2 ✕ 1

Pick Up That Can:
Control and Resistance in the Virtual World of Half-Life 2

Point Insertion

After the opening credits of Valve’s 2004 video game Half-Life 2, the player is left viewing a train from a first-person perspective. After the train drops them off at Point Insertion station, the player is able to navigate the station freely, through the fences of a security checkpoint, through its holding cells, and eventually into a wide corridor that leads onto the main station hall. The gate in this corridor is blocked by a figure clad in a uniform resembling a police uniform, and made anonymous by a gas mask. His mechanised voice urges you to “pick up that can,” as he strikes an empty soda can off of a rubbish bin with his baton (image 1). This is the first instance in the game the player receives a mandate to interact with an object. After you have operated the correct controls to interact with the object, you are told “now put it in the trash can.” You now have a number of options to choose from: if the can is picked up and deposited in the bin, the agent will step aside to allow you into the train station, and the players’ console profile will subsequently receive the “Submissive achievement.”1 If the can is thrown at the agent, however, the agent will directly strike the players’ avatar with the baton (image 4), until they have either deposited the can into the bin, or the avatar ‘dies’, meaning the screen blacks out and the player is returned to the previous saving point.2

Image 1: Still from Video 1, Point Insertion. 00:09

Image 2: Still from Video 1, Point Insertion. 00:15

I will consider in which ways this particular scene depicts certain aspects of Deleuze’s description of the transformation of disciplinary societies into control societies. With reference to Deleuze’s writing on new technology, I will look at how players’ experiences of this virtual action and the production of ‘machinima’ might occupy certain planes of resistance.

In the 1992 essay Postscript on Control Societies, Deleuze writes, “disciplinary societies […] operate by major sites of confinement. Individuals are always going from one site to another, each with its own laws…”3 In Half-Life 2, game progress is rendered as a virtual space through which the player navigates. The developers at Valve refer to these virtual divisions as ‘gates’, and subdivides them into three categories: soft, hard, and story gates. A soft gate is defined as “a simple task that a player must overcome in order to continue forward in the experience.”4 Hard gates are more accurately described as combat areas, where one or multiple non-player characters (NPCs) must be eliminated to break through the gate; and story gates require an action outside of the player to occur, often triggered by the player’s in-game navigation. We can read the division of this virtual layout into these ‘gated’ spaces as a variation of the “sites of confinement”: each space you enter consists of a puzzle which requires different solutions, according to different sets of rules, and executed by different types of agents.5 In this sense, Half-Life 2 resembles a disciplinary society where discipline can be navigated from confinement to confinement, on a single level.

Image 3: Still from Video 1, Point Insertion. 00:18

Image 4: Still from Video 1, Point Insertion. 00:19

Formally, Half-Life 2 distinguishes itself from other titles in the genre by its lack of cut scenes, levels, and objectives. From the moment the player sets out from within the train, design mechanics allow for the camera and avatar to be controlled from the players’ console for the entire duration of the game—without interruption. In terms of player experience, this mechanic is supposed to augment immersion into the virtual realism of the Half-Life 2 universe; In terms of narrative, the lack of instruction means that the understanding a player needs to navigate this environment world must come from within this world itself. We can identify the can as a didactic object; it is didactic firstly because it teaches the player how to physically act, in order to interact in the virtual environment; secondly, it establishes a clear position toward the authority that asserts control over this fictional world. The design mechanics here establish an immersion that ascribes agency to the virtual environment. In order to understand how this process is reflective of control societies, it is necessary to define virtual objects and actions ontologically. In Philip Brey’s 2014 essay “The Physical and Social Reality of Virtual Worlds”, virtual objects are defined as such:

A virtual object is a digital object that is represented by a computer, and that can be interacted with or used through a computer interface. Virtual objects are digital objects that appear to us as physical objects and that we interact with in a manner similar to physical objects.6

