February 19th, 2011

S.C. Hickman

On Agency: A Response to Robert Jackson and David Roden

"Discreteness – unless I’ve read him wrong – doesn’t seem to entail complete ontological separate between things. It only entails that things have discrete and independent dispositions by virtue of their causal powers – including their computational powers."
     - David Roden

"I’m not even sure myself how far discreteness goes in the ontological separation of things (hence a work in progress). One can easily find agency and discreteness in Latour and Whitehead for example, but these entities are nothing but relations, by the nature that they are only real if they perturb or translate. ...The issue of course is how do you account for that agency?"
     - Robert Jackson

I've been thinking about a conversation between both
David Roden and Robert Jackson on Agency of late. When it comes down to it Graham Harman's position is that an Object is both a discrete and autonomous agent that can withdraw from all relations at any time, anywhere; that is, it is not dependent or co-dependent upon any relation to other objects ever, although it can entertain relations with other objects if it decides to do so: therein lies the crux of the issue. This makes an Object seem like a Sovereign power that is concerned with two types of relations: those that relate to its own internal self-relation, and those that determine its acts as an agent in relation to all those other objects that it might come into contact with outside of this self-relation. [1] 

Agency implies both decision and authority: an agent acts on its own internal authority and decides pro or con, for or against certain actions based upon that internal authority. Typically autonomy is thought of as an idea in which the self or object is defined by its relations to others, but in Harman's position an object is defined by non-relation or withdrawness. This is the first rule. Yet, it does have relations to its own internal sensual machinic experience, therefore it can be either awake or asleep (dormant). Here we speak of the real as opposed to the sensual object and its relations. Harman has iterated repeatedly that we can never have direct contact with the real object but only indirect contact through its sensual notes or appendages. 

 The contingency of objects is based upon the push and pull or the tension between dueling powers, which can be less than, greater than, or equal too an object's authority, force, or buffering. This is where, I think, Harman's idea of objects being in a duel intervenes. I'll have to take that up another time. At the moment I am concerned with Jackson's point that "objects are clearly contingent on many factors to ontologically exist, both externally and internally. But if you reduce an object down to ‘just’ those contingent factors, it does not seem to resolve the crux." The crux that he is speaking of is "how do you account for that agency?":

"For an object to act, and furthermore act on other objects something needs to be expressed that was not expressed before. Endorsing a total relational system risks deferring agency to the pre-individual, or the whole before the parts strategy. ... The private, withdrawn object manoeuvre deals with this by arguing that a real object is always more ontologically speaking – a black hole of sheer formal agency, untouched and undisturbed."

Expression implies the power of an Object to exert its own force against the resistance of other objects, whether those objects be internal or external to its own real internal relations (i.e., sensual objects). In art I was reminded of the Impressionist and Expressionist movements. While Impressionism was all about the effect of the subject on the viewer, Expressionism was the artist expressing his or her own interior. In this sense Agency implies either acting on or being acted on by an Object.

I found it fascinating that Robert Jackson in his
post brings up the decisional power of an autonomous agent and the implications for social constructivism. He brings in Levi, Badiou, Agamben, Zizek, and Ranciere who all in one form or another accept the idea that a subject-as-object is always already "governed by relational networks of power", and because of this the implications are clear for Jackson who states that "agency becomes a real problem, discrete or not discrete."  Of course Harman would be the odd man out on this implication, since for him human or object agency is situated outside of any relational network; or, at least the real object as agent is always withdrawn, whether its sensual appendages are in relation or not.

It is at this point in Jackson's argument that he situates his conception of algorithm and agency. First an explanation of just what an algorithm is (from

"In mathematics and computer science, an algorithm is an effective method expressed as a finite list of well-defined instructions for calculating a function. Algorithms are used for calculation, data processing, and automated reasoning. ... Starting from an initial state and initial input (perhaps null), the instructions describe a computation that, when executed, will proceed through a finite number of well-defined successive states, eventually producing "output" and terminating at a final ending state. The transition from one state to the next is not necessarily deterministic; some algorithms, known as randomized algorithms, incorporate random input."

As Jackson implies an algorithm may or may not execute correctly (i.e., "The cake may fail to rise in the oven or even burn..."). The term that comes up throughout this essay of Jackson's is the relation of an object or agent to contingency or its dependence on contingent factors, and as he said in another section of his essay:

"...the object is one of three things; wholly contingent on it’s relations (systems), somewhat or partly contingent on them (my position I think) or not contingent on them whatsoever (Graham’s position I think). There could be a case for a ‘slight’ undermining of objects – perhaps being ‘partly’ contingent on relations is Graham and Levi’s position, not sure on this point."

He then decides if an algorithm is "contingent on certain factors" then the issues is twofold:

1. Can it be said that an algorithm has agency?
2. Does it possess something inherent in its form that is inaccessible to its contingent relations?

His answer it "yes, but it is a limited yes. A paradoxical tension between it’s contingent vicissitudes and it’s vigorous execution." Harman in Prince of Networks: Bruno Latour and metaphysics speaks of what he terms the "object-oriented model of time", which for him implies the "tension between an intentional [sensual] object and its shifting, accidental manifestations (PoN, p. 218)." Not sure if Harman was thinking of algorithms or not when he made this startling confession: "Try as we might, we cannot adequately translate the subterranean execution [my italics] of things into any discursive list of tangible properties; such attempts always fall somewhat short of the thing itself (PoN, p. 218)." What is implied here is the tension between a thing and its qualities, the point being as Harman states it, - situating his answer within the paradox of philosophy itself, using a passage from Plato's Meno in which Socrates says: 'If I do not know what something is, how could I know what qualities it possesses?'(PoN, p. 218). The paradox is that things can be known "only by way of listing their qualities", but ultimately as Socrates implies and Harman agrees "the thing itself must be known quite apart from those qualities" (PoN, p. 218).

Most of this comes back to Harman's concepts of vicarious causation in which two real or two sensual objects can never touch. The other side of the tale is his use of asymmetry to show us just how objects can in fact touch each other in a place where "real objects meet sensual ones"(PoN, p. 221). The paradox of the tension resides between two states buffered or alluring in which an object is either split from its own qualities or not.

Jackson leaves his essay with a final question: "If an algorithmic calculation has a bug that causes it to malfunction, and it is modified so as to achieve the end result and execute properly, is it a different algorithm?" At the heart of this question is another paradox: Who does the modifying?  Is it an external power, another object interfering with the internal operations of algorithmic agency or object? Or is it the internal processes of a real object working on and correcting the relations of its own sensual or intentional objects? If the former, then the power of the external object or agent must be greater than the force of the real object to overcome its resistance or buffering, thereby allowing the greater force of the external object as agent to enter into a third object interacting in that volcanic core where the sensual dynamics produces change. If the latter, the object itself is awakened from its own dormant state and acts upon its own sensual objects to affect change. Either way change is effect of an affective relation based upon the mechanism implied by Harman's dynamics. One might say that an agent exercises authority over what she does only once she is faced with a set of possible actions, whose possibility reflects their compatibility with the causal or vicarious power of her or another's desires: in predicting that she will perform one of these actions, she authorizes this action, and thereby strengthens her motives for performing it.   

1. Read both posts for David Roden (
here) and Robert Jackson (here)