Model development

There are three major sections of the model unique to INDIE^11.3 and each should be discussed individually:

• Experiment phase: Only Rembrandt and Nutrition got to the point where they needed more than one scenario, which obviated the need for most authors to use any of INDIE's built in experiment modules.

• Critiquing phase: The three teams who developed extensive critiquers were able to start writing rules in less than a week, although critiquer design was a major issue and required redesign, as discussed earlier. Authors (especially those of Volcano, which was the first project to use the critiquer after it had been designed for Rembrandt) needed help expressing complex thoughts such as ``This rules requires three of this type, five of this type, or anything but these four points.''

• Scenario management: Once again, Rembrandt and Nutrition were able to build more than one scenario and manage them, even though Nutrition and Rembrandt had different techniques for moving from one scenario to another--Rembrandt allowed students to choose among any of three scenarios while Nutrition showed the students each case sequentially. As the controls for switching scenarios were buried in model actions, authors usually needed direct intervention from the programming team to engineer scenario changes, and almost all had trouble with resetting the initial state of the interface between scenarios. We clearly need to come up with a solution, such as a meta-model action called ``Change scenario'' which is an ordered collection of the most common scenario-changing model actions.