De/centralization, Discretion, and the Anarchist Movement
This is a lead essay in the C4SS Mutual Exchange Symposium: "Decentralization and Economic Coordination." The related readings and list of all other articles can be found in the introduction here.

Anarchists have long made decentralization a core demand for the structure of our organizations and have placed tremendous faith in the decentralization of our organizations as a means by which to prevent the generation of hierarchies and abuses of power, among other issues. However, this faith is based on very little, and anarchists can stand to benefit by rejecting it as an all-or-nothing principle alongside the concomitant Manichean dichotomy between decentralization, representing the Good, and centralization, representing the Bad. It is, in fact, imperative that we do if we aim to build a movement capable of self-defense and growth in the face of growing authoritarian power.

Favorable feelings toward decentralized organizational schemes are one of the few things all flavors of anarchism have in common. Nearly every anarchist believes, to some extent, that the core values of the anarchist movement are best given life by avoiding the centralization of decision-making and power. To a certain extent, this is right. Some level of decentralization is necessary in an organization so that its organizational processes can be carried out smoothly. Besides abuses of power, there are purely practical reasons to avoid excessive centralization. An organization in which decision-making is highly centralized increases the distance of decision-makers from conditions on the ground, increasing the likelihood of a bad decision being made because of the decision maker’s ignorance of changing local conditions.1 High centralization also can overwhelm decision-makers with minutiae that’s better off being dealt with locally, negatively impacting the rate at which projects are undertaken and completed – i.e. excessive centralization can generate bottlenecks that impede progress.

This does not mean, however, that centralization is more trouble than it’s ever worth. Similar practical problems are associated with decentralized schemes. The most notable problem with decentralization is the lack of coordination between fully autonomous organizational subunits.2 In a decentralized scheme, there may be no actual coordination whatsoever, and the movement and development of the organization could be thought of as complexity generated by the interactions of the organization’s subunits rather than conscious coordination, i.e. mutual readjustment of plans, as such.

This is especially true in a context where there is no basis for cooperation, such as in a commercial market. We can reasonably argue that markets, taken as collections of agents making exchanges with one another, are decentralized systems. The interactions between these agents generate complexity – one example of which would be a “price system” – that then influences the further development of the system. Putting aside that prices in real world markets are not only the product of competition, but also contractual obligations and sometimes legal fiat, we can use this imperfect abstraction to investigate how a decentralized structure can inhibit coordination.

Market anarchists argue that markets enable coordination through the price system, which acts as a means of communication of local knowledge. It would be possible to attack this argument, derived from the Austrian aristocrat and economist Friedrich von Hayek, from a variety of angles. We could note that compressing local information, which is sometimes made up of non-quantitative observations and otherwise incommensurable figures, into a decimal value like price would entail compression to the point of obliterating any actually useful knowledge contained therein. A commodity increasing in price will not be enough for a decision to be made on the basis of that fluctuation. If the price of tin increases relative to copper, and these can be substituted in the production of a commodity, it still must be known to decision-makers at the firm-level whether that price increase reflects a temporary change in conditions, or a long-term one (Cockshott and Cottrell). The change in price alone is insufficient to guide investment decisions, and the knowledge of what drove the change is not arcane knowledge that is beyond our ability to transmit in a timely, useful, manner. The price change also involves some latency between the event that led to the price change and the actual change in price as felt by producers and consumers, meaning the timeliness of the price change may not reflect anything by the time it’s relevant to profit and loss calculations. 

We could also note that prices do not necessarily reflect truthful information, so much as the beliefs of agents in the market. Market actors routinely get things wrong, and this is only revealed after the fact by significant economic dislocations that affect the lives of workers and consumers. Speculative bubbles are a testament to this and perhaps the best example of it. It can always be argued that markets will adjust to these circumstances and that there are incentives for spotting these things. Indeed there are: these incentives to spot bubbles result in windfalls for financiers and other gamblers in real world markets, and speculative bubbles are typically resolved in a way that further centralizes and concentrates wealth.

