References for “Mechanical failures and anarchist freedom”

In this essay, I talk about metaphors for freedom among anarchists. I particularly discuss a metaphor concerning failure in complex systems, pointing out that anarchists relate freedom to the deep transformation of social patterns. Here are the references I cited in this essay:

  • BENALLY, Klee. Introduction: Indigenous Anarchism is a collect call. In: BLACK SEED (Ed.). Not on Any Map: Indigenous Anarchy in an Anti-political World. Writings from Black Seed, a journal of green and indigenous anarchy, and beyond. Berkeley: Pistols Drawn, 2021. P. i–xiii.
  • BOOKCHIN, Murray. The Ecology of Freedom: the emergence and dissolution of hierarchy. Palo Alto: Cheshire Books, 1982.
  • DAY-WOODS, Shaun. Dancing & Digging: Proverbs on Freedom & Nature. Night Forest Press, 2021.
    • I have reviewed this book here.
  • DIAS, Álvaro Machado. Por que algoritmos decisórios falham. Folha de São Paulo, 2023. Available from: Visited on: 4 Nov. 2023.
  • FERGUSON, Kathy E. The Feminist Case Against Bureaucracy. Philadelphia: Temple University Press, 1984.
  • GELDERLOOS, Peter. Worshipping power: an anarchist view of early state formation. Chico: AK Press, 2016.
  • [on Guy Debord:] GRAEBER, David. Direct Action: An Ethnography. Edinburgh: AK Press, 2009.
  • GRAEBER, David. The utopia of rules: on technology, stupidity, and the secret joys of bureaucracy. London: Melville House, 2015.
  • [on Diego de Santillán:] GRUPO DE ESTUDIOS JOSÉ DOMINGO GÓMEZ ROJAS. 101 definiciones del anarquismo. 2nd ed. Santiago: Editorial Eleuterio, 2017.
  • HECKERT, J. Anarchy without Opposition. In: DARING, C. B. et al. (Eds.). Queering anarchism: addressing and undressing power and desire. Edinburgh: AK Press, 2012. p. 50–59.
  • KINNA, Ruth. Kropotkin: Reviewing the Classical Anarchist Tradition. Edinburgh: Edinburgh University Press, 2016.
  • LAURSEN, Eric. The Operating System: An Anarchist Theory of the Modern State. Chico: AK Press, 2021.
  • MORGAN, Richard. The Making of Kropotkin’s Anarchist Thought: Disease, Degeneration, Health and The Bio-Political Dimension. Milton: Routledge, 2021.
  • [on Hobbes:] NEOCLEOUS, Mark. The Monster and the Police: Dexter to Hobbes. In: CORREIA, David; WALL, Tyler (Eds.). Violent Order: Essays on the Nature of Police. Chicago: Haymarket Books, 2021. P. 141–158.
  • PELTON, Josh. Engineering Wisdom: Tolerance & Failing Gracefully. 2018. Available from: Visited on: 31 Oct. 2023.
  • PELTON, Josh. How Complex Systems Fail. 2021a. Available from: Visited on: 13 May 2022.
  • PELTON, Josh. Social Rationality. 2021b. Available from: Visited on: 18 July 2022.
  • RECLUS, Élisée. A evolução, a revolução e o ideal anarquista. Translation by Plínio Augusto Coêlho. São Paulo: Imaginário, 2002[1898].
    • There’s an English version here, but it’s just excerpts, the one I’m referencing is much longer.
  • ROBINSON, Cedric J. The Terms of Order: Political Science and the Myth of Leadership. Albany: SUNY Press, 1980.
  • [on Zapatistas:] SHENKER, Sarah Dee. Towards a world in which many worlds fit?: Zapatista autonomous education as an alternative means of development. International Journal of Educational Development, v. 32, n. 3, p. 432–443, 2012.
  • TRESCH, John. The Romantic Machine. Chicago: The University of Chicago Press, 2012.
  • VARELA, Francisco. Reflections On The Chilean Civil War. In: BROWN JR., God. Edmund G. et al. (Eds.). Lindisfarne Letter 8: The Cultural Contradictions of Power. Crestone: the Lindisfarne Association, 1979. P. 13–19.
  • WOODS, David D. et al. Behind Human Error: Cognitive Systems, Computers, and Hindsight. Columbus: The Ohio State University, 1994.
  • ZUBOFF, Shoshana. The Age of Surveillance Capitalism: The Fight for a Human Future at the New Frontier of Power. London: Profile Books, 2019.


