Looks like now that OpenACS and .LRN are getting more and more features and optional packages we need to consider usability. Actually we should have been thinking about it all along, but at least we could claim there should be resources for it now.
One case is Assessment which provides almost infinite flexibility, with almost no clues how this should be used. An assessment author does not get advice or guidance from the system on how it should be used. A while ago Solution Grove and MGH started an effort to reduce the complexity of the admin interface to allow easy creation of the most common types of assessment.
https://openacs.org/xowiki/Assessment_Admin_UI
(Sorry the pages are out of date, I'll update them.)
I have recently been reading some usability guidelines for open source software, and I think they are very helpful.
http://humanized.com/weblog/2007/10/05/make_oss_humane/
http://www.catb.org/%7Eesr/writings/cups-horror.html
http://www.catb.org/%7Eesr/writings/luxury-part-deux.html
http://openusability.org/
http://developer.gnome.org/projects/gup/hig/2.0/index.html
etc...
Assessment is such a great example because it supports so many different things, and doesn't help the user decide which thing they might want.
The project we started to simplify the admin UI for assessment mainly focused on using reasonable defaults, by reducing the number of things the author has to think about, we can get a functional assessment more quickly without confusion.
The recent problem I had with the idea of "section points" is an example. Nowhere does the system explain what they are, why I would want them, or present a reasonable default for it. It's my guess the reasonable default for section points is equal weighting of the sections, each question is worth the number of points specified for each question. This is not the case, the default is if section points is not specified, the section is worth 0 points.
In my use, none of my users needs this feature and its quite confusing. Its probably simpler for most assessment authors to weight the individual questions since they need to already make that decision.
I invite all users of assessment to participate in this effort, I really don't understand most of the complex features of assessment, or the implementation, so I need help with this. The documentation does not explain how or why I would use these features and does not reflect the reality of the existing code in the package.