On a game theoretic level, it is also wrong to talk of markets as “coordinating” economic processes. A “communication system” is insufficient to enable coordination. Again putting aside whether the information conveyed by price changes is accurate or has meaning, the competitive structure of markets necessarily inhibits cooperation and therefore also inhibits effective coordination toward common goals. John O’Neill provides an account of this via his discussion of Hayek’s epistemological critique of socialist planning:

In Hayek’s defence of the market, there is an assumption that the communication of relevant information in the market is all that is necessary for the achievement of coordination. He treats the solution of the problem of information distribution as ipso facto a solution to the problem of coordination. The two are, however, distinct. The possession of information about the plans and actions of others does not of itself enable one to act so that one’s actions are coordinated with those of others. For example, producers possessing information that the demand for the goods they produce is falling relative to supply are not in a position thereby to ensure that their actions are coordinated with those of consumers and other producers. The problem is not simply that of the lack of relevant information noted above. Even given this information, the problem of coordination is not thereby solved. Where plans are inappropriately coordinated, a mechanism is required to adjust those plans. For example, knowledge that (given planned consumption and production of some good) production will exceed demand is of no use to a producer who aims to achieve coordination. Even given mutual knowledge of projected discoordination, no adjustment by any particular actor of his or her own actions will necessarily lead to coordination. There must be some mechanism whereby producers can mutually adjust plans in order that activities be coordinated.

The market, as a competitive order, has no such mechanism for mutual adjustment, for the same reason that it blocks the movement of information. While mutual adjustment might benefit all parties, if one or more cooperates while another does not, ceteris paribus, the non-cooperating party will benefit. Given that all parties are self-interested, the competitively stable strategy is non-cooperation: the market inhibits the mutual adjustment of plans. Eventual adjustments of actions are achieved in the market rather via sudden dislocations in economic life, in which overproduction of certain goods leads to the disappearance of certain competing producers, underproduction to an uncontrolled and excessive movement of productive resources to supply demand.3

The problem for inter-firm coordination within a market is simply that there is no mechanism which enables firms to actually coordinate their plans together and make mutual adjustments as necessary. The ideal market lacks not only a mechanism for coordination (as could exist in, e.g., a cooperative federation or a cartel) but also inhibits cooperation from the start because the competitively stable strategy within a competitive market is always non-cooperation. When we speak of “coordination” within a market, what we’re really talking about are the dislocations that competition generates: workers being laid off when a firm is forced to readjust plans, companies being liquidated with other companies consolidating the liquidated companies’ assets and increasing their share of the market, and so on.

Not every decentralized system has to be subject to such a constraint, however. It’s possible to imagine anarchist affinity groups with clear lines of communication between themselves, and enough shared trust to coordinate a plan of action and follow it through. As much as we can imagine it, however, these kinds of interactions are irregular at best. A commitment to total decentralization prevents coordination between anarchist groups from being a more regular occurrence. Anarchist groups have few if any stable lines of communication between one another, and seldom cooperate over long periods of time and distance in a recurring manner.

Here, we need to overcome the Manichean dichotomy of decentralization and centralization, and introduce instead the concept of discretion from the Viable System Model. Espejo and Reyes use “discretion” to refer to the distribution of decision-making authority through an organization’s units over an organization’s resources and processes. The distribution of discretion within an organization will determine the degree of that organization’s centralization and decentralization.4

Consider a regional anarchist confederation as an example. The confederation may have a central body with discretion over the publication of a collective journal chronicling member organizations’ theoretical output and practical activities. Although effectively centralized, such a structure would not be usurping control over local issues from member organizations, nor would it be dictating anything to them. A flexible enough federation could define such centralized functions as it pleases, without necessarily endangering the autonomy of member organizations over their own processes and resources. The autonomy of member organizations enables them to mutually recognize the need for discretion to some centralized function – not every organization needs to concern itself with running its own journal – and thereby enable them to focus their attention on what really matters. This means that centralization can, in some instances, enable units to better manage the complexity generated by their own organizational processes and their interactions with other groupings.

