Extracts from Luhmann - Zettelkasten

by Alex Nelson, 17 August 2025

There are some passages from Luhmann which I wanted to make noted.

Addendum . I initially collected these passages hoping to write about the Autopoietic nature of the Zettelkasten. After thinking about it for a while, I realize this may be a "red herring". Nevertheless, Luhmann's abstraction of autopoiesis is rather interesting in its own right. Luhmann's slips on autopoiesis seem to be rooted at 21/3d26g1i but has not been transcribed.

“Organization” and “Structure” in Systems Theory

It is worth reviewing some vocabulary idiosyncratic to systems theory (which Luhmann employs). Borrowing some of the terms and definitions found in Milan Zeleny’s “What is Autopoiesis?”:

(I dislike the adjective “topological” in “topological boundary”, but I guess it’s technically correct.)

It is unclear to me whether the use of the phrase “process” refers to process in the sense of Principia Cybernetica, or if it abbreviates “production process”.

For example, the abstract notion of an “automobile” is an organization, but the rusting bucket of bolts with 4 wheels and engine that is in my driveway is a “structure”. Together, they form a “system” which is my Honda Civic.

Autopoiesis is a particular example of self-organization.

Caution: I know “element” has a technical meaning for Luhmann, but I cannot find a clear discussion of the concept. It seems to be a “simple unity” which is a component (or subcomponent) of a system. There is a discussion in Social Systems (Ch. I, Section 2, Item 4) but it remains opaque to me. For Luhmann, since he is interested in a theory of Society (for which Communication is of critical importance), an “element” is always an “event” of some kind. But it is unclear if this is accidental (in the sense that “this is just how events look like in Luhmann’s sociology theory”) or essential (in the sense that “Luhmann really argues all elements are necessarily events”).

About Autopoiesis

Luhmann developed a “systems theoretic” approach to thinking about his Zettelkasten. For him, a “system” was distinguished from its “environment” by some kind of “boundary” (these quoted terms are all Luhmann’s terms). Without this boundary, there would be no “autonomy” to a system and it would merge back into its environment.

This played a critical role in the notion of “autopoiesis” for Luhmann, since it was characteristic of “living systems”. And Luhmann’s Zettelkasten was a “living system” of notes. Here’s what he had to say, specifically about communication (English translation from Journal of Sociocyberntics 6 (2008) pp 84-95):

Communications as the basic elements of social systems

To use ipsissima verba ‘autopoietic systems’ are systems that are defined as unities, as networks of productions of components, that recursively, through their interactions, generate and realize the network that produces them and constitute, in the space in which they exist, the boundaries of the “network as components that participate in the realization of the network” (Maturana, 1981: 21). Autopoietic systems, then, are not only self-organizing systems. Not only do they produce and eventually change their own structures but their self-reference applies to the production of other components as well. This is the decisive conceptual innovation. It adds a turbocharger to the already powerful engine of self-referential machines. Even elements, that is last components (individuals), which are, at least for the system itself, undecomposable, are produced by the system itself. Thus, everything which is used as a unit by the system is produced as a unit by the system itself. This applies to elements, processes, boundaries and other structures, and last but not least to the unity of the system itself. Autopoietic systems, of course, exist within an environment. They cannot exist on their own. But there is no input and no output of unity.

Autopoietic systems, then, are sovereign with respect to the constitution of identities and differences. They do not create a material world of their own. They presuppose other levels of reality. Human life, for example, presupposes the small scope of temperature in which water exists as a liquid. But whatever they use as identities and as differences is of their own making. In other words, they cannot import identities and differences from the outer world; these are forms about which they have to decide themselves.

Social systems use communication as their particular mode of autopoietic reproduction. Their elements are communications which are recursively produced and reproduced by a network of communications and which cannot exist outside of such a network. Communications are not “living” units, they are not “conscious” units, they are not “actions”. Their unity requires a synthesis of three selections: namely, information, utterance[1] and understanding (including misunderstanding).[2] This synthesis is produced by the network of communication, not by some kind of inherent power of consciousness, or by the inherent quality of the information. Also — and this goes against all kinds of “structuralism” — communication is not produced by language. Structuralists have never been able to show how a structure can produce an event. At this point, the theory of autopoiesis offers a decisive advance. It is the network of events which reproduces itself and structures are required for the reproduction of events by events.

The synthesis of information, utterance and understanding cannot be preprogrammed by language. It has to be recreated from situation to situation by referring to previous communications and to possibilities of further communications which are to be restricted by the actual event. This operation requires self-reference. It can in no way use the environment. Information, utterances and understandings are aspects which for the system cannot exist independently of the system; they are co-created within the process of communication. Even “information” is not something which the system takes in from the environment. Pieces of information don’t exist “out there”, waiting to be picked up by the system. As selections they are produced by the system itself in comparison with someting else (e.g., in comparison with something which could have happened).

