... Volcano^2.1

When referring to an application built in INDIE, we will frequently refer to it by a shortened, distinctive title for simplicity, such as ``Volcano'' for ``Volcano Investigator.''

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...[Bell 1996].^2.2

A full analysis of GBSB's success and failure be can be found in Chapter 7.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Our^2.3

When I say ``we'' and ``our'' in reference to INDIE's development, I mean the INDIE development team as a whole. The primary work on INDIE 1.0 and 2.0 was done by me, but vital support in both code and insight came from the rest of the INDIE development team who are listed in the Acknowledgments.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... scenario.^5.1

Volcano is a degenerate example of this as it currently has only one scenario, but as we'll see in other examples of INDIE projects, problem selection is more important than simply an introduction.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... gratitude.^5.2

Ruby doesn't express any gratitude since her condition is largely untreatable and her heart will be permanently damaged.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... INDIE.^9.1

The other three projects built in INDIE 1.0 were built with members of the INDIE development team acting as programmers, which required full-time effort on our part.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ``widget.''^9.2

Modgit should technically be spelled ``modgEt'' given the words we are connecting, and in some papers [Riesbeck and Dobson 1998,Dobson and Riesbeck 1998] we have listed it as such. After some consideration, however, modget looks more likely to have the stress on the wrong syllable (having the same foot as ``Gidget'' instead of having the same foot as ``Rockette,'' and rhyming with ``bit'' instead of with ``bet''), so we write it modgIt instead.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... points.^9.3

Simple applications like Volcano have one experimental notebook while more complex applications with lots of evidence points have multiple ones. Authors can associate points with any of the experimental notebooks and they will appear in the correct list in the interface when the student discovers them.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... plays.^9.4

This double-duty was a bit of disaster; see Section 8.2.2 for a discussion of both the problem a possible solution.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... users^9.5

In our case, this set of users are students using an INDIE-built application.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... finished^10.1

By ``almost finished' I mean that interviews were done within a month or two of the main authors being removed from the project. Frequently, projects would continue after interviews for a few more months by a subset of an authoring group to finish up cosmetic or other changes. Samir Desai, consultant for a major consulting firm once said, ``A multimedia project ends when the money runs out,'' and in the case of INDIE projects, this, too, was true, so completion dates are hard to pin down.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... finished^11.1

By ``almost finished' I mean that interviews were done within a month or two of the main authors being removed from the project. Frequently, projects would continue after interviews for a few more months by a subset of an authoring group to finish up cosmetic or other changes. Samir Desai, consultant for a major consulting firm once said, ``A multimedia project ends when the money runs out,'' and in the case of INDIE projects, this, too, was true, so completion dates are hard to pin down.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... often.^11.2

In response to this story and other author interviews, we added a facility for creating reports that look like spreadsheets for questions and answers so that authors can get new views on their data.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... INDIE^11.3

The ASK system was also a large portion of the model, but is little different than other ASK systems already built.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... evening,^11.4

Although, in the interview, this team also mentioned that the graveyard shifter found he worked better at night anyhow and this was at some level his preference.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... objects^11.5

The worst mistake on this part was naming evidence points ``firmers,'' since they could be confirmers or disconfirmers. I felt it was a good name because it implied they ``firmed up'' claims, but it was not only baffling to authors for its cuteness and obscurity, but they socially rejected the name as silly and called them ``bullets,'' which reflected their usual structure--a 49#49 followed by text.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... memory,^11.6

The fact that we considered this an ``authoring bug'' is because the version of the Macintosh operating system we were using didn't allow us to increase the overall size of the application while the program was running. Authors needed to quit, change the size of the application, and start it up again. This was a very common ``gotcha'' since authors on big projects needed to remember to increase their memory allocation each time they got a new copy of the tool.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... author.^13.1

At last look, GBSB only has 4 possible cases in its critiquing: if test 1 is complete, if test 2 is complete, if test 3 is complete, and if test 4 is complete. Thus, critiquing is done ignoring conjunctions of results. In the case of SCC, with only two tests and only one that produces useful results, this is probably a mostly valid simplification, but for more complex domains where each test gives a piece of the overall puzzle, this is unacceptable.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... make,^13.2

A good anecdote here comes from our first demonstration of Immunology to a Northwestern Medical School professor. Our demo path started with the simulated student saying to herself, ``Well, my patient has severe carditis. I wonder what causes carditis? I had better look it up in the `Research Immunology' section.'' The professor immediately shook his head. ``That's not what they'll do,'' he said. ``They all want to be doctors, so the first thing they will do is the thing that seems most doctor-like; they will want to interview the patient and see what she says.'' The teacher clearly had a strong insight on the first failing of medical students--act like a doctor before hitting the books and coming to the patient prepared.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.