The Assemblage - Working Together

by Stefan Morales | Last Updated: March 17, 2024

This set of notes elaborates on the philosophical/ social science definition of Assemblages, and it directly links to this earlier note from a piece on self-governing without guarantee:

“The appropriate scale to Intentionally Engage in system transformation is not just the Spatial Scale of the micro-level, meso-level, or the macro-level, but rather the Intensive Scale: navigating, and co-developing internally and externally connected and coherent assemblies, which operate via shared authority and action, in, through and between existing and new configurations of Social Objects (aka Assemblages).”

This longer set of notes on the assemblage can stand alone to a degree, but I would strongly suggest contextualizing it by reading about self-governing without guarantee, and then coming back here. OK, to define this concept for our purposes:

Assemblages are complex entities formed from the interactions and combinations of diverse, independent components, both material and expressive (and a mix of the two), which maintain their individuality while contributing to the whole’s function and identity.
In simple terms, we care about this concept because we wonder about how something came to be, what it is now, and what will happen next. The deeper our shared inquiry around this, the deeper our appreciation of the wonder (and ridiculousness) of this world, and the more we can step outside of our biases in how we respond to it.

While there are many different approaches to assemblage theory, I believe that the main sentiment beneath them all is similar, and the main differences are more a matter of degree, emphasis on different “moves”, and splitting hairs than anything radically divergent. So this set of notes will aim to:

Unpack a working definition of the concept of the assemblage from Manuel Delanda’s work; (vital to understand what I’m working on)
Lay out alternative takes from the philosophical and social science lit, specifically Ian Buchanan’s work; (also vital)
Show the connection between Assemblage Theory (AT) and Actor Network Theory (ANT); (optional nerd out).

Lastly, I recommend dipping in and out of this piece, as these are fairly dense notes!

Working Definition

For an already fairly complex concept, though, I find Manuel Delanda’s summary of “Assemblage Theory” from A New Philosophy of Society to be the most accessible way in as it tightly gives us the central characteristics of an assemblage, as well as the basis for a few different ‘thinking space’ diagrams which I will present in a future piece. Let’s unpack Delanda’s summary:

“First of all, unlike wholes in which parts are linked by relations of interiority (that is, relations which constitute the very identity of the parts) assemblages are made up of parts which are self-subsistent and articulated by relations of exteriority, so that a part may be detached and made a component of another assemblage…”

To elucidate, consider yourself, going about your day. At one point you are embodying perhaps the ‘role’ of a parent or caregiver or friend. Then you are at work, embodying other roles – other ways that you ‘tie into’ the organization that you generate your livelihood from. You might receive some mail later that day inviting you to participate in a survey your local government is running. The amount of roles and personas embodied in you (that are turned on or off, conflicting, overlapping, ‘tied into’ or not, their different paces of activation, their interfering frequencies, etc.) as well as the different social objects you are interacting with in varying degrees of intensity – and all of this dancing across the Cynefin domains.
And this is simply looking at the social object of YOU, as an ‘individual’! And here we are not even stressing the diversity of experience, capability, capacity, etc. that each unique individual has. What of other social objects that the social sciences account for and study? This is what we aim to get better at exploring, navigating, and assembling within – without the perverse dynamics of guarantee. Delanda goes on:

“…Assemblages are characterized along two dimensions: along the first dimension are specified the variable roles which component parts may play, from a purely material role to a purely expressive one, as well as mixtures of the two…”

Here we need to simply recognize that the ‘roles component parts may play’ can be taken very broadly: whatever we experience as a defined feature, aspect, or characteristic that plays a significant role in the assemblage one is considering. They can be purely material qualities (like energy, nutrient cycling, physical violence, raw materials etc.), or purely expressive (rules, codes of conduct, email communication, etc.) or a mixture of the two (facial expressions, gestures, policing and enforcement mechanisms, disciplinary regimes, etc.) that play a role in the assemblage.