The communicative synthesis of information, utterance and understanding is possible only as an elementary unit of an ongoing social system. As the operating unit it is undecomposable, doing its autopoietic work only as an element of the system. However, further units of the same system can distinguish between information and utterance and can use this distinction to separate hetero-referentiality and self-referentiality. They can, being themselves undecomposable for the moment, refer primarily to the content of previous communications, asking for further information about the information; or they can question the “how” and the “why” of the communication, focusing on its utterance. In the first case, they will pursue heteroreferentiality, in the second case self-referentiality. Using a terminology proposed by Gotthard Günther (1979), we can say that the process of communication is not simply auto-referential in the sense that it is what it is. It is forced by its own structure to separate and to recombine hetero-referentiality and self-referentiality. Referring to itself, the process has to distinguish information and utterance and to indicate which side of the distinction is supposed to serve as the base for further communication. Therefore, self-reference is nothing but reference to this distinction between hetero-reference and self-reference. And, whereas auto-referentiality could be seen as a one-value thing (it is what it is), and could be described by a logic with two values only, namely, true and false, the base of social systems is one of much greater complexity because its self-reference (1) is based on an ongoing auto-referential (autopoietic) process, which refers to itself (2) as processing the distinction between itself and (3) its topics. If such a system did not have an environment, it would have to invent it as the horizon of its hetero-referentiality.

The elementary, undecomposable units of the system are communications of minimal size. This minimal size, again, cannot be determined independent of the system.[3] It is constituted by further communication or by the prospect of further communication. An elementary unit has the minimal meaning which is necessary for reference by further communication — for instance, the minimal meaning which still can be negated. Further communication can very well separate pieces of information, utterances and understandings and discuss them separately, but this still would presuppose their synthesis in previous communication. The system does not limit itself by using constraints for the constitution of its elementary units. If need be, it can communicate about everything and can decompose aspects of previous communication to satisfy actual desires. As an operating system, however, it will not always do this to the extreme. Communication includes understanding as a necessary part of the unity of its operation. It does not include the acceptance of its content. It is not the function of communication to produce a consensus as the favoured state of mind. Communication always results in an open situation of either acceptance or rejection. It reproduces situations with a specified and enforced choice. Such situations are not possible without communication; they do not occur as natural happenings. Only communication itself is able to reach a point which bifurcates further possibilities. The bifurcation itself is a reduction of complexity and, by this very fact, an enforcement of selection. Automatically, the selection of further communication is either an acceptance or rejection of previous communication or a visible avoidance or adjournment of the issue. Whatever its content and whatever its intention, communication reacts within the framework of enforced choice. To take one course is not to take the other. This highly artificial condition structures the self-reference of the system; it makes it unavoidable to take other communications of the same system into account, and every communication renews the same condition within a varied context. If the system were set up to produce consensus it soon would come to an end. It would never produce and reproduce a society. In fact, however, it is designed to reproduce itself by submitting itself to self-reproduced selectivity. Only this arrangement makes social evolution possible, if evolution is seen as a kind of structural selection superinduced on selectivity.

Translator Notes

[1] In German I could use the untranslatable term Mitteilung.

[2] The source of this threefold distinction (which also has been usal by Austin and Searle) is Karl Bühler (1934). However, we modify the reference of this distinction. It refers not to “functions”, and not to types of “acts”, but to selections.

[3] This argument, of course, does not limit the analytical powers of an observer, who, however, has to take into account the limitations of the system.

See also Hugh Baxter’s “Niklas Luhmann’s Theory of Autopoietic Legal Systems” (doi:10.1146/annurev-lawsocsci-102612-134027) for a very readable review of Luhmann’s notions of systems and autopoiesis.

Curiously, it is intriguing to also compare this to Michel Bitbol and Pier Luigi Luisi’s “Autopoiesis with or without cognition: defining life at its edge” Journal of the Royal Society Interface 1 no.1 (2004) 99-107, doi:10.1098/rsif.2004.0012, especially section 6 (but also sections 5, 7 are worth reading as well).

Quotes from Organization and Decision

The passages here are from the Rhodes Barrett translation published by Cambridge University Press.

