<i>The features don’t matter. It’s all about the outcomes. What can people do with your software?</i><p>I've seen software with the opposite problem-- simulation tools applied to problems outside the range they were designed to handle. For example, a tool that idealizes gases in a room as well-mixed, used to analyze smoke movement during a fire. Or a program that assumes ducts have no air leaks, used to estimate energy losses in a duct system. In cases like these, the results aren't total garbage, but they do have to be interpreted very carefully.<p>For some software, a feature list isn't just "what the software does." It's also a list of warnings about where the results should be taken with a grain of salt.
"The creators of Garageband for iPad didn’t care about what their software could do. They cared about what people could do with their software."<p>Aren't they the same thing? Feature lists describe what your software can do, which shapes what people can do with your software.<p>That GarageBand for iPad is accessible to all might be more to do with the platform than the featureset.