Here, the physicality of virtual objects already complicates the objects’ agency: a virtual object is a digital object, meaning the object is physically equal to its code, yet completely dependant on hardware to be perceptible. Although represented by and interacted with through a computer screen, Brey adds that these objects “seem to have no physical existence,” and are therefore susceptible to an ontological confusion as to what their agency as objects could be at all. This confusion is addressed later on in the text with reference to John Searle’s categorisation of objects-in-reality as found in his 1995 book The Construction of Social Reality:

The distinction between physical, ordinary social, and institutional reality corresponds in large part with the previously made distinction between simulation and ontological reproduction in virtual environments. Physical reality and ordinary social reality can usually only be simulated in virtual environments, whereas institutional reality can in large part be ontologically reproduced in virtual environments. For example, rocks and trees (physical objects) and screwdrivers and chairs (ordinary social objects) can only be simulated in virtual reality… their simulations are not capable of reproducing the actual physical capabilities of physical and ordinary social objects.

So the can is not a real can; for the simulation fails to provide a physical simulation. The agent is not a real agent, because the physical attributes cannot be reproduced on a flat screen; but the agency he exercises falls into the third category of objects, the institutional objects:

… money and private property can literally exist in virtual reality. This is possible because institutional entities are ontologically constituted through the assignment of a status function, of the form “X counts as Y (in context C).”7

Status function is assigned through collective intentionality: if we’re all immersed in the belief that police can hold a certain type of agency to exercise a certain type of power, that power can be reproduced. (X) here is the agent to (Y), the function of policing. The difference between the agents’ status as institutional object, and representation thereof, here, is exactly the context (C): because the context of the agent in the virtual realm precisely doesn’t ontologically recreate but only simulates, the status function assigned to policing becomes obsolete. We can therefore speak of a possibility of control, but are left with a mere simulation of control that is not ontologically reproduced.

So in what sense can we consider the scene in the station corridor as representative of control societies? Deleuze writes:

In control societies… the key thing is no longer a signature or number but a code: codes are passwords, whereas disciplinary societies are ruled (when it comes to integration or resistance) by precepts. The digital language of control is made up of codes indicating whether access to some information should be allowed or denied.8

When we take into consideration the context of this particular fictional universe, we might find another way in which this scene might engender a control society. The main antagonist in Half-Life 2 is designated the Combine, described as a “collective of synthetic, alien, and human elements that have taken authoritarian control over earth.”9 It is made clear that the agents of the Combine are in constant communication with each other, through a narrative device aptly named the Overwatch, which constantly discloses mission objectives, locations and events to the agents in its network. Characteristics of control societies are those found where, Deleuze notes: “We’re no longer dealing with a duality of mass and individual. Individuals become ‘dividuals,’ and masses become samples, data, markets, or ‘banks.’”10 The apparently transhuman Combine cannot be separated into clearly designated units anymore; thus, although they might take the appearance of Foucault’s prison guards, they seem to be on a threshold between being representative of units of discipline and Deleuze’s concept of the dividuals that make up information banks, under constant surveillance, constantly feeding back new information. There also exists such a multiplicity outside of the games’ fiction—the player. Although there might be one player progressing through the game, this information is undisclosed to the code that constitutes events in this virtual reality. The player is not represented as an individual but as an avatar—a dividual entity that similarly functions as a bank in which information about game progress is stored.


Video 2 presents an alternative way of navigating into Point Insertion station. When the corridor is entered, the avatar is already in possession of a crowbar, visible on screen. Upon approaching the agent, it appears as if the soda can has been replaced by a large jerry can. Yet the agent still acts according to the script; again urging you to “pick up that can.” Suddenly, the crowbar is replaced by a bunch of bananas, which are cast towards the agent, followed by another bunch, and another, all bunches piling up in front of and around the agent. When as much as fifty bunches of bananas have been rejected from the avatar, the avatar moves back into the furthest corner of the corridor. A trigger is set off, causing the bananas to eject from their peel, after which an enormous, multicoloured explosion takes place, shaking the image vigorously for about eight seconds. The corridor is lit on fire. As the fire dies down, the avatar approaches the gate again. The agent and the bananas have vanished into charred smudges, and red stains.

