The authors observe that the task of planning is a type of configuration problem. This is interesting, especially in the context of M1 where one of the features of the system is the ordering of sub-problems to be solved. The ordering of sub-problems in M1 is not discussed even though agenda maintenance is a common AI topic and a paper in which M1 is discussed is cited.

      Another topic discussed was the common merging of function and component terminology. I wonder if there have been any attempts at a system which allows both a separate representation and yet facilitates the merging of the two when it is useful for problem solving. Could such a system identify when this merger was valid and/or useful?