**Links to**: [[Subject]], [[Xpectator]], [[Agency]], [[Model]], [[Abstraction]], [[Self]], [[Subject]], [[Behavior]], [[Freedom]], [[Volition]], [[Decision]], [[Freedom]], [[Orientation]], [[10 Bias, or Falling into Place]]. %% edit section below on assembly theory Use this, big time (to argue for and against it, compare to Levin’s definition of agency) https://www.mdpi.com/1099-4300/24/4/472 %% _N.B.: The term agent is heavily used throughout this project, but this can sometimes be better termed and understood as a [[Xpectator]]._ An agent, in this context, is not something that should be mostly associated with _agency_ (which we consider an often too liberal, voluntarist concept), but rather as a thing which _acts_, is the preferred word in this research for designating a **locus of interest in tracing the dynamics of individuating phenomena in/such as complex systems**. The description of whatever we intend to be an agent implies the methodology in its appraisal as something capable of bringing about change. The agent is thus whatever we desire remains “the same” in light of change(s it brings about). It is a way of _pointing_ at things (see also: [[E Pointing]] and [[Vantagepointillism]]). In this way, an agent can be something like a cell; or even a structure on a cell membrane; an atom; a multicellular organism; a concept; an ideology. The level of description is what matters: insofar as any of these things can be modelled as catalysts of further phenomena, they are _agential_. When we define an atom as a catalytic bringer-about of changes, the changes become the background effects of the atom. At another level, atoms may be the background effects of other things we define as agents, such as cells or ideologies. It all depends on the level of analysis, avoiding a so-called “flat ontology” where everything is an agent. (See also: [[Equivalence and difference]] and [[Homotopy Type Theory]]). It can be compared to but differs from “actor/actant” (in actor-network theory or ANT). It is comparably relational, constructive and material-semiotic, but does not want the baggage of ANT. In Michael Levin’s work we obtain a definition of _agential material_ as that for which “... optimal engineering protocols require behavior shaping (not just micromanagement) because the components are not passive but have some degree of autonomous behavior, preferred states, and decision-making capacity in some problem space.” (Levin’s definition in Blackiston, Levin et al. 2023). Here it becomes clear how the level of description implies the methodology: for the bioconstructive purposes of Levin (et al.), we observe agency in things that have volitional directionality such as preferred states, coupled to their observable degrees of freedom (capacity for motion, lifespan, etc.). Were we to define a _concept_ as an agent, we could say that it is agential as a “fractal” and largely indeterminate structure which couples things such as thinking agents to language, and language to certain ‘degrees of freedom’ in worldly structures. This happens when we say “this is a pencil”—the whole sentence is a concept, or each of the words, and they are all fractal in how they couple myriads of underdetermined (e.g., the category of _being_ travels in many, very interesting dimensions), memorious-forgetful instances to graspable objects and beyond. Closer to Levin’s concept of the (spatiotemporal) lightcone; we prefer thus to define agency as that which we observe as dominating and/or having a perspective (a lightcone) from _our_ perspective. Similarly, in assembly theory, objects such as complex life molecules are conceptualized “not as point particles, but as entities defined by their possible formation histories” (Sharma et al. 2013). In this case the agential methodology ensuing from this is the measurability of their _assembly index_, for **specific purposes** (see also: [[Assembly and assemblage]]). More thoughts on this continue in [[Scale-free dynamics]].