In the Forward the English translation of Social Systems, Eva M. Knodt writes,
What distinguishes the systems-theoretical approach to communication from semiological, hermeneutic, and action-theoretical accounts is a probabilistic framework that subordinates structure to function and allows the former to be seen as an emergent order that is dynamic and constantly changing. With his explicit subordination of structure to function, which cannot be emphasized enough, Luhmann breaks not only with the conservatism of Parson’s “structural functionalism,” but with all versions of linguistic structuralism as well. In accordance with the “order from noise principle,” systems theory starts from the assumption that communication is contingent--that is, neither impossible nor necessary–and subsequently seeks to identify the conditions under which the improbable becomes probable. Luhmann locates the major obstacle to the formation of social order in what Parsons described in action-theoretical terms as the problem of “double contingency,” a state of potential paralysis that results from a situation in which two black boxes make their own behavior contingent upon the behavior of the other. Luhmann agrees with Parson’s that action is impossible unless the problem of double contingency is solved–the “pure circle of self-referential determination, lacking any further elaboration, leaves action indeterminate, makes it indeterminable”–but rejects the idea that this problem can be taken care of once and for all, for example, as Parsons believed, with reference to a prior social consensus concerning cultural norms and rules of conduct. In Luhmann’s view, it is precisely the paradoxical indeterminacy of pure self-reference that makes any such consensus susceptible to fluctuations and the unpredictability of random events. In provoking “undecidable decisions,” the problem of double contingency fulfills a catalytic function in the emergence of a constantly changing social order whose instability is the only source of its stability. (xxix)
For social systems, structures are expectations. And expectational structures are subordinate to system function. Expectations (like programs) are dynamic and constantly changing; however, a system’s function does not change. If the function changed, the system would be a different kind of system. The function of an interaction system, for example, is to solve the problem of double contingency. Psychic systems are black boxes to one another, yet communication happens and social systems emerge.
Elsewhere, Luhmann discusses contradiction:
Rejection contradicts the expectation of acceptance or simply the assumption of continuity in “business as usual.” All variation therefore occurs as contradiction–not in the logical sense, but in the more original, dialogical sense. It can occur only as self-contradiction of the system.Theory of Society, vol 1, p. 278
To reject business as usual means to allow innovation/discovery/invention.
Variation therefore comes about through communication that rejects the content of communication. It produces a deviant element [element = event, which vanishes the moment is arises]. . . . The process focuses on the expectation of acceptance already articulated or implied or expected in communication. It thus looks into the past. . . . This separates variation from selection; for a proposed selection would itself be a positive meaning proposal, which is in turn subject to bifurcation of acceptance and rejection. (p. 277)
The introduction into communication of an undecidable decision means that the business-as-usual of communication is interrupted. Any interruption is a potential paralysis. This is the “potential paralysis” of double contingency. If autopoiesis is to continue, the structure must be changeable, and by nature it is constantly changing. Structure is “an emergent order that is dynamic and constantly changing” (Luhmann) Expectations change, but a system’s function remains the same.