Network synthesis means to be able to deal with overflows of reference, symbol, criticism, and control. Any one social action relies on reference, self-reference, and some way to account for the unmarked state in order to be able to selectively capture, number, order, and name these overflows in its orientation to different situations. We propose to solve the venerable sociological problem of the missing micro/macro-link between action selecting its way about, on one hand, and a system reproducing its organization and its structures by the means of the selection of action which does not know about the system, on the other, by introducing the idea of self-organization via double closure (Heinz von Foerster). This idea consists in the assumption that it is the very contingency, i.e., ambivalence, arbitrariness, and discretion, of communication which is able to orientate social action. Or else, social action consists in reproducing, by solving it, the problem it at any moment solves, thereby bringing forth, and calling on, a system which is a second-order device providing for orientations connecting among each other.

All we need when trying to explain the self-organization of social action is the idea of a loss of two degrees of freedom. This loss relates to overflow by solving the problem at any one moment, by giving a specific orientation to a situation, without dissolving the problem itself, which re-emerges afresh from the very contingency inherent in any one orientation. Other degrees of freedom instantly reproduce. The first one of the two degrees of freedom getting lost due to a system's self-organization is the possibility to act arbitrarily. Social action being given, to act arbitrarily is impossible. Even if arbitrariness should be intended, or should be attributed, both intention and attribution would already partake in a social action giving meaning, and thus determination, to the action. Indeed, social action consists in acting always in a way such that any of its endings is at the same time the beginning of something else. There is no conclusion of any action which does not open up for other action to connect. Any one social action selectively not only solves its problem of selection but also reproduces some set of both determinate and indeterminate possibilities the very next action is bound to choose among.

The second one of the two degrees of freedom getting lost is the possibility to call on new criteria, or categories, or schemata, in order to self-determine, describe, and account for one's own action. Action does not only have to connect to next action, it also has to confirm a certain redundancy in its various ways to do so. It has to confirm organization in its most general sense by being able to call itself by some name which is recognized among similar people and similar situations. This loss of a second degree of freedom makes up for the double closure of the system which consists in not only operations (first closure) but also organization (or regulation) having to relate to itself. Note that there is a space of distinction between the first and the second closure. This space provides for the organization of social action being at some variance with the ties and links it actually calls upon to connect to previous and to next action. This space of a distinction, of a severance between the first and the second closure again makes up for error and correction, indeterminacy and determination, overflow and framing being possible.

The sociological conundrum is solved by not only action but also organization informing itself, both of them taken together making up for social action, giving it nevertheless quite some free play in relating action to organization, because there is at any instant a more or less indeterminate set of possibilities interfering. We may also call this the autopoiesis of social action since autopoiesis relies on ways to reproduce a system which combine available and unavailable factors of production, looking at self-reference, as it were, in order to bear with the unmarked states of the network which is about to be explored.

If dissemination of communication defines the problem social action has to deal with, while all the same reproducing it, and if the four culture forms we identified extending on Luhmann's conjecture, and the numbers, orders, and reflections coming with them, all overlap and intermingle in constituting the realm of the possible in contemporary society, we must now ask how a social action reproduces which has no other way but to look at the problem it is bound to solve. And remember that social action does not wait for sociological explanation to solve its riddles, but reproduces via self-organization.

In this chapter, we try to figure out a model of the self-organization of social action. Sticking with the problem of network synthesis we assume that the control problem of the next society is setting the stage for both the dealing with all other dissemination media and for a sociological understanding of the self-organization of social action. That is, we propose to switch from ethnological (tribes), ontological (strata), and ethical (media) explanations of social action to an *ecological* (or knowledge) model, which takes the notion of form serious and looks into ways to self-select inside networks constituted of neighborhood relations, producing a synthesis which is always beyond action and self-similarly built into it at the same time. We choose the terms of an ecological model because we think that the notion of the ecological is apt to deal with both self-organization and network, or heterarchy, without feeling obliged to assume any kind of super-system pre-establishing the social order emerging from the selection of social action. In sociology, the notion of the ecological has a certain reputation to it to be able to describe and inform on heterarchical, loosely coupled, as self-organizing as surprise-ridden, and as robust as fragile, social relationships. We here emphasize this reputation and call it, along with evolutionary thinking, the tradition we call upon for our own model.