Exe0.1 David Gauthier

From Software Studies
Jump to navigation Jump to search

Politics of the "Cold Gaze"'

In order to start thinking about “execution”–and more precisely machinic execution–I would like to contrast two notions of computing from the start: the notion of computing as structure and the notion of computing as machine. This distinction, albeit a scaffolded distinction which certainly has shortcomings, will help me survey the field of seminal discourses in Media / Software Studies and help devise what I call the politics of the “cold gaze,” which foregrounds tensions inherent to scholarly camps theorising computation as structure versus those theorising computation as machinic.

The notion of computing as structure implies well-defined type definitions, inputs and outputs, routines and subroutines, as well as feed-back loops that speak to a concept of totalisation that is deeply grounded in the principle of the eternal return, that is, the repetition of the same or the equivalent. It puts forth a type of formalism or “imperial” order (symbolic, linguistic, mathematical or otherwise) that subordinates inner technical modulations and variations to the resemblance of their produced effects. Because a structure forecloses the alterity of its constitutive elements in order to function as a coherent whole, structures are thus possessed with a desire of fixed eternity translatable across sites.

On the contrary, the notion of machine is haunted by the desire of failure, error, catastrophe and death. It unfolds through a heterogenous and metastable assembling of matter and symbols, where the angst of undecidability looms large. Central to the notion of machine is thus its relation to alterity, that is, the recurrent causal relationship it establishes between the heterogenous elements it holds together (mathematical symbols, geological elements, bio signals, electrical pulses, etc.). A machine, thus, rather than foreclosing its constitutive elements, opens up, conjugates and articulates them as part of a contingent diagrammatic process or, in other words, a metastable event.

These two notions of computing—structure and machine—can be understood, I suggest, as two oblique interpretative apparatuses that put forward disparate views on what computation is and/or does. Neither of these notions can be said to harbour a more “human” or “non-human” bias than the other, as both are equally suited for the analysis and interpretation of technical media and its effects. Moreover, I suggest that both can engage with technical media in proffering a respective type of “cold gaze” (Ernst 2013: 25). If theory is to be understood from its etymological roots as theoría, which means at once contemplation and speculation, then the “cold gaze,” in being a mode of contemplation and speculation of the machine itself–its own peculiar worldview–, demands us to substitute the warm human-centred gaze of theories such as phenomenology and/or historiography with the cold one of the machine; here “the camera eye displaces subjective vision, which allows for a sharpened reading” (Ernst 2013: 45) of technical operations. The “cold gaze” allows theory to address and indeed observe the inner operativeness of media, and thus our concern of “execution,” in all its non-human, non-phenomenological, non-discursive and artificial dimension.

Some have argued that this “cold gaze,” promoted by a certain style of media archeology, is a type of “distancing” from the “messy world of political and social issues,” one that further amplifies the divide between the humanities and the sciences (Parikka, 2013: 67). I would like to differ from this perspective and, in taking up the call of this workshop, put forward the idea that the problematic of “execution” as it relates to the infrastructural operations of media “asks a question to which the political interpretation is the only coherent answer” (Galloway, 2006: 329) and even further posits that “desire is part of the infrastructure” (Deleuze and Guattari 2004: 104). To be clear at this point, “execution” is to be understood non-metaphorically as the digital computational process by which (source) code is compiled and executed on a given hardware site with its own peculiar architecture. What I propose to problematise is that both notions of structure and machine, in providing an analytical frame to theorise what and where “execution” is and/or does, produce contrasting regimes of visibility and invisibility, or, to put it bluntly, two diverging worldviews. These regimes, I shall argue, speak to their respective politics of what really matters in technical media when confronted with questions of operation and interpretation, and thus meaning.