Also worth quoting is Luhman’s Organization and Decision, chapter 2 (pp.29–35), where he observes:

The most important of the concretizing, circle-breaking assumptions can be briefly summarized as follows:

  1. The basal unity of an autopoietic system takes the temporal form of an event, thus an occurrence that draws a distinction between “before” and “after”; which can therefore be observed only on the basis of the distinction between before and after. When we are concerned with results, we will also talk of “operation” and, in the case of organizations, of “decision.” In comparing theories, it is important to bear this grounding in events (and not in substances) in mind. It follows that the theory presumes discontinuity — continuous disintegration — and considers continuity (thingness, substance, process) to need explaining. A theory of autopoietic systems construed in this manner is in radical opposition to all sorts of process theories, including dialectical theories. It rejects any sort of essentialism and, on the contrary, demands that every event (or in our field every decision) leave the following one to a subsequent event. Forms of essence are only directives for the repetition of selection. Autopoiesis theory is also to be seen in opposition to action theory. For action theories rely on the ideas (such as intentions, or purposes) of actors for linking up their “unit acts,” whereas events such as communications that form autopoietic systems produce a surplus of possibilities, so that something suitable can be chosen in the next step. What is then selected need not be anticipated; the decision is better and more typically made with a preceding event in mind.
  2. A system that generates itself has to observe itself; that is to say, it has to be able to distinguish itself from its environment. This is sometimes denied, but since “organization” cannot be taken to mean the entire world, some criterion or other has to be given for defining what is indicated as an organization. The theoretically decisive question is then whether this demarcation is carried out by the organization itself or, if this is not the case, by whom.
  3. In observing itself, the organization does not observe itself as a fixed object whose properties are recognizable but uses its own identity only to present and abandon ever new determinations. Autopoietic systems can therefore also vary their structures (one speaks of “self-organization”), provided that this is compatible with the continuation of autopoiesis. All reflection on identity that enduring self-descriptions with substantive characteristics propose therefore has to proceed highly selectively, opting for normative assumptions in the process and mostly remaining controversial.
  4. The variability of the self taken in each case as point of reference is guaranteed by the circumstance that the organization observes itself in observation. Even the organizational system itself operates on the level of second-order observation; it constantly diagnoses its own observations (albeit not in every single case). The theory of organization must therefore set in at the level of third-order observation. It observes a system that is observing itself and can consequently extend its observation to the matters that are inaccessible to self-observation. This brings us to the classical sociological problem of latent structures and functions.
  5. Autopoiesis is accordingly possible only if the system is in a constant state of uncertainty about itself in relation to the environment and can produce and monitor this uncertainty through self-organization. The system cannot convert the built-in (we shall also say self-generated) uncertainty into certainty. The absorption of uncertainty (which we will be dealing with in detail) can only be a transformation of the given, current form of uncertainty in adjustment to changing states of irritation. Reflection or self-description can do nothing to change this state of affairs. Every “transcendental” identity could endanger the further reproduction of the system by itself.
  6. The best possibility for coping with uncertainty is to stick by what has already happened. Organizations therefore clarify the meaning of their action largely in retrospect. This in turn tempts them to pay little attention to the given state of the environment. This outdifferentiation at the operational level must, however, be balanced out at the structural level. The appropriateness of structures (e.g., decision-making programs or the typical time needed for operations is mostly decided with an eye to the environment.
  7. Concepts such as self-reference, self-observation, or self-description presuppose operations that realize what is meant. These operations have to be carried out in the system (where else?). If, at the same time, one takes into account that this is not possible in the form of unconnected, singular events, one comes up against the problem of the recursive interconnection of these operations. To make itself possible, every operation has to presuppose recursion to and anticipation of other operations of the same system. This is the only way in which contexts can be identified and boundaries produced and reproduced in relation to the environment. Regardless of the fact that this state of affairs was first mentioned by a biologist, Humberto Maturana, we therefore also speak of autopoietic systems in our context. 29 When the term was first applied to organizations, it was therefore explicitly marked out as a metaphor. Too broad a concept of cognition and too narrow ties to biochemistry have strongly influenced further discussion and often led it astray.
  8. In view of a complex, often confused debate on the subject, a number of explanatory remarks are called for.
    1. As the term “poíesis” indicates, it is a question of producing a work, of generating the system as its own product. This naturally does not mean that the system itself has all the causes necessary for self-production at its disposal. No causal theory could accept such a concentration of all causes in one system (unless it be God). This is already true of concepts like product, production, and reproduction at the conceptual level. Indeed, only when a system not simply exists but has to reproduce itself out of its own products can it, in precisely this regard, be independent of the environment. But it is important for the system to have causes at its disposal (in an organization, for example, members bound by instructions), so that under normal circumstances it can ensure its own reproduction.
    2. The concept stresses not the regular — let alone unconditional — certainty of production but reproduction, that is to say, production from own products. With Heinz von Foerster we can also speak of a “historical machine,” in other words, a system that produces further operations from the state in which it has put itself.
    3. Autopoiesis is accordingly formally defined. As concept, it therefore leaves completely open what material operations it performs. They can be biochemical or neurophysiological operations, but also conscious disposal over attention or communications. Neither analogy nor metaphor enters the argument. What is involved are various applications of a general theory.
    4. The simple concept of autopoiesis serves to distinguish and indicate a state of affairs. It has no empirical explanatory value as concept. What it does, above all, is to oblige other concepts to adapt — for example, the concept of evolution or the understanding of the relationship between system and environment. Everything else depends on what operations materialize autopoiesis, and through what structures produced by evolution and learning.
    5. The autopoiesis of the system is realized at the level of operations. It is therefore compatible with all structures that permit operation to connect with operation. In this context the concept of structure correlates with autopoiesis, and not as it usually does with the division of work. Through operations, structures are generated and reproduced and possibly varied or simply forgotten for use in operations. We can therefore not infer structural conservatism from the concept of autopoiesis. On the contrary, it is the very closure of the system against the environment that gives it opportunities for structural variation that direct ties would not afford. Autopoietic modes of operation are typically one-off inventions of evolution, which in the course of history tend toward structural diversification. “Autopoiesis” thus refers only to a limit to possible structural variation. But as evolution over long periods and domains teaches us, it is precisely the difference between inside and outside that accelerates change. However, it naturally also teaches us that change does not obey the wishes of particular observers, so that to the observer the systems involved may appear rigid and immobile.