“A second dimension characterizes processes in which these components are involved: processes which stabilize or destabilize the identity of the assemblage (territorialization and deterritorialization)…”

Here we are talking about one half of the “double articulation” process (explored in the previous set of notes). And I do like the simplicity of ‘stabilize’ and ‘destabilize’ and it will probably find its way into the more accessible Liberating Structure I’m designing (…but it does lose the synesthesia quality of blending the spatio-temporal that deterritorialization and territorialization give us).
We are talking about the way our identity can be grounded in a place, cordoned inside or outside of a physical or conceptual border, made more or less of an ‘earthling’ on a planet that sustains life, etc.
Territorialization/ed is the activity or result of distributing or disposing persons or things properly or methodically, it’s about establishing localized well-defined boundaries, distinctions or identities around something that gives the entity its coherence, it’s an increase in internal homogeneity as a result, it is grounding and grounded.
Deterritorialization/ed is more the activity or result of lateral, non-methodical disposal and disordering, it removes the boundaries or identifying attributes of something, increasing internal heterogeneity in the process.
In the context of a ‘Socio-Ecological Object’ SEO, you can see how geomorphic constraints become critical to an understanding of territorial and deterritorial. And you can see how many human institutions and organizations today are highly mediated from these geomorphic constraints: that we basically design our social objects to be ‘at one remove’ from geomorphic constraints. This notion of ‘at a remove’ or ‘at one remove’ is archaic or academic English, but I like it as it stresses a certain quality of gradation or step, of connection and disconnection at the same time. For French we might say something like ‘de façon interposée’, for German something like ‘wegversetzt außen’, but there doesn’t seem to be an equivalent in other romance or Germanic languages.
Nonetheless, these concepts can feel clunky, and I imagine it’s also a matter of the indo-European language family needing to take a foray into other languages where we might open up new vistas for understanding this spatio-temporal blend. Another key word here is process, but I’ll speak to that below.

“In the version of assemblage theory to be used in this book, a third dimension will be added: an extra axis defining processes in which specialized expressive media intervene, processes which consolidate and rigidify the identity of the assemblage or, on the contrary, allow the assemblage a certain latitude for more flexible operation while benefiting from genetic or linguistic resources (processes of coding and decoding).”

This is the other half of the “double articulation” process, and the one that has arguably been given the most truck and trade in the social sciences over the past half century’s foray into linguistics, semiotics, social constructionism, postmodernism, etc. But this fascination with language has often stressed interpretation, literary theory, etc. Delanda is going much broader, allowing us to explore the ways in which coding processes and decoding processes are both material and virtual, operating across all media – human, synthetic (maths, computation), biological.
Zelig Harris is instructive here for linguistics, providing a materially rigorous account of the way constraints operate on information in language, sign language, math, and music; but so too is the more accessible read: Jeremy Campbell’s Grammatical Man. And it goes without saying that Sharma and Czégel are trying to articulate a theory that accounts for the gap between physics and biology (loosely, at least, for now), the latter of which is governed by genetic resources which constrain the set of biological possibilities that can emerge from the ‘non-ergodic’ corner of reality called life.
But what is most meaningful to humans assembling their social objects will tend to be the coding processes that are used to shape the content of shared purposes, rules, agreements, legal frameworks, protocols, actions, strategies, etc. as well as the equally important (but much less developed) decoding processes that hospice old, rigid forms.
And as for the SEO… these must be more expansive in their recognition of coding and decoding processes, leaning into the consolidation and flexible operation of identities via human and non-human coding and decoding dynamics. I wrote about this elsewhere many years ago, but it is a more philosophical piece, using the concept of a ‘chemical sign’ to expand the reader’s awareness of non-human forms of information/meaning making, and not geared towards my current focus on scalable practice. A more simple way of putting it would be to say that the coding and decoding processes of an SEO would be as grounded in information, meaning, constraints, etc. generated by humans as by non-human entities, ecosystems, etc.

“All of these processes are recurrent, and their variable repetition synthesizes entire populations of assemblages.”

Process is key in all of this, as we are invited to consider time here: the history of the assemblage, it’s present capacities and potentialities and future possibilities, probabilities, plausibilities, and preferences. Population thinking is critical to this aspect of variable repetition, where historical contingency and selection pressure limits a population of assemblages from becoming whatever adjacent possible form they may take, or whatever adjacent possible ways of stabilizing, developing, growing, etc. they can adopt. Sharma and Czegal unpack this concept further, and Delanda explores concepts like ‘population thinking’ in his book on Intensive Science.
As far as I can tell, however, these theorists do not directly address the flip-side of this process: where chaos dynamics, collapse, etc. unmoor and release existing assemblies, or destroy them entirely. The ‘loosening-up’ or ‘freeing dynamic’ that follows from collapse and release is equally important to consider, as new opportunities and niches for radical innovation emerge from a previously more homogeneous context. For this lens we must turn to the backloop of the Adaptive Cycle and explore Panarchy theory.

