INDIE only allows one copy of the GBS file to be active at a time. As with all alpha-tested software, INDIE developers encouraged authors to save ``early and often,'' and after a few scares and the occasional bit of data lost, authors tended to save several times a day; Rembrandt ran from 2 to 17 different versions of their data file each day while Nutrition, which was a larger program and took over a minute to save, tended to run around 4 to 6 versions a day.
Since only one person could work on INDIE at a time, authors tended to change their work habits so that they didn't need the data file at the same time; Rembrandt authors worked in shifts with one pair working during the day and the third author working largely in the evening,11.4 while Nutrition, Clinical Monitor, and Volcano authors tended to divide work so that one author was the main interface builder and other authors did work on the rest of the program. This meant that there was one main author who understood INDIE well and several subsidiary authors who tended to have a weaker grasp of how the interface mechanisms like triggers worked.
To mitigate bottlenecks in authoring, INDIE developers allowed authors to develop new material separate from the main data file that could be merged additively, although once merged the data could never be separated. Both Volcano and Rembrandt built their critiquers separately from the interface and merged their data in once their first draft was done. The Rembrandt critiquer author even built a simple interface so that she could test her rules, and then deleted this testing interface when she merged the data in.
An obvious suggestion for improvement would be a shared object database so that two authors at the same time could change objects at the same time, with restrictions to stop authors from working on the same object at the same time. Authors had differing opinions on how helpful that would be; although there were obvious cases where authors working on, say, the interface might have been separate from authors working on the ASK system, many authors suspected this would create more confusion that it would save time. Said one content developer, a veteran of many GBSes who had learned to trust no one but himself to keep the data coherent: ``I'm not going to let anyone mess with my database!''