Consider the following argument: In a framework that allows uncertain
initial states, nondeterministic effects
are just a notational convenience, not a source of additional
representational power. For any action schema $a$ with nondeterministic
effect $P \lor Q$, we could always replace it with the conditional
effects ${~R{:}P} \land
{\lnot R{:}~Q}$, which in turn can be
reduced to two regular actions. The proposition $R$ stands for a random
proposition that is unknown in the initial state and for which there are
no sensing actions. Is this argument correct? Consider separately two
cases, one in which only one instance of action schema $a$ is in the
plan, the other in which more than one instance is.