Observation of Mental Models


Kurt seemed to enter the session with no particular model of the site or of how it should look. He was, perhaps, an unusual sort of user for the site, since he knows virtually nothing about its subject material, whereas most users are probably musicians looking for information, people who have at least some model of the domain, if not this site.

In the first few queries, Kurt found one or two things that worked repeatedly (the "information" and "images" areas) and stuck with those until they failed.

Only then did he notice the "search" option, which he then tried to use for many tasks (unless he remembered seeing something specifically relevant).

The "information/roland" directory mostly contains files about particular synthesizers, with standardized naming. For example, information about the JD-800 is in a file called roland.JD-800, and information about the D-50 is in roland.D-50. Additionally, there are certain lines about which there is so much data, they merit their own directory, for example the SH-101 and the Juno-106. These are simply named "SH-101" and "Juno" (no roland prefix). This is a clear naming inconsistency. On the first task, however, this didn't seem to pose a problem for Kurt, presumably because he had no model of how things should be named. He found "Juno" fairly quickly. On the seventh task, however, Kurt had presumably modelled the naming as being only of the "roland.SH-101" kind, and so when he didn't find an appropriately named file, he concluded that nothing relevant was available, missing entirely the "SH-101" subdirectory.

Kurt's trouble with the last task and in general with knowing what the appropriate first link to follow (broadly speaking, is something information, images, searchable, an interesting highlight, etc) suggests that things need to be more standardized from certain points of view. Some attempts to do this have been made but could be improved. The model of what goes where is not at all obvious.


Mike Perkowitz