Image 5: Still from Video 2, Banana Edition, 00:14

Image 6: Still from Video 2, Banana Edition, 00:33

This video, which is accessible online, is a piece of ‘machinima’. In the 2004 book 3D Game-based Filmmaking: the Art of Machinima, which coins the term, writer and filmmaker Paul Marino defines machinima as, “animated filmmaking within a real-time virtual 3D environment”:11 machinima consist of the creation of a fixed animation or event within an (already existing) 3D rendering of a space. In his writing on the politics of technology in Deleuze, William Bogard notes:

Networks are not uniform or homogeneous. Control, as well as resistance to control, functions in a multiplicity of ways on a multiplicity of planes. […] The use of networks in art, network practices like file-sharing, hacking, encryption, the use of proxies for anonymity, podcasting, denial of service attacks, open-source application development, along with more traditional methods like refusal to use networks, unplugging, and so on—there are many ways to resist information control.12

Through looking at the machinima video ‘Pick Up That Can! Banana Edition!’ I will argue that we can add machinima to the list of network practices in art that can resist information control. Deleuze already writes about possible dangers for information technologies, “control societies function with […] information technology and computers, where the passive danger is noise and the active, piracy and viral contamination.”13 If machinima is a type of viral contamination, it occupies some planes of resistance. Through considering the network that creates machinima as well as the materiality of machinima production, I will try and demarcate these planes of resistance.

Garry’s Mod, developed independent of Valve by Garry Newman, was introduced as a ‘sandbox’ modification to Half-Life 2 in 2004. It uses the open-source code of the Source game engine that supports the game,14 and thus enables its users to add new characters, weapons or other objects, and effects to the virtual game environment. Although it was later introduced as a stand-alone game, it has no formal objectives or levels, but functions as an open world for users to build and experiment with the physical possibilities of the game engine.15 The codes that this modification produces can then be reintroduced in the game, and the user—who is now a player-turned-author—can alter the virtual environment after their desires. When this process is performed and recorded, the resulting product is machinima. In their essay “Participatory Fan Culture and Half-Life 2 Machinima”, Schott and Yeatman write:

Machinima practices arise from and exist within a game culture propagated by the hybridization of players and authors and in which the act of playing itself constitutes “presence” or virtual embodiment within a gamescape, thus granting players the capacity to effect change by activating, producing, and reworking assemblages…15

The production of machinima requires a network of actors—in this case, the hybridisation of authorship produces an assemblage of codes and coded information. Bogard claims that “in control societies, the form of content, the machinic form, is the distributed network… Distributed networks deterritorialise the disciplinary assemblage.”16 The author-machine displaces what was previously identified as a disciplinary assemblage. What was scripted to be a ‘soft gate’ has transmuted into a ‘hard gate’, or a combat area; the task that the player was scripted to carry out has been resisted to the point of lethal virtual violence. The difference in rules in these spaces of confinement, then, do not really matter, as the mandates supplied by the script can be rejected by a transmutation in the code, over and over again.