Anarchists often employ metaphorical imagery when discussing liberty, but it is often hard to tell when metaphor ends and literal allusion begins. For example, David Graeber conceived of freedom as a tension between a “play” principle in human activity and the rule-bound games we generate when we play. Is this a metaphor, or are we literally to conceive of all human institutions, like the family, capitalism, or the state, as akin to football or poker? What about Murray Bookchin’s “ecology of freedom”, seeing as we are, indeed, animals?

Some turn to zeitgeist-capturing technology to make a point: for Guy Debord, writing during the rise of the television in 1968-ish France, people were becoming an audience to their own lives; for Eric Laursen, the State is like a computer’s operating system. Others prefer physical terms, in different levels of abstraction. Heckert wrote that to hold tightly — to shame, resentment, or any emotion or any story of how the world really is — is to be held tightly, and this is not freedom; to hold gently is to be held gently, which is freedom. Kropotkin, as Richard Morgan writes, employed images and tropes from bio-political science but took them literally and made them real in a radical political framework.

Élisée Reclus once warned us that proverbial formulas (often metaphorical) are dangerous, as one happily acquires the habit of repeating them like a machine, as if to avoid reflection. This message also contains a metaphor about machines. I think this is actually a very useful figure of speech to discuss what liberty means within anarchism. This is for lots of reasons. It symbolises — as noted by Reclus — unconsciousness even amidst action. Also, anarchism took shape as the industrial revolution took flight, the latter demanding, as Cedric Robinson wrote, a kind of submission which invaded every recess of the worker’s existence. It therefore references the anarchist criticism of humans literally made to perform inhuman rhythms.

In this essay I want to discuss anarchist freedom by digging a little deeper into a particular metaphor using the figure of the machine: the notion of “failure in complex systems”, an idea from the field of engineering, at least as I read it through the work of David Woods and colleagues. I am also going to be quoting Josh Pelton a lot: he’s an engineer with a penchant for philosophy and sociology and a marvellous Youtube channel called Thunk, which I highly recommend.

When operating large-scale machinery, such as aeroplanes or nuclear reactors, people tend to notice and fix what looks broken, but if everything is shut down every time a light bulb needs changing, nothing is ever going to get done. Such systems are thus usually designed with enough margins for errors that allow them to keep running in spite of issues; repairs happen when they’re necessary or when there’s a convenient opportunity. Complex systems, then, run in a constant state of slight disrepair.

Sometimes, however, a combination of things left unfixed leads to catastrophic failure, hurting people. When this happens, it is common to search for a single cause that explains the disaster, especially if it allows for attribution of guilt. Quote-unquote “Human error” is quickly found, for even if the cause was mechanical, someone may, for example, have missed evidence that something was wrong, or not have planned adequately for it from the beginning. To avoid future issues, the culprits might be exiled, punished, or simply retrained.

However, Woods and colleagues explain that there is no such thing as human error. Different knowledge of events and context, or different goals, lead to different judgments of people’s performances. As Pelton summarises, human error is above all an artefact of hindsight bias, which recontextualises something as if it obviously contributed to the disaster in a way that anyone should have seen coming. In other words, there will always be ways to explain failure through negligence or incompetence, just because in a complex system there’s always something going wrong. Erroneous actions should only be the starting point for an investigation, but one that inquires the system itself. Human error, Woods and colleagues write, is not some deficiency or flaw or weakness that resides inside people, being rather a result of their interaction with the system.