Central to the concern at hand is the question and notion of infrastructure. To illustrate this problematic I will focus on the rich and illuminating debate about source code and ideology that took place a few years ago between Galloway (2006) and Chun (2005, 2008, 2011). This debate was partly prompted by the nascent field of Software Studies which elected Software as the prime object of study of new media objects (Manovich, 2001: 48). Through her seminal articles, Chun warns us that in divorcing software from hardware and in focusing on its discursive and linguistic aspects, one makes a political move since “software perpetuates certain notions of seeing as knowing […], creating an invisible system of visibility. The knowledge software offers is as obfuscatory as it is revealing” (2005: 27).

In her articles, Chun highlights how the advent of Computer Science, with its emphasis on symbolic programming languages, drastically changed the ways in which computing has been conceived from the 1950s onwards. Programming and coding practices, prior the advent of computing languages per se, were an affairs of “crafty” and often gendered local conventions and customs that were highly tailored for individual machines across various sites (Nofre et al., 2014: 49). With the growing commercialisation of computing machinery, the concept of programming language came about as a means to standardise these local conventions and customs, encapsulating them into a linguistic form that would present traits of both mathematical notations and natural language:

The notion of a programming language, which is connected to the idea of universality, became central to this exercise of boundary work that sought to disengage the activity of programming from local conventions, and to transform it into a transcendent and universal body of knowledge. From this endeavor, programming languages and algorithms emerged as epistemic objects stripped of any marks that would associate them with specific hardware. (ibid.: 66)

Consequently, through the advent of “universal” languages, not only does programming acquire a type of “machine independence,” (source code able to be built and executed on a variety of machines) but more importantly, linguistic objects produced with a given “universal” programming language acquire, much like mathematical notions or even poetry, an epistemic and discursive life of their own. Programming languages could thus carve out their own computing invariant–a transcendent “island of semantic stability” (ibid.)–by rendering invisible the machine that was once literally in plain sight. It is clear, then, that the universalisation of programming produced a kind of stratification of computing that cuts off the tacit and innate relationship programming had, and indeed still has, with the material, processual and “crafty” aspects of hardware which, consequently, becomes an invisible “black box” (Brown and Carr: 89 via Nofre et al., 2014: 54).

Speaking of this disjunction between the visible symbolic programming language and the invisible “black box,” Chun posits that, as a result, “software is a functional analog to ideology” (Chun, 2005: 43). This analogy between software as an object in itself and ideology stems from the fact that software instantiates a strict division and upholds an illusory dialectical logic of cause and effects (input and output) between infra-structure–obscure “black box”–and super-structure–manifest programming languages. Inevitably, then, questions of operations and meaning are foreclosed by the linguistic regime of software alone in that it is the only regime capable of lending itself to “objective” interpretations, and thus legitimatising itself in its own right and asserting itself as the dominant interpretative regime. Chun concludes by noting that “because of the histories and gazes [it] erases; and because of the future [it] points toward [, ] software has become a commonsense shorthand for culture and hardware a shorthand for nature” (Chun, 2005: 46).

To grasp the potency of Chun’s warning, let’s turn to Galloway’s intervention and show how his framings, according to Chun, further highlight the illusory conflation of code (software) and execution (hardware). In his article entitled “Language Wants To Be Overlooked” (2006) Galloway acknowledges that code necessitates a hardware infrastructure in order to function; he writes, “code exists first and foremost as commands issued to a machine. Code essentially has no other reason for being than instructing some machine how to act” (ibid.: 326). We can clearly see that Galloway sustains this split between infra-structure–the material hardware machine–and a super-structure–code as linguistic commands issued to control the machine–when he famously declares that “code is the only language that is executable” (ibid: 325). The paramount problem with this conception of command and control, instruction and execution, code and machine is that, as Chun rightly puts it, “[in making] the argument that code is automatically executable, the process of execution itself must not only be erased, but source code also must be conflated with its executable version” (2008: 305).