    6. The recursive interconnection of operations follows neither logical nor rational rules. It merely produces connections and the prospect of connectivity. Sales figures, for example, can be treated as proof of the success and quality of the given organizational structures. Information can thus arouse suspicion of interest-specific distortions and encourage further efforts to confirm this suspicion. In international relations between organizations, for instance, ecological criteria for products may be interpreted as trade barriers. Recursions, therefore, ensure the maintenance and reproduction of suppressed paradoxes. Anything particular is always something else at the same time.
  9. Autopoietic systems are operationally closed, and for this very reason they are autonomous systems. The concept of operational closure allows for no gradualization; in other words, it does not allow the system to operate in its environment or the environment to operate in the system. A system cannot be more or less autopoietic, but it can be more or less complex. If only for mathematically demonstrable reasons, operationally closed autopoietic systems cannot be described in terms of input/output functions. This gives rise to impressions and descriptions like freedom, arbitrariness, and intransparency, which we shall be going into in detail. The concept of operational closure, too, abstracts from causal assumptions, and thus by no means claims (even relative) causal isolation. A system can be operationally closed and, like the brain, rely strongly on the constant input of resources of a very specific sort (in this case blood circulation). Operational closure means only that the system can operate only in the context of its own operations and in so doing has to rely on structures generated by these very operations. In this sense we can speak of self-organization or, as far as operations are concerned, of structural determinedness.
  10. These theoretical propositions have far-reaching consequences for the relationship between system and environment. In this case, operational closure does not mean that an organizational system can have no contact with the intra-societal environment. Society makes intra-societal communication possible across subsystem boundaries. On the other hand, an organization cannot participate in communication without observing itself as participant. As recipients of communications, the organization’s own structures control what information the system is irritated by and stimulated to process information itself. As sender of communications, the organization makes decisions about what it wishes to communicate and what not. To this extent, the environment remains for the organization a construction of its own whose reality is naturally not questioned. In this, we agree with Karl Weick. What is observed as environment in the organizational system is always a construct of its own, filling in the otherreference of the system. As it were, the environment validates the decisions of the system by providing the context that allows the system to determine in retrospect what has been decided (Weick speaks of enactment). It allows uncomfortable causes for the system’s own decisions to be externalized, thus “punctuating” its own operations. It is a backup area for problems, which allows the system to ignore the part it plays in generating problems. In sum, it allows the system to relate its own operations to a niche without asking why the world and society in particular contain such niches. This is what the old concept of “milieu” means.
  11. Although biology has given us the concept of autopoiesis, we can very well leave it open whether and how the reproduction of relatively stable large chemical molecules in cells can be understood as autopoietic: perhaps because it is possible only in cells, perhaps because they are highly unstable entities that have to be constantly replaced. In the case of social systems, autopoiesis is much easier to recognize, or at least quite differently structured, for social systems are not entities requiring replication that constantly have to be replaced. Like consciousness systems, they consist only of events, which pass in arising and which have to be succeeded not by the same events but by others. The ongoing transition from one element to another — the ongoing reproduction of otherness — can be understood only as autopoiesis, for it presupposes connectivity generated within the system itself. No environment could input anything suitable at the speed required. Only the system itself can stop its own decay, which takes place from moment to moment. And this makes very specific demands on structures; they must not strive for repetition but first of all regulate the transition from one to the other. As we have seen, this requires orientation on highly referential but also determinable meaning.
  12. Autopoiesis relies on a system being able to generate internal improbabilities deviating from what is usual. Structurally limited contingencies then take effect in the system as information — as information not from the environment, which the system cannot contact, but at best via the environment (not in biological systems such as cells, immune systems, brains, but only in systems that can distinguish between themselves and the environment in the medium of meaning). An autopoietic system can thus inform only itself, and in the system information has the function of selectively limiting the options for continuing the system’s own operations, with the further function that decisions can be made relatively rapidly through connectivity options.
  13. Closure in this operational sense is a condition for the openness of systems. Older systems theory had, with respect to the law of entropy in thermodynamics, spoken of open systems to explain how order is developed and maintained against the trend. But it failed to ask what enables a system to be open; in other words, what systemic order has to exist for a system to be able to afford openness and possibly even increase the complexity of the aspects in which it can be open. This question was not put because empirical examples and/or the input/output model had been the orienting factors. Although even older cybernetics had spoken of systems that were closed with regard to information and open with regard to energy, only the more recent theory of self-referential systems clearly states that operational closure is the condition for openness.
  14. The theory of autopoietic systems distinguishes strictly between the continuation of autopoiesis and the maintenance of certain structures that serve to ensure sufficient redundancy and connectivity, and thus make autopoiesis possible in the first place — in one way or another. Structures are thus assumed to be functional, contingent, and differently possible. From the disposition of the theory, this permits understanding for the ambiguity of structural arrangements, the need to interpret them, and their circumventability. We could also say that the theory of autopoietic systems draws the attention of the observer particularly to the circumstance that structures have meaning, and thus have to be constituted in open horizons of reference to other possibilities, whereas autopoiesis itself is not a topic in autopoietic systems. This brings us very close to theories of “symbolic interactionism” or theories of the hermeneutic “interpretation” of reality, without, however, having to take recourse to behaviorist (Mead) or subjectivist assumptions. In what follows, we will repeatedly see that uncertainties have to be reduced and ambiguities clarified in the decision-making process; but also that uncertainty and ambiguity in the processing of meaning are always regenerated, and that the autopoiesis of organizations, in particular, is kept in motion precisely by uncertainty being both reduced and renewed. The impressive gain this complex conceptual maneuver affords is to shift the basic problem in systems theory from maintaining resources to that of maintaining a difference. This also means that one no longer speaks of “existential” necessities (an organization can exist only if…) but of conditions of possibility for observing organizations. If they cannot be distinguished, they cannot be observed. If we describe organizations as autopoietic systems, we are therefore always concerned with the generation and reproduction of a difference (systems-theoretically: between system and environment), and the concept of autopoiesis means that an observer who uses it presupposes that this difference is generated by the system itself and reproduced by systemic operations.