It is also important here to make the distinction between the modification of the code and machinima. We can identify the modification of the code as a modulation that opens up different lines of flight: Garry’s Mod opens up an infinite range of altering the objects, environments and narratives that we are presented with in the game. Deleuze argues that disciplinary societies “mold individuals,” while control societies “modulate dividuals.” A mold is a confinement, whereas a modulation would be more appropriately described as figure in a constant state of variation. Bogard likens the concept of modulation to a performance: “modulation is […] like editing the parameters of a piece of music live, as it plays, or as it is being recorded (for example, varying its tempo, velocity, attack, reverberation, and so on). Modulation is not intended to produce an individual, which would be equivalent to making a fixed recording. Instead, the format itself is opened to variation in real time.”17 Machinima, however, is a fixed performance, and not one open to constant modulation:18 once the machinima is recorded, regardless of the endless possible fictions that have been introduced by the modification of coding, it is temporarily fixed. Machinima is also fixed spatially: both the virtual space where the performance takes place and the space in which it is viewed can not be altered. The locus of publication, such as the Banana Edition video on Youtube, fixes the possibility of transfiguring the practice as a live configuration. In ‘Half-Life 2 Machinima’, the practice is described as, “a cartographic practice, that is, mapping of particular actions that trace out particular desires, including those of narrative, technical prowess, and expressions of pleasure.”20 But this mapping practice operates spatially as a reterritorialization of control onto the virtual world; at the same moment that it rejects a script, it does so by creating a new script, for it doesn’t allow itself to be operational in a modulation of space or time.

A lengthy definition of resistance, based on Deleuze’s concept of control societies, is recorded by Bogard:

Resistance is not just anti-control. Networks have liberatory and democratic potentials that function immanently within control... The point is that resistance, though common, is always specific and immanent within a concrete assemblage. Just as there is no universal form of control, there is no universal mode of resistance to control, only experimentation with the abstract machine and the possibilities networks create for us to have positive encounters with the outside, with the common—encounters that are joyful, create solidarity, abandon hierarchies, and denounce power, that generate lines of flight and multiply connections beyond what the network can dominate.21

When Banana Edition is read through this account, we might identify it as a form of resistance that operates as such: from within various control networks, such as the Combine or the locus on-screen, code modification can offer a real encounter that produces humor and skews the representation of power relations—yet the way it is reproduced (as machinima) is still set within a concrete assemblage, such as the virtual realm of Half-Life 2, or the console hardware.

There are various different ways in which the opening scenes of Half-Life 2 embody a particular kind of transformation between disciplinary and control societies. The single level of different types of confinements, or ‘gates’, may be read as a signifier of a disciplinary society. The simulation of control that the agent represents is not an ontological reproduction but a mere simulation. The dispersive nature of the physicality of virtual objects is a line of flight into transmutation, as the virtual object is itself already a network: It cannot be reduced to a code, or a physical piece of hardware, but does equally not exist only as an on-screen representation. The Combine embodies the dividuals and data banks that are characteristic of Deleuze’s control societies, and furthermore, the virtual world also renders the player a dividual, as the avatar that represents the player merely holds the code for player progress, and has no individual properties.

The modification of game codes can open up lines of flight to resist the way control is rendered to operate. Resistance can never be total because control can never be total. Machinima is a recording of the performance of resistance created by a distributed network. However, the machinima considered here (Video 2: Banana Edition) acts more as a reterritorialisation of control, as it’s presented as a spatially and temporally fixed entity. The resistance offered up here takes place in a concrete assemblage: however the avatar passes the gate, and however that gate is defined, we will arrive on the other side, in the concrete assemblage we’ve come to know as the station hall of Point Insertion.


1. If the can is thrown at the figure first, the player makes it past the gates, they will receive the “Defiant Achievement”; both achievements are rewarded with the same amount of points. See: wikiHow, 2015 “How to Unlock the Submissive Achievement in Half-Life 2”, wikiHow: to do anything. Available at: [Accessed 8 January 2015].

2.The events described here are possible ways to navigate the game, as exemplified by various walkthrough videos and reviews referenced throughout the text.

3. Deleuze, G., 1992, “Postscript on the Societies of Control”. In: Deleuze, G., 1995, Negotiations: 1972 – 1990, trans. Joughlin, M. New York: Columbia University Press, pp. 177.

4. Mark Sivak deducts this from the developers’ commentary on Lost Coast, an additional level to Half-Life 2, which was released in 2005. As referenced in: Sivak, M., 2009, “Half-Life 2: Being Gordon Freeman”, ETC Press. Carnegie Mellon University, 3 October 2009. Available at: [Accessed 8 January 2015].