This erasure of execution, by conflating linguistic commands and machine operations, has the corollary of reducing notions of contingent computing events and processes solely to written instructions, or, in other words, conflating logos with action (ibid.: 303). I would like to emphasise this point, that is, the non-processual or eventless notion of execution, as it seems to be symptomatic in both Galloway’s and Manovich’s conception of transcoding. For Manovich, “to “transcode” something is to translate it into another format” (2001: 47). Similarly, for Galloway, software is a prime exemplar of “technical transcoding without figuration” (2008: 319) where the various layers composing the subsystems of the machine (code, logic gates, etc.) are put into into a relation of pure equivalence. As Galloway notes, “one of the outcomes of this perspective is that each layer is technologically related, if not entirely equivalent, to all the other layers” (ibid: 327). We clearly see that for both theorists, translation and transcoding are conflated, thereby completely bracketing the process by which the machine codes and decodes while privileging the outcome of this process, that is, the resulting format; Galloway writes, “there is a privileged moment in which the written becomes purely machinic and back again,” as if the process of transcoding is a pure “technical” formality that can be over-looked (Galloway 2008: 319). Guattari, in addressing this foreclosure of structure through this logic of equivalence, writes bluntly that “Essentially, no break is any longer accepted. That the structures have no specific identifying marks means that they become 'translatable' into one another, thus develop­ing a kind of indefinite logical continuum that is peculiarly satisfying to obsessionals” (1984: 117). As I wrote earlier, both Manovich’s and Galloway’s accounts privilege an imperial order that subordinates and forecloses metastable technical processes and events to the resemblance of their produced effects, effectively obliterating all the various and precisely non-equivalent processes, technics, components and tactics that are conjugated together as execution.

What I have tried to expose here, through the influential work of Chun, is that the notion of computing as what I’ve colloquially called structure leads to an impasse when confronted with the question of execution. In privileging the “imperial” order of (semantic) code–as super-structure–over the obfuscated order of the “black box”–as infra-structure–one cannot conceive of “execution” in its own right since code, in this ideological setting, is the only omniscient and omnipotent sovereign. But how can we get out of this impasse? Could a critique of ideology untangle this mess? I do not believe so. Rather, following Deleuze and Guattari, I think it is more productive to say: “there is no ideology, it is an illusion” (Guattari, 2009: 38). Does this mean then that we need to follow Kittler’s mantra and declare that “there is no software” (Kittler, 1995)? I do not believe so either. What is needed is to confront the desire of upholding a strict and illusory division between super- and infra- structures that is reinforced by the distinction between soft- and hard- ware. The question of execution demands problematising the advent of a machinic event unfolding within a specific site. The challenge, then, is to think of and account for computational events in and for themselves by exposing the various material and symbolic processes involved in their production. The question of execution can then move from being one of programmatic structure to become one of catalytic event conditioning.

In a sense, this move from structure to event is similar to what Niels Bohr required of quantum physics with his notion of the Copenhagen Interpretation (Bohr 1937; 1950) which stipulates that abstract physical concepts, such as energy, momentum, velocity, etc., ought to be framed as events unfolding within experimental physical arrangements, rather than being understood as direct attributes of things (Barad 2003: 814). Historically and indeed politically, Bohr stood against the idea that instrumentation can be disregarded as an after-thought of the “imperial” order of mathematical formalism, and asked that measuring devices themselves be incorporated in the interpretation of experimental results, as they are responsible for the very existence of a given measured entity. Just as a physical particle traverses and physically couples with a given experimental apparatus, so does the thread of execution of a program that traverses the various software strata and silicon wafers it encounters along its path. As media theorist Wolfgang Ernst notes, “material archaeological strata and the symbolical order of the archive are progressively being conceived as essentially processual by nature, to be deciphered as operative diagrams” (Ernst, 2013: 17 my emphasis). What I want to highlight here is that the notion of machinic execution calls for a type of theoretical and methodological apparatus that can account for the multiple heterogeneous physical and symbolic manifestations and temporal processes that are articulated and conjugated together as to produce execution proper. The notion of diagram, which was pioneered by Pierce and later reworked by Deleuze and Guattari, is highly potent in that regard as it presents “the articulations whereby a system operates” (Guattari, 1984: 170) and, at the same time, can also work to expose, as an oscilloscope, the intensive time-based physical signals of a given machinic execution (Ernst, 2013: 178). I am not suggesting here that, much like the earlier discussion on code, diagrams are to be conflated with machinic operations (Chun, 2008: 307), but rather that diagrams are to be produced from execution itself as a diagnostic of its non-discursive dynamics.

