The most rigorous development of in this direction comes from Manuel DeLanda who delves into the framework of assemblage interactions. For DeLanda “the ontology of assemblages is flat since it contains nothing but differently scaled individual singularities.”(10) Thus all relations of belonging, whether of an individual to a population, or an entity to an organization, are considered equally to produce new individuals and not new types. A flat ontology is supported by the Deleuzian position that entities are not defined by the assemblages they participate in. Such “relations of interiority,” where the relations constitute the defining properties of elements—characteristic of organic holism(11)— are contrasted to “relations of exteriority,” in which defining properties and “capacity to interact” are separable,(12) allowing individuals to enter or exit assemblages “without the terms changing.”(13) This leaves the formation of assemblages contingent on development embedded in the temporal dimension rather than dictated by abstract necessity. The historical generation of the assemblage prompts empirical investigation because “there is no way to tell in advance in what way a given entity may affect or be affected by innumerable other entities.”(14) The full capacities of an assemblage may go unexercised and are ultimately unknowable, suggesting a redundancy of causality. Graham Harman sees a difficulty in the fact that assemblages “withhold themselves from their relations with the outer world insofar as they are never fully actualized, and withhold themselves from their own pieces by exceeding those parts and forming a new reality,”(15) arguing that such withdrawal also separates the assemblage from its generative process. The concern is that as an assemblage crystallizes into a particular pattern of existence, the accidents of its formation are no longer differentiating features.(16) Harman forgets, though, the potential for undiscovered capacities to distinguish the assembly in ways that reanimate previously redundant history.
DeLanda is careful to distinguish assemblage ontology from an atomism that only enables causation from the bottom up. He insists on coexistence of differently scaled assemblages and points out that, though always composed of smaler entities, assemblages are most often composed by larger ones(17) often as effects of other expressive or territorializing actions. Naturally, assemblages are also capable of interacting with one another as well. The implication of this (because the interaction must revolve around some relation of exteriority, and because relations of exteriority constitute assemblages by definition) is that relations always produce new assemblages however briefly.(18) This excessive access to the novel is, perhaps, the most revelatory strength of DeLanda's assemblage ontology and it helps to return the active, agentic capacity to a description that at times risks becoming overly schematic. Furthermore, the ties to the material–expressive axis of the assemblage further cement the social character of the assemblage among all those entities which in reality come together to effect the staging or execution of an event.
The other primary strand of thinking in assemblage theory follows Bruno Latour's description of Actor-Network Theory.(19) Latour is similarly motivated by a desire to remove the over-arching frameworks that reduce the specificity of the case at hand(20) in favor of situated non-linear interactions between actors. “when a force manipulates another, it does not mean that it is a cause generating effects; it can also be an occasion for other things to start acting.”(21) The actor is induced to act, but diverse ways in which the agency of the action is figured(22) (in connection with many actors) is highly contingent.(23) Though the actor-network branch has produced some of the more ontologically radical assemblage urbanists,(24) Latour's development of Actor-Network Theory is much more methodological, primarily emphasizing how a researcher or theorist can remain committed to a flat ontological grounding, while remaining intentionally quiet about existence within this ontology. He gives almost no description of groups themselves—“the word ‘group’ is so empty that it sets neither the size nor the content. It could be applied to a planet as well as to an individual.”(25)—offering only the performative definition that the assembling (or disassembling) of groups exist is itself the mapping of social context and that the groups do not exist outside of this action.(26) Latour sees no inherent difference between this action and the acts of a researcher, concluding that “the network does not designate a thing out there… It is nothing more than an indicator of the quality of a text about the topics at hand.”(27) What does exist for Latour is nothing if not concrete, however these actants “are fully relational in character, with no distinction between object and accident, object and relation or object and quality… to change one's relations is to change one's reality.”(28) Again assemblage theory promotes a prodigious new production of entities, but through the complete inverse of what we find in DeLanda. Furthermore, it means that there is no place for withheld properties, all actants are defined by their efficacy.(29)
Despite these contradictions, agent-network theory does illuminate some relational aspects which can be incorporated with the approach adopted in this text. In particular, the production of scale through the construction of linkages or the action of an individual.(30) Even more promising is the potential reversibility of scale relations, by which an individual can incorporate larger assemblages which it may even belong to itself.(31) These topics will return in the next chapter, to establish how the reserve complexity of objects prevents a reduction to merely relations and how the effectuality of relations establishes the social interplay that generates difference and preserves the agency of actors.