other systems that many administrators and creatives use. For example, tax software, payroll processing software, video editing tools, etc. I also wonder if such approaches using data mining can be universalized, or if it is dependent on the user, context, and tool. Sometimes all a user wants to do is crop an image, and not do something more complex. Other times, users may already be following an online tutorial, after searching for the one they like. Perhaps tools like DiscoverySpace should allow for other users to create their own on-boarding workflows instead of having a fixed one for all users. This is because, as mentioned, creative tools allow for multiple flows that can lead to the same output, and some may make more sense to users than others.
Finally, I really appreciate the discussion section. in this paper, since it presents ideas for designing a more universal toolkit. While we can’t build a dataset of all possible actions for complex tools and creative processes, it would still help.
Questions:
1. Do you think everyone should undergo the same on-boarding process when using creative tools, or be given the choice to go through different pathways?
2. Why are these tools so complex? How can we provide more features without introducing more complexity into the information architecture?
3. What are some drawbacks to this approach?