The concept of discretion allows us to discuss centralization and decentralization without imputing moral content onto the terms, or treating them as an either/or choice. It enables us to see that centralization and decentralization can even exist alongside each other within an organization, and to understand how to appropriately structure our organizations so as to maximize the autonomy and effectiveness of members in performing their various tasks while also enabling the membership to coordinate their efforts toward a collectively-decided upon goal.

The reflexive fear of centralization among anarchists has contributed to the anarchist movement’s incapacity to respond to authoritarian threats and to pose as a counter-power to authoritarian movements. It has also generated  an informal centralization that is hidden by its lack of formalization (that is, by the lack of formally recognized discretion). Thus, Castoriadis in Workers’ Councils and the Economics of a Self-Managed Society argues:

It is not centralization as such which has made of modern societies such outstanding examples of political alienation or which has led to minorities politically expropriating the majority. This has been brought about by the development of bodies separate from and “above” the general population, bodies exclusively and specifically concerned with the function of centralization.

Bureaucratic centralization is a feature of all modern exploiting societies. The intimate links between centralization and totalitarian bureaucratic rule, in such class societies, provokes a healthy and understandable aversion to centralization among many contemporary revolutionaries. But this response is often confused and at times it reinforces the very things it seeks to correct. “Centralization, there’s the root of all evil” proclaim many honest militants as they break with Stalinism or Leninism in either East or West. But this formulation, at best ambiguous, becomes positively harmful when it leads as it often does – either to formal demands for the “fragmentation of power” or to demands for a limitless extension of the powers of base groupings, neglecting what is to happen at other levels.

Autonomy, the fact that we give ourselves our own rules to live by, is on one hand enabled by decentralization, and on the other hand undermined by it. Enabled, obviously, because decentralized decision-making provides local decision-makers significant authority over their own conditions. However, total decentralization imposes limitations on autonomy by restricting the control we have as members of the organization to what’s in our local reach. While the autonomy of the individual subgroup is maximized, the autonomy of the whole collective is nullified: the whole collective can not aggregate and put forward its own collective meanings and rules, it can’t even manage the complexity generated by the interactions between components parts and themselves and their environments. The effect of this on individual members is that they are confronted by complexity they have no autonomy over, which exists as an impersonal power above them. Decentralization, rather than securing autonomy, becomes a limitation on it.

One of the words that proponents of decentralization within the wider anarchist movement today will throw around is stigmergy. Stigmergy is a word which has its origins in entomology, specifically the study of eusocial insects. It’s used to describe the ways that eusocial insects, such as ants or termites, seemingly coordinate their activities without being given commands from a central body. Eusocial insects can read scent tracks left by previous hive members to determine where to search for food, where to build new additions to the hive, and so on.

Stigmergy has its downsides. The same mechanisms that enable a collective of ants to build a complex hive can also result in them fooling themselves and forming an ant mill, following each others scent trails into a spiraling death march, eventually marching themselves to death. Ant mills are unintentional effects of the way eusocial insects organize themselves, and if markets and anarchists are truly “stigmergic,” we have good reason to believe that something like the ant mill is always a possibility, that the mechanisms of “self-organization” contain failure modes which spiral out of the control of any given individuals and which would demand conscious collective intervention (obviously an option unavailable to the ants) to avoid them or steer out of. Our insistence as anarchists on decentralization at all costs may have us in an ant mill of our own, unable to confront growing authoritarian movements and powers, flailing from ideological fad to ideological fad, and unable to speak of any lasting victories since the heyday of anarchism in the nineteenth century workers’ movement.

Decentralization is a shibboleth that has hindered the ability of anarchists to organize ourselves and effectively manage our common affairs. A totally decentralized system lacks what the Viable System Model terms a “metasystem,” a subsystem that provides cohesion and coordination between primary activities. While a metasystem may necessarily feature centralization, this need not necessitate bureaucratic hierarchies and chains of command. The Viable System Model, in fact, aims to provide a conceptual blueprint for building non-hierarchical organizations. 

