I had a sudden hypothesis after reading this thread last night. (No proof, mind you, just an idea.) I wonder if HyperCard's success was for the same reason as UNIX's success: a well-chosen, small set of interoperable tools that allowed the user to do far more than the sum of their parts. And, as computing and interfaces have become more complex, we've just tacked on complexity to them rather than reinventing them to adapt.

For instance, UNIX's treatment of everything as a file is a great tool, but I think there was a time where that simplified a much greater percentage of computing than it would today. Similarly, I suspect HyperCard died because the abstractions it was using had to be reimagined to stay competitive.

Spreadsheets (as mentioned elsewhere) are an example of a product that has managed to stay focused on the simple set of tools, and, when features were tacked on, they often stay out of the way instead of adding complexity to day-to-day operations. However, I'm not sure they've aged well, but I guess they are the best we have at the moment.