Notes on: Nail, T. (2017) What is an assemblage?. SubStance 46 (1) ( Issue 142) pp21--37. Retrieved from https://muse.jhu.edu/article/650026
Deleuze apparently said in an interview that the concept of assemblage represents the '"general logic"' at work in ATP, but the general logic was never formalized, producing problems for later commentators, including DeLanda, who developed his own neo-assemblage theory. This paper is an attempt at formalization, based on to the What is Philosophy and ATP.  Apparently 'core formal operations' shared by all assemblages can be identified (21).
The first problem is that assemblage is a translation of the French word agencement, which seems to imply a more active arrangement or piecing together.  Assemblage by contrast means the junction of two things or a coming together.  [The difference is in activity implied by agencement]. 

Further, the French term implies an arrangement of heterogeneous elements.  As further implications, unity is replaced by multiplicity, and essence by events. 

Unity implies intrinsic relations between the parts, functioning to produce a whole, an organic relationship that means the parts only makes sense in terms of the whole [only functionalism? Nothing like a 'mode of production'?].  The whole maybe organized by 'principle, spirit' (23).  The implication is that the parts cannot recombine without destroying the whole, that they do not exist independently.  The mechanism [can still be functionalist] on the other hand relates the parts by external relations of 'composition, mixture, and aggregation', and elements can be recombined.  In WiP, an assemblage is seen as a dry stone wall rather than a jigsaw puzzle, and given that assemblages are multiplicities, the 'in between' relation becomes important [in fact it seems to involve a 'set of inseparable relations'. It also implies radical heterogeneity of the authors in the case of a writing relationship?].  The self subsisting elements can be thought of as singularities [this is the first case where heterogeneity is overdone in my view—can elements not also be normal points as well as interesting ones, repetitions instead of singularities?]
Similarly essence is rejected, partly because it involves circularity of definition.  The question for D and G concerns genesis—how things have emerged and from what viewpoint we are grasping them, 'questions of events' (24). Assemblages consist of ' only contingent and singular features' (24), so there is nothing internally necessary to them.  The 'network of social and historical processes' produce assemblages [with a silly and extreme example of all the factors that have led to the production of a particular book—so we are not allowed to select among these equal heterogeneous elements?  How on earth could we account for all of them anyway?  This is a kind of worship of singularity and heterogeneity as in 'aversion to the general' ].  Accident and event are what really produces essences [citing Difference and Repetition].
We have to remember that there is a constructive process behind specific arrangements, not just mixtures.  At the same time, there may well be shared features to the arrangements, even though 'all assemblages may be singular and heterogeneous' [implies rigorous separation between structure and content?]. 

This is what D and G referred to as the abstract machine, a network of specific external relations to hold the elements together, the 'conditioning relations'.  The machine is abstract because it does not actually exist in the [non virtual] world but produces concrete elements and agencies [so far it is just like a conception or a theory then?].  However it is 'absolutely real' (25)—that is the relations are real and these act extrinsically on 'concrete elements'. Abstract machines function as ' the kind of local condition of possibility' [but then a circularity because these relations are detectable because 'elements appear to be meaningfully related']. D& G themselves refer to abstract machines as causing [my emph]  the given to be given although it is not itself given [classic—ATP], something which is 'Real-Abstract' [and thus] not transcendent.  It follows that it is difficult to represent an abstract machine, especially since 'the relations are immanent to the elements', so it is common to describe the abstract machines as a proper noun referring to a unique set of relations, [e.g. May '68, or people like Lenin].  Common nouns assume essentialist categories. 

The proper name given to abstract machines is a-signifying, [not a signifier, but an iconic sign] just a name like the ones given to hurricanes [also occurring once only—except in September 2017!] [So it can be arbitrary—the example given is the name given to constellations of stars, which can be seen only as 'the proper name for the set of conditioning relations that arrange a set of stars' {according only to our perceptions, though, surely?}].  We need such relations to avoid 'radical heterogeneity'[but that's the problem—the whole thing depends exactly on radical heterogeneity or takes for granted particular arrangements, or tries to trace them to historical and social origins.  For the latter, the work's done by Guattari's Marxism, in a rather assertive manner].  A constellation is a singular event that can be changed as new stars emerge [I still think it would have been more forceful if he had chosen something more like a solar system, something more 'real']. 

All assemblages have concrete elements as well which are arranged by these abstract relations, as the 'existing embodiment of the assemblage'.  Conversely, existing embodiments can be understood as composing an abstract machine [hence the up and down links between science and philosophy in DeLanda].  The relation between abstract and concrete has to be 'constructed piece by piece...both are mutually transformative'[invites circularity] [unhelpful metaphors ensue such as the abstract machine is the single wave that unrolls all the specific waves].  Changes in concrete elements change the relations in the form of 'reciprocal determination' or even 'coadaptation' in ATP [functionalism again], or even 'reciprocal presupposition'[definite definitional circularity] in WiP..  We cannot ask what the essence is but we should rather ask what an assemblage can do, as 'an empirical question'[and do we need an empirical methodology here or not, or will windy generalisations based on salon chatter do instead?]. We cannot predict events or their consequences [not even probabilistically? clear break with Marxism here than?].  We have to work one step at a time, asking about the forces at work or what the implications might be, where the limits are and so on.  However, it seems that 'the concrete elements are always changing along with their conditioning relations'—'this requires a constantly renewed analysis'[and endless deferral].