On the other hand, there is a counterproductive way to focus on the system. One may, Woods and colleagues write, restrict the range of human activity, perhaps by policing practitioners so they more closely follow the rules, or by introducing more automation to quote-unquote “eliminate people” from the process altogether. But, as Pelton notes, adding more moving parts to take humans out of the loop mixes in new variables that might fail, and therefore tends to produce even more errors — and crucially, more judgments of human error. Emphasis on increasing efficiency generates more pressure on operators, write Woods and colleagues, and additional technology creates new burdens and complexities for already beleaguered practitioners, leading to new modes of failure.

How can we reliably avert disaster, then? Well, failures must be seen as opportunities to learn and change. Instead of using investigations to find out how others failed (presumably to punish them), information flow must be rewarded so that better decisions can be made and issues are less likely to be overlooked. Pelton summarises the best way to prevent critical failures in complex machines as simply… Empowering people to spot and fix problems as they crop up, making the system more adaptive and robust. He also notes that repair is messy, idiosyncratic, unpredictable, and unlike assembly lines, not conducive to automation or rote procedure; it is artisanal work, requiring artistry, problem-solving, expertise, as well as the time needed to experiment. Hence empowering autonomous, decentralised repair includes training and experience, but also providing ample resources, as well as understanding what pressures we’re applying to individuals that will influence how they make everyday decisions about safety.

If the metaphor isn’t obvious yet, I’m going to spell it out. Anarchists can be seen as making a similar case, with the “complex machines” in question standing for “societies” or “social patterns”, machines made of their own operators, who “run” themselves to perform the “task” of… living their own lives! Interaction in a sufficiently large group always involves conflict (that is, “we” too collectively live in constant “slight disrepair”). Yet, people do not usually suspend every commitment they have as the result of a quarrel until their relationships are perfectly harmonious. Everything goes on until “rituals”, from phone calls and meetings to interventions and assemblies, provide “convenient opportunities” for “fixes”, that is, for restoring people’s willingness to cooperate with one another. Still, a lot can be missed because of all the complexity involved. Problems might compound and combine until they cause harm on a greater scale.

Anarchists were not the first to point out that such harm is not a matter of human error. A lot of post-republican Roman thinkers, for example, basically said as much when they concluded that the Republic ruined not because people lacked virtue, but because of constitutional shortcomings. However, for anarchists the solution cannot be bureaucratic, like “taking humans out of the loop”. With machines, after things go wrong we often notice overlooked signals that could have indicated vulnerabilities. With social relations it’s similar, and improving social consciousness, that is, seeing and appreciating the significance of such signals, should be the bread and butter of how we relate. This would empower us to recognise potential problems and act directly on them (you know, “to spot them and fix them as they crop up”) in a collective, yet decentralised manner, even if this means deep social transformation (that is, in the metaphor, the system is made “more adaptive”). Doing so requires education (in the metaphor, “training”), sharing and rotating responsibilities (in the metaphor, “experience”), and common access against the artificial scarcity of private property (in the metaphor, “ample resources”) to combat relevant inequalities (in the metaphor, the “pressures applied on individuals that influence their decisions”) so that people retain their capacity to keep on “fixing” and therefore improving their relations.

Moreover, this metaphor also fits anarchism well in the sense that ongoing negotiation and adjustment is emphasised over social blueprints — and by this I don’t mean to contrast repair with creation but to think of repairing as creative activity. Of course it involves forecast and planning for robustness (also known in engineering as designing for tolerance); it encompasses, speaking of societies rather than machines, accounting for path dependence and keeping in mind the coherence between means and ends. Still, obsessing over the initial moment of creation misses the point that, by fixing something, someone might in the future, as Pelton discusses the fixing of broken objects, think of how fun it was to see the insides of the thing, how they’ve learned the right way to do something, and how a friend helped them out. These are all things that can’t be designed into a system: people have to decide how to restore its functionality, and that decision may change or subvert the values of the designer.

Metaphorically, then, this is all about institutions that are constantly reassessed by confident, socially conscious individuals. Of course sovereignty-supporting political traditions, I’m talking here about Marxists, liberals, social democrats, etc., of course they also want to “avoid disasters”. But if you think about it, for them a disaster is when predetermined ideal outcomes are threatened — not necessarily causation of harm. After all, they are in favour of there being national armies and police and prisons and whatnot, which are meant to cause harm and threaten those who would threaten their ideal constitutions.

