Is it any different nowadays? AFAIK the goal isn't to advance the field, but to prove to the people giving you the Bachelor/Master's degree that you're able to create a "paper". Some theses stand out, but most of them are probably quite boring.
It seems to me like their tasks were way heavier. For us, most of us did basic CRUDs with some implemention twist, and our papers were your standard: do a short theoretical introduction to the domain, make a few UML diagrams, explain how it works + some code examples, then testing.