One thing I appreciate about assemblage theory is the way it can allow you to pull in diverse perspectives, theoretical tools, practices, and empirical studies about what matters for a given assemblage in terms of dynamics, constraints, etc. It’s not about which one is right, but more about which combination of concepts, tools, etc. helps us be less wrong about ‘the what’ and more attuned to how we might best navigate uncertainty and complexity together (the ‘what if’ and the ‘now what’). In a sense, by considering the historical formation of assemblages, we can create a better collective ‘probe head’ through accessible and shared practices, scaffolds, mindsets, tools, etc., that we then use to attend to the present and future. For me, all this is in service of iteratively designing better socio-technical instrumentation for navigating the Bayesian middle that we find ourselves living out at the human scale.

The last bit I’ll say here on Delanda’s notion of the assemblage is this helpful characterization: an assemblage is not simply a collection of parts. A single, simplified example from Deleuze and Guattari’s work can help sharpen all of the above, I believe. A skilled archer riding and battling on horseback is an assemblage of human-bow-horse, whereas as Delanda elaborates, an unskilled person trying to fire a bow and arrow while riding on horse for the first time is a ‘collection of parts.’ To transform the latter into the former requires:

a process of repetition characterized by…
a double articulation of identity formation (constituting the boundaries for practice, purpose, inclusion, etc. and the instructions, lessons, etc. that constitute the practice), that…
synthesizes a recurrent population of obvious (human-bow-horse) and non-obvious (index finger muscle, etc.-tensile strength, etc.-breed personality, etc.) material and expressive components into the working assemblage, over time.

So, when we are talking about navigating and assembling for transition, for example, we are talking about making this process of assembly formation explicit to serve the specific articulations of transition, energy descent, adaptation, permaculture, etc., which broadens the types of ‘skilling’ well beyond the human. Which takes us to the concept of a SEO, which I’ll share more about in future notes.

Alt. Takes: Agencement, Komplex

I think it’s helpful to show how some aspects of this central concept were lost in translation when Gilles Deleuze and Felix Guattari’s work was brought to English readers. And Ian Buchanan’s rejoinder to Delanda is a helpful place to map this, as Assemblage was the chosen English translation of the French word agencement, where

“Agencement derives from agencer, which according to Le Roberts Collins means ‘to arrange, to lay out, or to piece together’, whereas assemblage means ‘to join, to gather, to assemble’. Agencement, as John Law has noted, encompasses a range of meanings that include ‘to arrange, to dispose, to fit up, to combine, to order’.“ (Assemblage Theory and Method, 20)

It is in this sense that we can see a more expansive understanding of assemblage that allows us to consider other ways in which a ‘collection of parts’ can be ‘joined up’ beyond a materialistic reading that tends to prioritize the degree to which the assemblage is a ‘working assemblage’, or ‘does something useful’, or ‘produces an effect.’ All the ways this idea can fold so well into concepts like Constraint Closure, etc. To this end, Buchanan argues the extreme opposite: that what makes an assemblage an assemblage is the fact that it doesn’t work:

“The second observation I want to offer is to note that in his comments on Man Ray’s piece ‘dancer/danger’, Guattari (in his essay ‘Balancing-Sheet Program for Desiring Machines’, which was appended to the second edition of Anti-Oedipus and can therefore be read as a kind of bridging piece linking Anti-Oedipus to A Thousand Plateaus) observes that what is crucial about this assemblage is the fact that it doesn’t work. He means this quite literally. The working parts, its cogs and wheels and so on, do not turn or intermesh with one another in a mechanical fashion. It is precisely for that reason, he argues, that it works as a piece of art. It works by creating an association (i.e. a refrain) between the human dancer and the inhuman machine, and thereby brings them into a new kind of relation which he and Deleuze would later call the assemblage, but in their first works they called the desiring-machine.” (Assemblage Theory and Method, 22)

So in this sense, Buchanan’s interpretation of Guattari weights the expressive components of an assemblage, and how a relationship between these parts emerges, not due to instrumentality, but because they are art. In a way, the assemblage becomes an assemblage because of the wider socio-cultural assemblages that ‘dancer/danger’ fits within. And so one could also argue that there is a lot of work happening around the ‘work of art’ to make it work as art, if you catch my drift. Now, what if one has a profound encounter with a collection of parts that is not a work of art, but that aesthetically ‘hits’ you? I think we start to approach the individual as an assemblage of sub-personas, affects, etc. and other components (some material, some expressive), and the way that one ‘ties into’ the world beyond skilling, work, etc. and into the realm of aesthetic intensity, emotions, imagination, ritual “entrainment”, altered states, the abnormal, etc. And this is why I especially appreciate what Buchanan draws our attention to next:

“…it is useful to note that agencement is Deleuze and Guattari’s own translation, or perhaps rearrangement would be a better word, of the German word Komplex (as in the ‘Oedipal complex’ or the ‘castration complex’). Although it is Guattari himself who defines the assemblage in this way in the various glossaries he has provided, the connection between Freud’s notion of complex and the concept of the assemblage has been almost completely ignored. If there is any word whose meaning one should explore as a way into assemblage theory then it is complex. According to Laplanche and Pontalis’s exhaustive account, there are three senses of the word complex in Freud’s writing: (1) ‘a relatively stable arrangement of chains of association’; (2) ‘a collection of personal characteristics – including the best integrated ones – which is organised to a greater or lesser degree, the emphasis here being on emotional reactions’; (3) ‘a basic structure of interpersonal relationships and the way in which the individual finds and appropriates his place’. Laplanche and Pontalis also note that there is an underlying tendency towards ‘psychologism’ inherent in the term. Not only does it imply that all individual behaviour is shaped by a latent, unchanging structure, it also allows that there is a complex for every conceivable psychological type. The key point I want to make here is that none of these ways of thinking about the complex actually requires that we give any consideration to a material object.” (Assemblage Theory and Method, 21)

Yes, all this is well and good, in fact, it’s necessary to bring in as we are dealing with emotion, affect, etc., and while Delanda’s outline of Assemblage Theory can help us rigorously work at different spatial and intensive scales of reality, we can lose this psychological quality if we don’t also understand the degrees to which a given assemblage is also a Komplex – and we can see the power of Gauttari’s concept of the Three Ecologies here: environmental (the “it’s”), social (the “I’s, we’s, theys, etc.'”), and psychological (“the hows” as in how something is perceived).
Of course, at the individual level this is clear to see, but in what ways can a conversation be a Komplex? Or a team? Or an organization? We speak of individuals as ‘having a complex’ – is this the same? Is it something that is possessed, or something that one is possessed by?

ANT & AT

I couldn’t go as far as I have unpacking the Assemblage without spending some time exploring Latour’s framing of Actor Network Theory (ANT):

“The alternative I have proposed in this book is so simple that it can be summarized in one short list: the question of the social emerges when the ties in which one is entangled begin to unravel; the social is further detected through the surprising movements from one association to the next; those movements can either be suspended or resumed; when they are prematurely suspended, the social as normally construed is bound together with already accepted participants called ‘social actors’ who are members of a ‘society’; when the movement towards collection is resumed, it traces the social as associations through many non-social entities which might become participants later; if pursued systematically, this tracking may end up in a shared definition of a common world, what I have called a collective; but if there are no procedures to render it common, it may fail to be assembled; and, lastly, sociology is best defined as the discipline where participants explicitly engage in the reassembling of the collective.” (Bruno Latour, Reassembling the Social, 247)

