Two different approaches are tried to justify Minsky's conjecture. In the first approach, two different methods are used to model the effects of conditional branching. The lack of general information on program behavior makes it difficult to derive any general result. Minsky's conjecture is derived under certain assumptions. Since the program behavior under some of the assumptions is inclined to keep more processors active, this gives a gloomy perspective to array processor organization. The second approach, evading the problem of program behavior, tries to evaluate the space--time product from computational complexity. This approach is far too general and fails to provide any significant result. 5 figures.