Against Holism: The Argument from Entropy

In the last couple years I’ve spilled a lot of ink arguing against holism or the thesis that everything is related to everything else.  This is not because I believe that relations are unimportant– indeed, the only thing that really interests me are relations –but rather because I think this position says too much.  From a holistic perspective, for example, I’m unable to make sense of ecology.  While many ecotheorists defend holistic ontologies, the actual practices of ecologists and environmental scientists reveals something different.  In their practices, they show supreme attentiveness to the fragility of relations, and focus on what happens when elements are subtracted from ecological networks (e.g., the extinction of a species) or added to a network (e.g., increased carbon dioxide in the atmosphere).  This wouldn’t be possible unless entities possessed some minimal autonomy from their relations.  So while I can agree with the holist when they argue that it is important to attend to interactions and interdependencies between entities, I don’t think holism can be coherently sustained as an ontological position.  We can call this the “argument from ecology”.

A second argument could be called the “argument from politics”.  Here the argument is similar, perhaps identical; which shouldn’t come as a surprise as societies are ecologies.  Here the argument runs that if social and political change is to be possible, then holism cannot be true.  If everything were related to everything else in holistic wholes, then social and political changes wouldn’t be possible because individual entities– whether they be classes, identity groups, or whatever –would be overmined (Harman’s nice term) insofar as they would have no being beyond their status as elements within a whole.  Insofar as they would have no being beyond their status as elements in a whole, there would be no way of departing from existing social regimes so as to form new sets of social relations.  Put differently, there would be possibility of emancipation, as the concept of emancipation is only possible where there is something to be emancipated and where there is something from which that thing can be emancipated.  This is only possible where relations are separable and don’t constitute the essence of a being.  Social changes are only possible where elements constitutive of a whole– if wholes even exist –are irreducible to those wholes.

read on!

As an aside, it’s worth noting that this is no different than what Badiou says in his set theoretical ontology.  When Badiou defends the axiom of extension, he is defending the thesis that sets are identical to their members.  Given a set {a,b,c}, this set is identical to a set {c,a,b} because both sets contain exactly the same members.  This is equivalent to saying that it is the elements that determine the being of the set, not the relations between those elements (a point that more dialectically– relationally –inclined Marxists should pause and think about before enlisting Badiou in support of their positions).  What Badiou is, in effect, saying is that there is no natural or intrinsic ordering of social relations.  There is no great chain of being (right) that orders how people ought to be related, nor ought marriage “naturally” be “between a man and a woman”, nor ought whites hold dominion over slaves because people of other races are allegedly “like children” needing a “father” to look after them, nor do people necessarily need kings, parties, or authorities to look over them.

Rather, Badiou argues that societies are just sets that take on a variety of configurations and where, above all, relations among elements or members can be changed.  There’s a reason that Badiou calls his politics a politics of subtraction, an operation that wouldn’t be possible were relations intrinsic and holistic.  In making this claim, I take it that Badiou is just presenting a formal articulation of every genuinely revolutionary political orientation against every reactionary, holistic politics (Edmund Burke and his heirs):  relations are extrinsic, external, and separable, such that there is no natural or divine order to the form social relations ought to take.  Social relations can be transformed and reconfigured precisely because the being of beings is not identical to “being-related”.  Things relate, sure, but their being is not exhausted– nor is it reducible to –by these relations.  Harman nicely makes these points about Edmund Burke in a number of places.  People seem to confuse the thesis that things often exist in assemblages or networks of relations and it is important to recognize this, with the thesis that therefore the being of things is their relations to other things.  Given that social and political change has, in fact, taken place throughout history, we can conclude that ontological holism is mistaken.

A third argument occurred to me as I was composing my talk for the Networked Humanities conference at University of Kentucky.  We can call this the “argument from entropy”.  Before proceeding, it’s important to note that entropy has two different (though related) significations depending on whether we’re talking about thermodynamics in physics or information theory.  Generally people are only familiar with the first of these significations.  In thermodynamics, entropy signifies the tendency of closed systems to lose energy available for work over time.  In carrying out operations, some energy is always dissipated into the environment or lost, such that it is no longer available for work in the sense that physics uses the term.