Indeed, the point of all conformist social engineering is to induce people to not mind being harmed; as Kathy E. Ferguson writes, bureaucracies seek to “tie up our loose ends” and reduce us to a reflection of an organization. In this sense, society under patriarchy, white supremacy, statism, capitalism, and other forms of intersecting domination really begins to resemble an aeroplane or a nuclear reactor: a machine tied to a specific purpose regardless of what kind of machine people might require in their lives now or in the future – in yet other words, the designer’s values cannot be subverted.

Even further, each individual becomes like an expendable, replaceable part, evaluated according to their “performance”: they are increasingly rendered… machine-like. To respect state representatives, ponders Peter Gelderloos, is to mistake them for reasonable human beings rather than the organic masks that an insatiable machine wears in order to extend its power.

Of course, every social system is self-reinforcing, and that is actually good; anarchists also want anarchy to be sustainable. But the kinds of social systems anarchists envision are qualitatively different. They include in their own enacting a way to undo themselves, as Francisco Varela puts it; anarchism, Klee Benally analyses, is a dynamic politic that invites its very destruction while maintaining composure of core principles; it tries to create a healthy culture as Shaun Day-Woods defines it: one that doesn’t keep anything that can’t be destroyed.

Instead of humans becoming more like machines, the institutions that anarchists are intent on promoting are supposed to be more organic: balanced, open-ended, diverse. Creation is always anarchist, wrote Diego de Santillán, and so are creators if they do not create in view of automating themselves. As Ruth Kinna discusses regarding Kropotkin’s vision, his defence of organisational proposals being always open to revision and dissent ultimately meant that society should be a living, evolving organism. Even more, in the end, there is no “one”, single society; no single “social machine”; like the Zapatistas, who walk toward a world in which many worlds fit, anarchists also conclude that the challenge really is to integrate multiple utopias; to render diverse social machines interoperable without disasters.

In the end, the problem is not exactly machines themselves. Analysing romantic tropes, which were somewhat influential among anarchists, John Tresch notes that machines drew forth virtual powers and brought about conversions among hidden forces; they could be used to create new wholes and organic orders, remaking humans’ relationship to nature and renewing nature itself. Anarchists seem to draw a line between creating and using machines on one hand and, metaphorically or not, being one, being absorbed into one, on the other. If Hobbes himself described the Leviathan as a huge machine, anarchists often attacked this “mechanical” political reasoning, exploring what “more organic” relations would look like.

Of course, to heed Reclus’s warning, this metaphor is far from straightforward. If social machines are made of ourselves and our relational patterns, there may be a difference in how organic or mechanic they are, but the boundaries between creating, using, and being them are definitely blurred, to say the least. Even if the metaphor is not perfect, however, I still think it deserves attention given recent technological developments.

For Shoshana Zuboff, present-day global technology conglomerates commodify our behaviour, shaping it at scale through automated machine processes that nudge, coax, tune, and herd us toward profitable outcomes. It is no longer enough to automate information flows about us; increasingly the goal is to automate us. The enormous amount of data produced about us enables automated decision-making, private and public, leading to what Álvaro Dias calls dystopian iteration: when the future, conceived as a straight line, without surprises or chances of being transformed, is reified by algorithmic decisions, crystallising imbalances and externalities.

In other words, if anarchists criticise the building of agency-crushing sovereign “machines” out of our own actions and dispositions, this is becoming less metaphorical as time passes. The automated, racist control of predictive policing, credit scores, and algorithmic bosses; advances in the emulation of human likeness and knowledge; the grim suffocation of war by drone surveillance and AI-targeted strikes; the invasiveness of pregnancy-guessing gadgets: all of this seems to strengthen the anarchist case for a rethinking of what “freedom” means. At least, that is, if this notion is at all to remain a guiding aspiration for the future.