A theory that does not make definite predictions about what we observe can still be scientific.
A theory that is not better supported by observations than a competing theory should be preferred for that very reason.