In Organization and Decision, Luhmann also writes (p.53):

The abstraction of the theory of autopoietic systems serves to include such states of affairs. What ultimately matters is who observes a system, and which one, with the help of the distinction between “inside” and “outside.” This does not mean that systems theory is at the mercy of unrestrained relativism. The thesis of autopoietic self-constitution asserts rather that what primarily matters is the self-observation of the given system, thus the distinction between self-reference and other-reference. All other observers have to comply if they cannot otherwise find their object but only follow their imagination or model constructions.

Quotes from Theory of Society

In Luhmann’s Theory of Society, vol I, Chapter I, page 42 of the English translation published by Stanford:

The general theory of autopoietic systems requires precise identification of the operation that carries out the autopoiesis of the system and thus demarcates the system from its environment. In the case of social systems, this is done through communication.

Communication has all the required properties: it is a genuinely social operation (and the only genuinely social one). It is genuinely social in that, although it presupposes a multiplicity of participating consciousness systems, it cannot (for this very reason) be attributed to any individual consciousness. Moreover, under the conditions of its own functioning, it excludes consciousness systems’ knowing the given inner state of the other or others: in oral communication because interlocutors participate in utterance/understanding simultaneously; in written communication because the partners participate in their absence. Communication can therefore only assume that adequate understanding also has psychic correlates. In this sense (and “interpenetration” can mean no other), it depends on operational fictions that have to be tested only occasionally and only through communication.

Communication is also genuinely social in that a “common” (collective) consciousness can by no means and in no sense be produced, and that consensus in the full sense of complete agreement is hence also unattainable; communication operates in its place. It is the smallest possible unit of a social system, namely, the unit to which communication can still react through communication. In another version of the same argument, communication may be said to be autopoietic in that it can be produced only in recursive relation to other communications, thus every single communication contributes to reproducing only in a network. A unit of communication is completed with understanding or misunderstanding regardless of the essentially infinite possibilities for clarifying what has been understood. But this completion takes the form of a transition to further communication that can pursue this clarification or turn to other topics. The production of elements is autopoiesis. Communicating acceptance or rejection of the proposed meaning of a communication is already another communication, and, regardless of all thematic ties, does not arise of itself from the preceding communication. An essential precondition for the autopoiesis of society and its structure formation is that communication does not of itself contain its own acceptance; this has to be decided through further, independent communication.