5. The Half-Life 2 wikia lists all the types of agents and their specific tasks; examples can be found at Half-Life 2 Wiki, Civil Protection. Available at: [Accessed 8th January 2015].

6. Brey, P., “The Physical and Social Reality of Virtual Worlds” In: Grimshaw, M., ed., 2014, The Oxford Handbook of Virtuality. Oxford: Oxford University Press, pp. 44.

7. Ibid., pp. 47–48.

8. Deleuze, G., 1992, “Postscript on the Societies of Control”, p. 178.

9. Sivak, M., 2009, “Half-Life 2: Being Gordon Freeman”, ETC Press

10. Deleuze, G., “Postscript on the Societies of Control”, p. 180.

11. Marino, P., 2004, 3D Game-based Filmmaking: the Art of Machinima. Phoenix: Paraglyph Press, p. 1.

12. Bogard, W., “Deleuze and Machines: A Politics of Technology?”. In: Poster, M., and Savat, D., eds., 2009, Deleuze and New Technology. Edinborough: Edinborough University Press, p. 28.

13. Deleuze, G., 1992, “Postscript on the Societies of Control”, p. 179.

14. Schott, G., and Yeatman, B. “Participatory Fan Culture and Half-Life 2 Machinima: A Dialogue Among Ethnography, Culture, and Space”. In: Lofield, H., and Nitsche, M., eds., 2011, The Machinima Reader. Cambridge, Massachusetts: The MIT Press, p. 305.

15. Ibid., pp. 305, 310 & 312.

16. Ibid., pp. 302–303.

17. Bogard, W., “Deleuze and Machines”, pp. 18–19.

18. Ibid., p. 22.

19. Schott, G., and Yeatman, B. ‘Participatory Fan Culture and Half-Life 2 Machinima’, p. 308.

20. Ibid., pp. 308–309.

21. Bogard, W., “Deleuze and Machines”, p. 28.


Deleuze, G., 1995, ‘The Actual and the Virtual’. In: Dialogues II, 2002, trans. Labert, E. R. New York: Columbia University Press.

Deleuze, G., 1992, ‘Postscript on the Societies of Control’. In: Deleuze, G., 1995, Negotiations: 1972 – 1990, trans. Joughlin, M. New York: Columbia University Press, pp. 177 – 182.

Grimshaw, M., ed., 2014, The Oxford Handbook of Virtuality. Oxford: Oxford University Press

MacMillan, F., 2010, ‘Analysis: On Half-Life 2 – Build It and They Will Come’, Gamasutra, 19th March 2010 [Accessed 8th January 2015].

Marino, P., 2004, 3D Game-based Filmmaking: the Art of Machinima. Phoenix: Paraglyph Press.

Nazaroff, N., 2011, ‘Pick Up That Can: Human Head’s Norm Nazaroff on the problems with interactive objects’, Gamesindustry.biz, 6th April 2011. [Accessed 8th January 2015].

Schott, G., and Yeatman, B. ‘Participatory Fan Culture and Half-Life 2 Machinima: A Dialogue Among Ethnography, Culture, and Space’. In: Lofield, H., and Nitsche, M., eds., 2011, The Machinima Reader. Cambridge, Massachusetts: The MIT Press, pp. 301 – 313.

Searle, J., 1995, The Construction of Social Reality. New York: Simon and Schuster.

Shapiro, J., 2012, ‘“PICK UP THAT CAN!” Storytelling in Half-Life 2’, Gamasutra, 25th October 2012. [Accessed 8th January 2015].

Sivak, M., 2009, ‘Half-Life 2: Being Gordon Freeman’, ETC Press. Carnegie Mellon University, 3rd October 2009. [Accessed 8th January 2015].

Valve, 2004, Half-Life 2, [video game], PC / Mac, Playstation® 2, Xbox 360, Xbox. Bellevue, WA: Valve Corporation.

Compression Augmentation