As a concluding remark, I would like to point out a possible direction in carrying out a type of operational diagrammatics of execution by namely looking at the object and the advent of computational “errors.” Constructing and developing software and hardware involves a “crafty” relation with matter and logic in taming the antagonistic causes and effects of entropy, noise, bugs, transients, undecidability, errors and the like. For this reason, advanced equipment, methods and protocols for the “clinical” diagnosis of software and hardware, such as debuggers, logic analysers and in-circuit emulators, have been devised to observe, trace and monitor the micro operations and failures of a given computing machine. In bringing into focus this debugging equipment as objects of study and in framing questions of technical operations as questions of diagnosis and prognosis, I suggest that we might be able, then, to devise proper methods and related theoretical apparatuses that can indeed directly engage with the nuanced problematic of execution.

What I have gestured towards here theoretically and methodologically with the notions of cold gazing, machine, metastable events, diagrams and debugging equipment is to move from theories of technical media based on critique to ones that are based on clinique. Part of this pursuit is to debunk the totalising drive of a certain type of critique tradition, which I believe goes hand-in-hand with the totalising drive of ideology, so it can be grasped as producing relaying pieces as part of a more extensive programme. Critique is not to be execrated, but neither should it be valorised for itself. From this perspective, critique, apprehended as a component rather than a whole, is nothing but a movement towards clinique.


Barad, Karen. 2003. “Posthumanist Performativity: Toward an Understanding of How Matter Comes to Matter.” Signs: Journal of Women in Culture and Society 28 (3): 801–31. doi:10.1086/345321.

Bohr, Niels. 1937. “Causality and Complementarity.” Philosophy of Science 4 (3): 289. doi:10.1086/286465.

_________. 1950. “On the Notions of Causality and Complementarity.” Science, New Series, 111 (2873):51–54.

Chun, Wendy Hui Kyong. 2011. “Crisis, Crisis, Crisis, or Sovereignty and Networks.” Theory, Culture & Society 28, no. 6: 91–112.

___________________. 2005. “On Software, or the Persistence of Visual Knowledge.” Grey Room 18: 26–51.

___________________. 2008. “On ‘Sourcery,’ or Code as Fetish.” Configurations 16, no. 3: 299–324.

Ernst, Wolfgang. 2013. Digital Memory and the Archive. Edited by Jussi Parikka. Electronic Mediations, vol. 39. Minneapolis: University of Minnesota Press.

Galloway, Alexander R. 2006. “Language Wants To Be Overlooked: On Software and Ideology.” Journal of Visual Culture 5, no. 3: 315–31.

Guattari, Félix. 1984. Molecular Revolution: Psychiatry and Politics. Peregrine Books. Harmondsworth, Middlesex, England ; New York, N.Y., U.S.A: Penguin.

____________. 2009. Chaosophy: Texts and Interviews 1972-1977. Edited by Sylvère Lotringer. Semiotext(e) Foreign Agents Series. Los Angeles, CA: Semiotext(e).

Kittler, Friedrich. 1995. “There Is No Software.” http://www.ctheory.net/articles.aspx?id=74.

Manovich, Lev. 2001. The Language of New Media. Cambridge, Mass: MIT Press.

Nofre, David, Mark Priestley, and Gerard Alberts. 2014. “When Technology Became Language: The Origins of the Linguistic Conception of Computer Programming, 1950–1960.” Technology and Culture 55 (1): 40–75.

Parikka, Jussi. 2011. “Operative Media Archaeology: Wolfgang Ernst’s Materialist Media Diagrammatics.” Theory, Culture & Society 28 (5): 52–74.