Yes, the point of unstable crosses is that you will get diversity to select from, as much as possible.
There is also some merit in crossing F1 or F2 generations to maximize the chance of breaking linkages and getting unusual results iirc...
There's a paper I was reading this spring... will check my notes and try to find it this evening.
I have to say that for me, my most interesting results have come from side crosses involving an F2 parent.