``Locality'' often means that the distance is small. However, distance is not defined in CCM. Instead, ``locality'' means that the number of data (atoms) concerning a reaction is small.
Each reaction is basically independent in CCM. Reactions can be regarded as microscopic independent processes or autonomous agents. The relation between two reactions are nondeterministic, or randomized. Because they are independent, the locality has an essential meaning in CCM. On the contrary, although the number of data concerning a rule application is usually also small in conventional production systems, the rule applications are deterministically ordered, and thus, not independent nor autonomous.
If the information used for a reaction is very local, a reaction requires very little computation. However, the number of reactions required for finding a solution is very large. Thus, it is not usually possible to find a solution, or to get into a globally stable state. (?) For example, the N queens system does a complete random search (random walk in the search space) if the most local rule is used (no catalyst rule is used). The system almost never visit a solution state and it does not stop even when it visits a solution state.
If the information used for a reaction is very non-local, the number of reactions required for finding a solution is minimized. However, a reaction requires much computation, and the system may easily fall into a locally optimal state.