Latour wants us to step beyond ‘social explanations’ because he sees them as carriers of well-worn clichés with no relevance to the actors they purport to ‘explain’. ‘Sociologists of the social’ (as opposed to sociologist of associations) take the word social to designate an already assembled bundle of ties or stabilized state of affairs (Latour, 1).
Instead of taking the social to be something that glues associations together, it asks what is already glued together by other types of connectors that would fall outside of a typical sociological account (such as ‘inanimate’ objects, humans, plants, and, and, and…). There are associations that are ‘always already there’ and they need to be traced, thus, Latour’s focus on the need to trace associations and design their assemblages—‘re-association and reassembling’ (5, 7).
ANT proposes that instead of starting with a social explanation and moving to the fieldwork, one should start with the fieldwork and reassemble what is social about it, what its social aggregates are, and what is new about this association. Sounds simple enough? Well…
For ANT, as we now understand, the definition of the term [social] is different: it doesn’t designate a domain of reality or some particular item, but rather, is the name of a movement, a displacement, a transformation, a translation, an enrollment. It is an association between entities which are in no way recognizable as being social in the ordinary manner, except during the brief moment when they are reshuffled together (65).
Because we’re tracing movements, displacements, transformations, translations, enrollments, and so on, we need to slow things down—Latour emphasizes this again and again—in keeping with the repeated claim to relativity, ANT follows a similar logic of slowness.
To draw an example from Einstein’s theory of relativity: in order to make-sense of the relativity of space-time that results from an object travelling near the absolute speed of light, it is necessary to slow-down the observed reality to make sense of the transformations space-time undergoes at 99.999…% of the speed of light. Before Newton, physicists instead explained the relativity of light by suggesting that it travelled through ether, and was thus, liable to be sped or slowed depending on the degree of fluidity or solidity of the ether, and depending on what aspect of reality one was looking at. Like the ether of the classical physicists, Latour accuses the social of being in a similar obscure position: “an artifact caused by the same lack of relativity in the description”; the social is like a skeleton key or Rosetta stone (102).
‘To follow the actors themselves’ in defining what the collective existence has become in their hands, is to ‘slow down at each step’ and to shy away from quick fixes to complicated meshworks of interaction (i.e. giving the meshworks a unified sort-of glue, or status as an whole or entity, rather than tracing out the associations that hold the group together) (12, 17).
Latour identifies a crossroads in sociology, between the work of Gabriel Tarde and Emile Durkheim, where the ‘sociology of the social’ (Durkheim) came to predominate over the ‘sociology of associations’ (Tarde). According to Latour, we can slowly ‘reassemble’ the ‘sociology of associations’ in the hope that describing will become more predominant in the face of social scientists who, in thinking they are describing society, end up prescribing, or acting as social engineers (14-16).
Delanda’s Intensive Science and Virtual Philosophy challenges traditional metaphysics with a focus on the ‘virtual’ — a realm of potentialities that are actualized in the world through processes driven by intensive differences (such as temperature gradients or pressure differences). This approach emphasizes the role of dynamic interactions and the unfolding of matter and form over time, suggesting that the structures and forms we observe in reality emerge from self-organizing processes that are modulated by these intensive properties. His book tries to draw a lot of different threads together, and present a synthesis of many different contemporary scientific theories (some of which I’ll outline here), arguing that by understanding the virtual capacities of matter and the intensive processes that guide their actualization, we can better grasp the complexity of the natural and social worlds, moving beyond static entities and fixed categories to a more fluid, dynamic understanding of existence.
You can also see then, why Delanda turned his attention to the social sciences: the assemblage offers a way to think about and act within virtual capacities and intensive processes, without falling into the sorts of biases I briefly touch on when I talked about ‘shitty social ontologies’ or that Latour resists in the form of the ‘sociologists of the social’.
But although AT and ANT are very similar, ANT’s emphasis on the network, mobilizes a much flatter social ontology than AT, which recognizes a much broader and diverse ‘legend’ of virtual capacities and intensive processes than those that occur in the social object of the network. The individual, the conversation, the institutional organization, the community, etc. – each of these social objects have different ‘legends’ of factors to consider, and that Delanda explores and gives examples of in his books (here, and here) and lectures on Assemblage theory.
For ANT, human and non-human ‘actors’ constantly work at forming associations, together, and the aim is to toy out the myriad networks of association to illuminate this dance between human and non-human. This is why Latour flips the static notion of the network to the more dynamic notion of a worknet: a broad set of human and nonhuman actors that work at forming the network of associations they are in. This is also why so many studies that use ANT, include non-human actors in their descriptions (Michel Callon’s work on scallops and fisherman – an early inspiration to ANT – is a good example).
While I love ANT (and reading studies like Callon’s!) I find that I can get a bit lost in the descriptions. The question of what was important to the formation, continued existence and possible future development of a given SEO can fade a bit into the background – the entire social and material world becomes embedded in an extensive network of relations that starts to feel a bit flat, all-pervasive, all-possible… When in fact, this isn’t the case: along with a tacit or implicit worknet, there is also a double articulation of identity (de)construction at work for any given social object. From this perspective then, ANT is tremendously useful for understanding and exploring networks – while the static concept of ‘the network’ is a social object in its own right, the myriad virtual capacities and intensive processes of networking or ‘the worknet’, show up between many different social objects (a network of individuals – a network of groups – a network of institutional organizations). In a sense then, while a double articulation can reveal potentiality, it’s surrounding worknet can reveal tacit or explicit associations that are actively being made.

thx for going full-nerd,

Stef