Assemblages also need agents [indeed!], but these are personae, 'not autonomous rational subjects, nor are they simply decentred or fragmented subjects incapable of action'(27).  they are 'mobile operators' that connect concrete elements according to abstract relations.  They too are immanent to the assemblage, 'mobile positions, roles, or figures of the assemblage'.  They actually relate the concepts on the plane [although, curiously '"the plane itself needs to be laid out"'[ citing WiP -- by the philosopher, the other crucial but neglected agent in this ensemble?] The operation [laying out] is not what the persona does.

So, the paradox finally emerges.  We need a persona, which is immanent to the assemblage but 'one cannot have an assemblage without agents that bring it about'.  [Luckily this is solved by philosophical sleight of hand] because each one mutually presupposes the other, although empirically, sometimes personas emerge before planes and sometimes after them.  The persona is necessary to draw 'the relational diagram' of the abstract machine and to establish actual correspondences between concrete elements, so that they can fill out the abstract machine.

We have to think of persona not in the first person but in the third person, as 'collective subjects of an indefinite event (one, everyone, any one)'.  Hence D and G swearing never to use the word 'I' any more [sic -- apparently in AO 23], but to use the third person instead.  This could be 'irony'.  However agents of assemblages always speak in the third person, and even when they speak in the first person they are implying second and third persons that are 'collectively immanent'[sounds a bit like phenomenology again when an individual subject presupposes another?].  We all belong to larger collective assemblages that do the actual arranging of relations and elements which makes the world of the individual agent meaningful.

There are different kinds of assemblage, however—'territorial, state, capitalist, and nomadic' (28). This is not to be understood as 'biological, literary, musical, and linguistic types – that is a matter of content rather than arrangements.  These arrangements will be political—'everything is political', and politics comes before being, practice '"actively participates in the drawing of lines"'[apparently quoting ATP 203].  Construction of the assemblage is practical or political.

Territorial assemblages involve coding the elements 'according to a natural or proper usage', but this is 'arbitrarily delimited', as when houses are segmented differently from factories [not arbitrary at all then].  Territories are divided into coded segments which locate each concrete element and the plans of every persona.  Thus we proceed from one coded segment to another, from home to school to work.  Each concrete elements [now the same as a segment?] can expand.  This looks natural normal, and can claim a continuity with the past.  However, segments of code arise from particular cuts: a 'selection cut' which allows something to 'pass through and circulate';a 'detachment cut' that blocks off part of the circulation; a 'redistribution of the remainder' which begins a new chain (29, quoting AO 247).  These are understood as syntheses of coding, connections, designed to ward off chaos and meaninglessness.  Selection can [always?] involve a 'repression of non codable flows', which would threaten chaos.  A '"coded stock"' or identity emerges, and that can develop along A 'filial line', involving 'genealogical or hereditary descent'.  Most human activities are coded like this.

The detachment cut can be seen as a 'disjunctive synthesis', blocking some connections and appearing in the form of taboos or prohibitions.  However what has been detached can be seen as a form of '"residual energy"' producing new chains of codes.  The redistribution of the remainder can be seen as a conjunctive synthesis, again serving to prevent the fusion of all codes [so having political significance?]: the remainder it isolates can be a surplus leading to a new line of code.  Different mechanisms are apparent in things like potlatch ceremonies, practices of raiding and theft, dowries, gifts and so on.

Together coded territories can offer '"relatively supple segmentarity"'(ATP, 208).  Only one concrete element is changed at a time.  Every [?] limit invites an attempt to cross it, every enclosure leaves a surplus which can be redistributed, 'perpetual disequilibrium, making its very dysfunction an essential element of its ability to function'[that is a functionalist account of change].  'Change happens progressively, one concrete point of the time' [if it is to be functional and not schismatic].

State assemblages aim at unity or totalization [not functionalist change?].  Any surplus of code is used for accumulation, not always economic .  This requires a specialized body which can play an active part in harmonizing concrete elements 'around a centralized point of transcendence'[totally uncritical again?]. D and G talk about stratification both vertically and horizontally, which externally constrains the relations which elements conform with each other.  It exploits 'resonance'.  This specialization means neglecting the wider network and strict controls over it.  That includes the domination of abstract machines, which appeared to rise above concrete relations and personae.  This would be full '"state over coding"', with centralized accumulation and control producing homogeneous striated space.