The task for anarchists is to find ways to organize ourselves that do not necessitate or reproduce violent hierarchies, but a centralized organization need not imply hierarchy at all, while decentralization can also be a feature of a hierarchical social organization, e.g. feudalism, or typical capitalist markets. In an organization in which every member has voting rights, the centralized democratic body containing all members of the organization, making a decision in common, represents the effective merger of decentralization and centralization. What we must avoid is not centralization, but bureaucracy: the creation of a centralized decision-making power separate from the general membership whose business is to run the affairs of the organization. 

The Leninist party provides a perfect example of bureaucratic centralization, with its Central Committee forming a body separate and above the general membership of the party, accountable only to itself and concerned with the day to day affairs of running the Party. The Bolshevik seizure of power and transformation of revolutionary institutions into military bureaucracies sealed the fate of the Russian revolution through the abolition of autonomous workers’ power. But this workers’ power, no matter what, would have taken a centralized form if only at the most immediate level, through the institution of workers’ councils and collective self-management.

Centralization and decentralization are practical matters. Anarchist discourse should cease to center a supposed conflict between the two, so that we can approach and appreciate the real value of both. Centralized organizational components, when properly designed and bound by proper rules and scope, do not pose an inherent threat to the political decentralization of power within anarchist organizations and communities. Rather, in some instances, they are practically necessary and inescapable, and sometimes even beneficial to our individual autonomy. 

Decentralization, rather than a goal in and of itself, is to be preferred when it is practically necessary or beneficial – e.g. adopting a cell structure (or some variant of it) to mitigate the amount of information any given police infiltrator can have access to. Importantly, in this instance the decentralized structure does not guarantee that any mitigation will occur. If the police forces can infiltrate every cell (which is not extremely difficult given the small number of anarchist organizations in any given place) then they have the potential to still map out the organization’s membership and initiatives. A decentralized cell network of anarchist organizations with five cells will still only take five police infiltrators to get a sufficiently complete picture for law enforcement. That is not beyond police capabilities – as green anarchists who followed the cell model and were still rounded up in the Green Scare will tell you. Decentralization and centralization have trade offs, and we have to ask ourselves, in each instance, whether the trade offs are worth it. This depends on anarchists approaching the subject with nuance and open-mindedness toward centralized solutions. 

If the anarchist movement is an ant mill, the viable system model tells us that we need a metasystem to bring ourselves out of it. This will necessarily entail centralized structures in our institutions, in order to set operational policies in response to changing internal and external conditions, and to mediate disputes that arise between operations. What matters for anarchists is that these structures do not become bureaucratic, that the individuals who make up the metasystem do not become solely and exclusively responsible for its management (and by extension, the management of the rest of the system). In order to prevent the generation of formal and informal hierarchies while still maintaining our effectiveness and cohesion, our organizations must become more thoroughly democratic. The organizational task for anarchists, in an ironic sort of way, is to prove Lenin’s dictum that “every cook can govern.”

  1. Raul Espejo, Alfonso Reyes, Organizational Systems: Managing Complexity with the Viable System Model, p. 166.
  2. Ibid.
  3. John O’Neill, The Market: Ethics, Knowledge and Politics, pp. 136-137.
  4. Espejo, Reyes, Organizational Systems, p. 166.

Mutual Exchange is C4SS’s goal in two senses: We favor a society rooted in peaceful, voluntary cooperation, and we seek to foster understanding through ongoing dialogue. Mutual Exchange will provide opportunities for conversation about issues that matter to C4SS’s audience.

Online symposiums will include essays by a diverse range of writers presenting and debating their views on a variety of interrelated and overlapping topics, tied together by the overarching monthly theme. C4SS is extremely interested in feedback from our readers. Suggestions and comments are enthusiastically encouraged. If you’re interested in proposing topics and/or authors for our program to pursue, or if you’re interested in participating yourself, please email or

Anarchy and Democracy
Fighting Fascism
Markets Not Capitalism
The Anatomy of Escape
Organization Theory