Week five of HCI class. I left a bit early today to attend an ITRC meeting, so I missed the latter half of the class. Nonetheless, I talked to Marchionini about presenting the KMT and Helpsite to the class, and potentially setting them up as a test documentation group so that they might offer their feedback on the interface. In addition, if I treat this as a user study, I might be able to get Marchionini's feedback on that process, for future testing with students, faculty, etc.

Searching vs. Organizing:
We talked a bit more about Microsoft's search-based content-indexing research, known as "Stuff I've Seen". This would apparently work something like Google Desktop or Gmail, indexing all of the content in your system (files, email, bookmarks, etc.) and presenting it to you through a search interface. The trend here is, apparently, moving away from categorization and hierarchy as organizational tools and focusing instead on making better ways to add metadata to your content.

This trend in web technologies, from Internet powerhouses like Google and Microsoft, suggests that we ought to give this careful thought in developing the KMT and helpsite. Should we even bother maintaining the topic tree, if users are going to become more and more accustomed to searching as a primary means of finding files?

Fisheye Interfaces
We spent some time discussing the "Fisheye view" study done by G. Furnas. the fisheye concept concerns the aspect of human perception wherein we "perceive information in a 'context+detail' orientation. Objects at the center of our retina are at a high resolution, while objects at our periphery are seen in far less detail."

We looked at this largely as an example of a good HCI analysis, rooted in human psychology. My initial response, however, was prototype interfaces based on this model- such as the one linked above where menu text becomes larger when it's in-focus, and smaller when it's in the periphery- is inherently broken in terms of accessibility (in the disability sense). A nice widgit, but maybe not the best implementation.