In Ch. I (p.39) discussing “self-reference” and “other-reference”:

The same can be formulated with the aid of the distinction between self-reference and other-reference. Meaningfully operating systems reproduce themselves in ongoing implementation of the distinction between self-reference and other-reference. The unity of this distinction cannot be observed; it is carried out only operationally and only internally (otherwise we could not speak of self-reference and other-reference). Like living systems, systems operating with meaning never go beyond their own boundaries with their own operations. But in the medium of meaning, boundaries always have another side, as forms they always exist as two-sided forms (and not as pure facticity of the operational act). This means that the observation accompanying progression from operation to operation always notices the selectivity of the recursive connection and hence something that belongs not to the system but to the environment. In communication, information about something is actualized and changed that is not communication itself. In the network of communication, other-reference always co-occurs in the search for suitable connections. The boundary of the system is therefore nothing other than the self-produced difference between self-reference and other-reference, and as such is present in all communications.

On page 174,

Communication, one should remember, is to be understood in terms not of the action of uttering but of understanding. Written language accordingly presupposes readers if it is to be used not only to record but also to communicate.[183] This makes it clear that the immense increase in what would come to be called the reading public revolutionized societal communication.

Footnote [183]: Communication research and especially historical research tends for understandable methodological and source-related reasons to take the opposing view; for texts are easier to find and analyze than what goes on in the reader. On this problem and the “priority of reading over writing,” see Havelock, Literate Revolution, 56ff.

On page 266:

We should recall that autopoietic systems always perform their operations only in the actual present. The recursive interconnection of operations also takes place in the present on the basis of currently available conditions and connectivity options. For the operation (and this also holds for communication if it is to be an autopoietic operation), there is accordingly neither a beginning, because the system must always have already begun if it is to be able to reproduce its operations from its own products, nor an end, because every further operation is produced with an eye to further operations.[53] Only an observer (and this can be the operating system itself) can make out a beginning and an end if he takes a corresponding before/after construction as his basis. Only when the system is operating and has built up sufficient complexity to describe itself in the temporal dimension can it “postcipate” its beginning. The determination of a beginning, an origin, a “source,” and a (or no) “before” is a myth fabricated in the system itself—or an account by another observer.[54]

Where autopoietic systems are involved, evolution can therefore not be understood as a mere meeting of particulars where what already exists enables something to be added that could not have been done without this precondition. We do not have to completely exclude such a development; but it alone cannot explain the speed of evolution. Autopoietic systems make evolution possible, rapid evolution, precisely because they close themselves off on the basis of their own distinctions and can thus recruit their own requirements whenever the simultaneous environment suffices.

Footnote [53] See Luhmann, Gesellschaft, 2: 816, on interactions envisaged as episodes, showing that episodization is possible only with the help of the distinction between interaction and society, and hence only in a society that is itself infinite.

[54] N. Katherine Hayles, “Making the Cut: The Interplay of Narrative and System, or What Systems Theory Can’t See,” Cultural Critique 30 (1995): 7–100, is thus right when she suggests that a transition could be useful to narrative patterns that can give an account of how autopoietic systems, and, among them, observing systems, set themselves as difference. But this solution does not get us very far, although it makes other means of establishing plausibility available and also permits us to inquire into the historical situations in which the narration of autopoietic evolution has been able to take root at all. But ultimately this merely brings us back to the question of who tells the tale.

Quotes from Social Systems

In the Introduction to Social Systems, Luhmann defines some criteria for autopoiesis as (p.11):

The theory of self-producing, autopoietic systems can be transferred to the domain of action systems only if one begins with the fact that the elements composing the system can have no duration, and thus must be constantly reproduced by the system these elements comprise. This goes far beyond merely replacing defunct parts, and it is not adequately explained by referring to environmental relationships. It is not a matter of adaptation, nor is it a matter of metabolism; rather, it is a matter of a peculiar constraint on autonomy arising from the fact that the system would simply cease to exist in any, even the most favorable, environment if it did not equip the momentary elements that compose it with the capacity for connection, that is , with meaning, and thus reproduce them. Different structures may exist to accomplish this, but only ones that can withstand the radical trend toward immediate (and not merely gradual, entropic) dissolution of the elements.

In Social Systems, Luhmann writes (p.169 of the English translation published by Stanford University Press):

Autopoietic reproduction does not mean that a specific action is repeated in every appropriate case (e. g. , every time a person wants to light a cigarette he reaches for his lighter). Repeatability must also be secured by the formation of structures. Reproduction means only production out of what has been produced; for autopoietic systems this means that the system does not end through its actual activity, but goes on. This going on depends, however, on the fact that actions (whether intentionally or not) have communicative value.