As an aside, I find it amazing that concepts of work and energy are almost entirely absent in the humanities (and generally in the social sciences as well).  It’s as if we believed that being is composed solely of the discursive and things, and are then left– when raising social and political questions –left to ask whether it is the discursive that holds social relations together or things.  We forget that holding anything together requires work.  When I propose the concept of “thermopolitics” (I don’t know whether anyone else has used this term), I’m suggesting that we need to attend to work and energy as additional mechanisms of power that lead people to tolerate oppressive assemblages (Reich’s, Deleuze & Guattari’s question).  It’s not simply because people are duped or stupid that they tolerate oppressive assemblages that act against their own interests, but also because they are depedent on certain assemblages for the calories they need to sustain their bodies, as well as the fuels they need to maintain their homes, transportation, and so on.  A similar point can be made about time.  When I propose the term “chronopolitics” (and again, I’m not familiar with how others use the term), I’m referring to the manner in which the sheer structuration of time for people, groups, and institutions can become so overwhelming that they’re left with no additional time to try and change their circumstances.  Chronopolitics would be the analysis of technologies of time as mechanisms of power that 1) exhaust people physically and mentally, and 2) so fill the day that other forms of engagement are foreclosed.

Anyway, back to entropy.  In information theory, “entropy” refers to a measure of probability with respect to information.  A high entropy system is a system in which there is an equal probability that any element will be related to any other element.  In other words, in a high entropy system it’s impossible to make inferences from one element to other elements because there’s an equal likelihood that any of the other elements will be related to this element.  By contrast, a low entropy system is a system in which it is possible to make inferences from a given element to another element.  A system in which only one relation to another element is possible would be a system with 0 or nil entropy.

In this framework, it’s possible to see why I make the claim that questions of interpretation and social and political thought are questions of entropy.  When we talk about texts that can be interpreted– and here I’m not talking about meaning –and when we talk about social orders, we’re talking about low entropy systems.  In the case of interpretation, for example, we’re saying that given this element, we can infer a relation to this other element.  This is above all the case with the hermeneutics of suspicion or “depth readings” in the Freudian and Nietzschean tradition, as well as in ideology critique.  In these contexts we’re saying that there’s an improbable relations between two elements that repeats in a variety of different actions, thoughts, or social domains.  In saying this, we’re saying that there’s an order here (order is always more improbable than disorder).  It is this improbability that supports the veracity of the depth interpretation.  Likewise, when we say that there’s a social order, we’re saying that given a particular element in a social system– say a person of a particular race –we can confidently infer their relation to other things:  their economic status, range of possible occupations, and so on.  We’re rejecting the claim that the person’s relations are stochastic or highly entropic.

This allows us to give a precise definition of power:  power is the mechanisms by which a society reduces entropy.  Order never comes for free, but always requires operations, energy, and work precisely because it’s more probable for anything to be related to anything else, than for anything to maintain improbable and selective relations to a delimited range of elements in the order of being.  “Onto-cartography” would be the investigation of the mechanisms by which improbable orders are maintained; and would therefore include investigations of discursive mechanisms (not surprisingly, the favorite of the humanities), chronopolotical mechanisms, geographical mechanisms (geopolitics), and thermopolitical mechanisms.  Whenever encountering an order in the social and political world we should be surprised and ask ourselves how it maintains itself.  In understanding how it maintains itself, we can begin to devise strategies undermine it where those negentropic mechanisms are oppressive.

All of this aside, we now see a third argument against holism.  A world in which everything is related to everything else is identical to the definition of a highly entropic system.  It is a world where no inferences can be made to other elements precisely because there’s an equal probability of anything being related to anything else.  Such a world would be a world of Brownian motion, where there was no language, mathematics, ecologies, or social orders, precisely because all of these orders are orders in which relations between elements are selective or improbable.  As Deleuze said of Hegel’s categories, the world proposed by holists is a world that is too baggy, too ill fitted, to get at the real of the world that we, in fact, encounter.