Versions 1 through 6

When I started working on the field filter, I made a list of all the schools and departments on all eight IU campuses. I then tried to create a "bottom-up" organization system, but I wasn't satisfied with what I came up with.

After five versions based on taking the departments and trying to organize them into established categories like science, social science and humanities, there were still about 15 fields and departments that weren't easy to categorize.

I decided I was going about this the wrong way.

The reason I started from all the departments was to try to make the controlled vocabulary robust, so it could accommodate the fields of future projects. Rather than have an ad hoc system for adding new fields as new projects are added, I want the fields to be ready for use when they are needed.

What occurred to me Thursday was that all those departments are already in schools. 

By starting at the school level, putting comparable schools from each campus together, I was able to create a sixth version of the controlled vocabulary for fields. And I like it. School level is still too broad for some areas (humanities, sciences, social sciences), but most schools are already narrow enough to work well. For the others, I want to expand them (science -> biology, chemistry, earth science, physics and astronomy).

Matching schools across campuses.

I realized that"top-down" is the best way to create this particular controlled vocabulary and organization scheme because:

1. When staff need to add a new project, they will know what school or department they were working with. Rather than artificial groupings, by using the organizational structure already in place at the university, it will be easier to categorize the projects.
2. If someone from a department or school is looking for projects that are in their field or a related field, they will have an easier time picking the "correct" field by being able to associate it with the school they are in.

I also used a few of the sample projects as test cases, listing what fields each would be tagged with. The differences between version three (the previous version I liked best) and version six are subtle if they exist at all. But the sample size is small, with biases toward particular fields. The big differences are in the fields that aren't as well represented in my sample. (The biggest difference is actually just the organizational structure and not trying to shoehorn fields into categories.)

I feel pretty good with version six. Which is good, because I need to present a solution at a meeting on Monday.

I also made a preliminary wireframe of the homepage based on feedback from the previous meeting and implementing the filters I have been working on. 

Takeaway: Sometimes you have to start over. Sometimes you have to step back and rethink. And it always helps to keep the audience and end users in mind, because that helps inform more usable choices. (Sometimes the best choice isn't the objectively "correct" choice.)