Later, Luhmann writes in Social Systems (p.360):

It would be a crass reification to reduce this state of affairs to a distinction between life and science (or something similar). The difference between autopoiesis and observation is a very elementary one, and both occur in all autopoietic systems, even in those that-like science-specialize in observation and in predictions and explanations that depend on it. Correspondingly, contradiction has a double function in all self-referential systems, namely, to block and to trigger, stopping observation that encounters contradiction and triggering connective operations that cope with contradictions and owe their meaningfulness exactly to this coping. Thus one comes to the conclusion that contradiction is a semantic form that coordinates autopoiesis and observation, mediating both types of operation, and separating and combining them. Contradictions achieve this because switching off operations that connect with observation means simultaneously switching on operations that precisely then are still possible.

…and (p.402):

Thus what holds for all autopoietic systems also holds here: observation (operative distinction) is only possible on the level of elements and only so that the observer is provided with a description that at once accomplishes the self-reference of the elements and thereby reveals that they belong to the system and not to the environment. Even self-observation is bound to this precondition.

Luhmann writes (p.434):

The unity of autopoiesis is nothing more than its on-going self-renewal.

…and (pp.443–444):

  1. We will speak of basal self-reference when the basic distinction is between element and relation. In basal self-reference, the self that refers itself is also an element, for example, an event or, in social systems, a communication. Basal self-reference is the minimal form of self-reference, without which autopoietic reproduction of temporalized systems would be impossible. We showed this above in the discussion of Whitehead’s concept of event [in Ch. 8 Section III]. Basal self-reference is a constitutive requirement for forming self-referential systems, but it is not a system reference, since the indicated self is intended as an element, not as a system, and since the guiding distinction is element/relation and not system/environment. This of course does not deny that the concept of element presupposes a system and vice versa. But that does not negate the distinction between different forms of self reference; it merely grounds the expectation that they correlate to each other. […]

[…]

[2.] […]

[3.] […]

Systems formed and unified by basal self-reference (autopoietic systems) are always closed systems. But this concept acquires a new meaning in comparison with earlier systems theory. It no longer indicates systems that exist (almost) without environments, that is, that can determine themselves (almost) completely. Instead, it means that such systems create everything that they use as an element and thereby use recursively the elements that are already constituted in the system. How is this to be understood for meaning systems, especially social systems?

“Basal self-reference” seems to first appear on page 144 of Social Systems and discussed further in Chapter II Section III of that book (pp.41 et seq.), albeit implicitly mentioned in Section III.

“Basal self-reference” in the context of Luhmann’s theory of communication is discussed in Social Systems, Chapter 4 Section II, pages 139 et seq.

Luhmann writes in Social Systems (p.189, bold emphasis is mine, italics are Luhmann’s):

Internal differentiation (system differentiation) uses an entirely different procedure. While environmental differentiation relates to the requirements for the system to observe the environment and is both stimulated and limited by this,[29] internal differentiation results from the process of autopoietic reproduction. The connection between reproduction and differentiation becomes comprehensible if one views reproduction, not as the identical or almost-identical replication of the same (e.g., as replacing supplies), but as a constantly new constitution of events that can be connected.[30] Reproduction always implies reproducing the possibility of reproduction. For social systems, this means restoring double contingency. On the one hand, reproduction is subject to the conditions for connectivity; it must be able to suit a situation. On the other, it offers possibilities for forming within the system a new system having its own system/environment difference-perhaps a system that will last longer than the initial one. At a party one sees a woman reach for a cigarette, and (if she dawdles suggestively), one may offer her a light from one’s own cigarette lighter.[31] Settled system differentiations stabilize the possibilities for reproduction by constraining conditions on the comprehensibility of communication and the suitability of behavioral modes. But the meaning surpluses that must be produced alongside provide ever further chances for innovative system formation; in other words, they provide the chance to include new differences and new constraints and thus to increase the ability to constrain the initial situation via differentiation. Only thus can system complexity increase.

Footnote [29]: We have defined observation as recording information with the help of a difference.

Footnote [30]: An extensive investigation of this connection is given in Yves Barel, La Reproduction sociale: Systemes vivantes, invariance et changement (Paris, 1973).

Footnote [31]: To give another, less interactional example, one could point to the discussion of formal versus informal organization. A formally organized social system can be differentiated formally as a result of planning, but it necessarily also offers occasions for informal system formation, which then involve an ambivalent relationship with the formal rules. This demonstrates better than earlier organizational research, which worked with the concept of groups, that there are interconnections between ongoing reproduction, differentiation, internal growth, complexification, and the increased channeling of the spontaneity of further differentiation. In contrast to the hitherto dominant opinion, one might suppose that it is formal, not informal, organization that provides the means of regaining elasticity and adaptability.