Stratification can be binary, circular or linear. Binary divisions are simplified and one pole is prioritised.  Circular segments offer concentric circles which resonate around an single axis, 'a single point of accumulation' (31). Linear segmentation offers a progressive link between segments, not necessarily a straight line but something more supple, including homogenous segments 'geometrically organized' around a single point [are not necessarily a circle, maybe a polygon --excessive classification again?], or sub space.  Examples might be found in state science, state art, or state linguistics.

Capitalist assemblages specifically involve a further stage of quantification so that relations and codes cease to have qualities of their own.  Change is not produced by progressive development of elements nor by abstract machines [which can include those hijacked by the state].  Instead, an 'agent or persona' forces quantitative relations.  This gives capitalism its axiomatic quality, forcing elements into a relation by quantifying everything.  Codes are reduced and captured to produce 'strictly economic general equivalence between purely unqualified (decoded) elements' (32).  [Lots of this in AO of course].

This axiomatic approach is identical to capitalism.  Indeed 'capitalism is the offspring or result', that which regulates the axiomatic [this is the result of reworking Marxism in AO in linguistic terms].  Capitalism attempts to extend and saturate the axiomatic.  Once assemblages are stripped of inequalities, they can be organized to generate wealth.  Capitalism goes further than the other assemblages in this sense, and organizes them by relations of exchange, understood here as 'a certain version of immanent relation'.

Nomadic assemblages can offer internal changes and combinations 'without arbitrary limit', including natural ones and hierarchical ones.  It was invented by historically nomadic peoples.  The point is to focus on relations not points, so that the 'in-between has taken on all the consistency and enjoys both an autonomy and a direction of its own'.  As a result, nomadic assemblages can offer 'truly unlimited qualitative transformation and expansion'[ideal ones of course not real ones that always require some kind of social stability, often in repressive forms].  They form 'a participatory rearrangement'(33) collecting all the elements of an assemblage and allowing conditions, elements and agents to 'participate equally in the process of transformation'[probably impossible once we get to more than about 20 people].

Nomadic assemblages are the creative and revolutionary ones, in that problems do not receive pre-given solutions but involve full participation in the formulation of problems and the choice of solutions. ATP refers to 'direct participation with our representation or mediation', rendered here as 'participation and self management' instead of territorial hierarchies.

Of course, none of these assemblages are ever pure.  There are mixtures, different tendencies, a politics, and we have to understand them specifically.

There are thus four different processes of change, involving combinations of lines of escape and points intended to cut or recuperate them.  In other words, four mixed types of deterritorialization [to be grasped again through an empirical process of trying to find out how assemblages work].  The four types include: (1) the 'relative negative' that maintain and reproduce established assemblages; (2) 'relative positive' that are ambiguous in their effects, either reproducing or changing; (3) 'absolute negative' that destroy assemblages; (4) 'absolute positive' processes that create new assemblages [but what makes them positive is neglected here—needs Guattari].

(1) This is reproduction [dynamic, reformist, incorporating]—'adaptation to popular demands' (34), or what D and G call '" compensatory reterritorialization obstructing the line of flight"'[conclusion to ATP].  Lines of flight here mean 'expressions of political realities different from the established ones' [in fact lines of flight themselves are of the same relative and absolute positive and negative types] in this way, particular assemblages can be simply reproduced.

(2) This is a strangely ambiguous type [either from weaselling or obsessive classification I imagine].  It is sometimes not clear what the effects might be of a line of flight.  It might be merely an anomaly, or some kind of exceptional individual with a potential [ATP cited again, 247].

(3) This undermines all assemblages, although it is easily recaptured by relatively negative forms.  D and G say this is the worst that can happen, 'suicidal collapse' that only brings back the old forms of repression.

(4) This will lead to a completely new assemblage, [sometimes?] reconnecting with other elements that have escaped.  The new world is suggested or created [these two options are quite different of course].  [What makes it practical is that] it's based on existing processes of deterritorialization which get connected up.  It is a creative process, revolutionary and constructive.

So we need to clarify these different possibilities if we are to use the analysis.  Otherwise we will see all forms of changes valuable, and even support 'spontaneism, "the worst that can happen"'(36).

Once we've done this work, we can see that there is a fully fledged assemblage theory, providing 'all the core concepts and typologies' that we need.  We are offered a basic structure of abstract machine, elements and agents shared by all assemblages regardless of content.  We have four basic political types, and four different kinds of change or deterritorialization.  [Naturally] actual assemblages will be mixed, and we will need specific analysis to understand 'how the the the assemblage functions'(37).  Then we can direct assemblages 'towards increasingly revolutionary aims'—this issue has been explored elsewhere, including in SubStance [by him].  Nevertheless we need this clarification to produce more 'rigorous and consistent...  methodology'.  [but this is not pursued—it just means philosophical speculation again].

Deleuze page