Data diversity relies on the observation that
software sometime fails for certain values in the input
space and this failure could be averted if there is a
minor perturbation of input data which is acceptable to
the software. This approach can work well with
Bohrbugs and is cheaper to implement than design
diversity techniques. Data diversity can also deal with
Mandelbugs since retrying the operation (with or
without changing the input data) may not lead to
another failure occurrence.