And (pp.286–287):

The theory of autopoietic systems presented here brings together two different components of reproductive self-determination, which are called “structure” and “process” in the standard nomenclature. Structure keeps ready a range of possibilities because (!—not only although) it emerges by selection. Given structure, the ongoing determination of the next element comes about by excluding other available (systemically possible) possibilities. For a process, the before/after difference is what counts. The process determines itself by departing from what is momentarily actual and making the transition to a suitable but different (new) element. Both procedures—exclusion as well as the search for connection—are contingent. Therefore they can work hand in hand and reduce the contingency of the component on the other side of the difference to a minimum that can be taken in stride, so to speak. Perhaps the best, in any event the most far-reaching, example of this is the way of talking that uses language.

Regarding autopoietic systems capabilities for forming connections—particularly “internal capacity [Anschlussmöglichkeiten] for forming connections”, a nearly identicalphrase used in Luhmann’s “Communicating with Zettelkastens—Luhmann writes in Social Systems (pp.288–290):

This reinstatement of expectability is a requirement not of stability but of reproduction. Expectations are the autopoietic requirement for the reproduction of actions, and to this extent they are structures. Without them a system lacking an internal capacity for forming connections in a given environment would simply cease, indeed, would cease of itself. This is not a problem of incapacity to adjust in relation to the environment. (The system does not react to such a problem through structures pure and simple, but by structural flexibility and by steering the selection of structures.) Structures of expectation are basically the condition of possibility for connective action and thus the condition of possibility for elements’ self-reproduction through their own arrangement. Being temporally bound, elements must continually be renewed; otherwise the system would cease to exist. The present would disappear into the past, and nothing would follow. This can only be prevented if the meaning of an action is constituted within a horizon of expectation that anticipates further actions—whether by expecting that a meaningful sequence will continue (as with the next digit in the sequence when dialing a phone number) or by expecting complementary behavior of various sorts (as in opening the door when one hears the doorbell ring). Action then seems to escape its momentary transitoriness, to go beyond itself. [35] This is possible, however, not because of an immanent energeia, a force, an elan vital of action but by structures of expectation that arc pre-given and constantly reactivated, reducing the uncertainties of the future (and along with them the temporal self-reference of the individual elements, i.e., actions) so that action can specify itself by selecting relations. How far this holds for systems other than social ones would require a separate investigation. The stability of expectations rests on the constant cessation and renewal of actions, on their “eventuality,” their being events. The fluctuation of the material in the basal events is the precondition for forming and retaining expectations that distinguish themselves from what is changing.

Thus the concept of structure complements the conceptualization of elements as events. It indicates a condition of possibility for basal self-reference and the system’s self-referential reproduction. Therefore, structures can-as the verb “complements” indicates-never be conceived as a sum or mere collection of elements. The concept of structure indicates a level of order in reality different from the concept of event.

Correspondingly, the concept of event must be understood as complementing that of structure. The concept of basal self-reference brings this to light. […] But self-reference is a more complex concept than this would indicate. It includes the capacity to determine itself internally through a combination of “self-identity” and “self-diversity” and at the same time to leave room for external codetermination. One cannot bypass this level of articulation; it enables an adequate reconstruction of what Weber might have had in mind when he spoke of the “subjectively intended meaning” of an action.

One can describe what one attains in this way as a nexus of several variables that, on the surface, contradict one another, namely, as the unity of (1) the selective linkage of elements, (2) the binding of free energies from other levels of reality through interpenetration, (3) the constant instantaneous dissolution oflinkage and binding, (4) the reproduction of elements on the basis of the selectivity of all the linked and bound relations, and (5) the capacity for evolution in the sense of a deviant reproduction that opens up possibilities for a new selection. Such a system has no temporally fixed essence. It is subject to time not merely in the sense that it must adapt and if necessary alter structures . Not even the interchangeability of elements (the theory of autopoiesis began with a consideration of macromolecules or cells) grasps the temporal reference radically enough. Action systems use time to force their continuing self-dissolution and thereby guarantee the selectivity of all self-renewal; and they use this selectivity to enable self-renewal in an environment that makes continuously varying demands.

See also item (7) (which starts “Concepts such as self-reference, self-observation, or self-description presuppose operations that realize what is meant.”) from the passage from Organization and Decision

What this means

Communication as an autopoiesitic system means communication is in response to earlier communications.

A dissenting opinion would point out that Luhmann did not really adequately incorporate Autopoiesis into his thinking until the early 1980s, which is about 20 years after starting his second Zettelkasten.