I never used Director, so I can't speak to that. And, by '96, I had graduated from HyperCard into C and Pascal.
But: I spent a lot of time in HyperCard, and I can assure you that the field of HyperCard complexity was not limited to just "simple back and forth stacks" or "anything truly useful beyond that". HyperCard afforded a really nice way of gradually ramping up complexity and approaching complex-on-the-outside problems with simple-on-the-inside code.
I doubt I could remember any of my own HyperCard projects at this point if my life depended on it. But, I can tell you about a HyperCard project a good friend of mine did: he called it "MusicMaker", and it came with a piano keyboard, multiple synthesizers, and its own unique musical notation system which allowed you to easily translate any sheet music into text which the HyperCard stack could synthesize and play on the piano keys. It could teach people music better than just about any other piece of software at the time. This was not, as I recall, remarkably challenging for him, and we both had a lot of fun with it.
I think any network-related programming is going to suck. Tricks of the Mac Game Programming Gurus had an entire chapter devoted to it, as I recall, in which most of the chapter could be summed up as, "OpenTransport sucks". So, I don't think it's fair to use a networking-related program as an example of a challenging HyperCard (or SuperCard) project.
> I appreciate your romantic notions to the contrary however.
(edited for snark)
Please try to keep the snark to a minimum. Thanks.