Is mixed-signal simulation a fantasy?

The world is a mixed-signal one, or at least that's what were told.  The concept of "mixed-signal" simulation has been around since my days at Daisy (some 18 years ago).  And yet, the concept still struggles along. 

Some blame the lack of behavioral modeling expertise.  Others, that the concept of bringing together the hardest aspects of custom and standard cell design together in one fell swoop, is just too much for any one person to shoulder!  And yet for others, mixed-signal simulation is a standard practice employed with all their designs.

 So, what do you think?  Do you run mixed-signal simulators?  Do you avoid them like the plague?  Either way, why do you do what you do??