Is there any way, without modifying/hacking the core, to provide users with better and more meaningful feedback when something goes wrong?

...