Computational Haze:A Mixed Method Meta-Analysis of Control, Absence, and Reorientation

by Zhenzhen Qi and Yang Wang


As cybernetics developed in the 20th century, it became apparent that mechanical systems could exert control over their agents, including humans. Foucault and Deleuze continually constructed and studied how modernity has transformed our social system into a more covert and potent controlling system. As a result, the mediation effect of computing-backed devices has permeated every aspect of modern life, resulting in the radical homogenization of both cultures and ideologies. This article examines the historical context of multiple-order cybernetics, Foucault’s theory of society and control, and the most recent control-assisting technology – machine learning – from a statistical and mathematical perspective. Then, the researchers introduce the concept of a Zone of Computational Instability, a potential method for intervening in a complex, multilayered cybernetics system and fostering openness. Under this framework, the phenomenon of modding in computer video games is examined. Frequently, modding positively affects and expands the original game, and in many instances, it dramatically increases a video game’s lifespan. The article then proposes bringing the practice of modding into physical reality, which shares similarities with a control-oriented, computationally simulated virtual world in gaming. 


computation, cybernetics, control, machine learning, modding

Predictive Computing

Homogenizing Culture

The 21st century has witnessed unprecedented growth in artificial intelligence (AI) research, particularly in machine learning, a subfield of Machine learning that has garnered increasing interest from both the technology and cultural industries. The global market demand for machine learning (ML) is expected to grow from $21.17 billion in 2022 to $209.91 billion by 2029.1 Rapid progress in computer vision and natural language processing enables us to enjoy the convenience of automation. However, in the past decades, research in critical computing culture has questioned the increasingly predatory behavior of machine learning, more specifically, how it influences cultural shifts through an increasingly ubiquitous yet withdrawn presence within everyday communication. For example, Anany and Crawford have questioned the limitations of the transparency ideal and its impact on the limitation of external governance of algorithmic platforms.2 Mackenzie questions how contemporary economic theories, based on mathematical models of markets, are not simply external analyses but intrinsic forces shaping economic processes.3 Eslami et al. describe how algorithmically filtered social media feeds, recommendations, and personalized search results shape users’ experiences without their awareness.4 Browne examines contemporary surveillance technologies and practices concerning the long history of racial formation and methods developed for policing black life under slavery, such as branding, runaway slave notices, and lantern laws.5

Within the visual culture industry, an example of predictive computing that emerged recently is the phenomenon of crapstraction. Jerry Saltz, the winner of the 2018 Pulitzer prize for criticism, highlighted a disheartening trend in the online art auction market. According to Saltz, artists increasingly turn against authenticity, mimicking a high-reward, low-risk, decorative art style that speculative art collectors and their friends trade online. Best-selling online artworks look disarmingly similar, all conveniently tailored for online auctions – “frictionless, made for trade. Art as bitcoin.”6

Fig 1, Top row, from left: All You Hear Is Beads Rattling (2012), by Leo Gabin; Untitled #0904 (2009), by John Bauer; Untitled (JS06198) (2006), by Josh Smith. Bottom row: ST-AA (Transfer Series) (2013), by Angel Otero; Big Squid Ink (2014), by Jamie Sneider; I (2011), by Rosy Keyser. 

The commodification of artistic creativity alone is not a new problem. In every period of art history, the market value of a particular type of visual style tends to be overmined at the expense of the others, such as Impressionism in the 1800s, Expressionism in the 1900s, Pop Art in the 1950s, and Minimalism in the 1960s, among others. However, according to artist and philosopher Hito Stereyl, what is new about Crapstraction is the emergence of online auction platforms and an invisible force of steering hidden behind the graphical user interface,  in the form of reductive logic transcoded into software code running on modern-day personal computers. 

Reductive Logic

Understanding a dataset often begins with grouping unlabeled raw data points, also known as clustering. Below is an illustration of one type of clustering analysis using the Gaussian algorithm. The raw data may have been gathered from user interactions with software or online sources such as e-commerce websites.

Fig 2. Contours of a standard Gaussian cluster density estimate. Author Unknown. Retrieved from

As part of the process, a few significant clusters are created, and data within proximity to each cluster are grouped. Then, a labeling algorithm is performed on each collection, selecting a singular point from each cluster. Finally, a complete data set is reduced to a few key points, as the rest are disregarded. 

Clustering analysis has a vast technical application, one of which is image compression. A digital image consists of 3 channels, RGB, each having values ranging from 0 to 255. Therefore, any digital image can have up to 255*255*255 different color fields. As an example, the image below contains 16,777,216 color fields. We aim to reduce its file size by representing it with only 16 color fields. With the smaller number of color clusters,  the image compression algorithm offers a higher compression ratio at the cost of image quality.7

Fig 3. From left, Compressed Image using K-means-clustering algorithm, with K set as 16 colors. Original Image containing 16,777,216 colors. Retrieved from

As personal computing becomes ubiquitous, the volume of personal data grows exponentially. In 2018, there were 2.5 quintillion bytes of data created each day. As a result, more sophisticated clustering algorithms are required to process complex datasets. Data clustering is just the first step in a more complex machine-learning process, allowing similar data to be processed more efficiently in the subsequent steps. 

On the other hand, an autoencoder is an artificial neural network that could be considered one of the more advanced clustering methods used in algorithms like facial recognition and machine translation. Deep autoencoder networks are used in unsupervised learning mode for dimensionality reduction. In the illustration below, when interacting with software, user faces are captured as a digital image and delimited with a bounding box(see input layer). Then, the input layer is compressed via a 3-layered Convolutional Neural Network(CNN). It performs hierarchical facial feature extraction with the help of an autoencoder(see hidden layers 1 and 2)  and then recomposes the faces together using an autoencoder(see hidden layer 3). By shrinking image input into simpler, more abstract vector forms, the algorithm minimizes the time needed for pattern matching. However, more granular levels of details are also filtered out. 

Fig 5. From left, Pursuit of a goal (one operator) and its localisation (two operators) without radar. Pursuit of a goal with radar and the localisation, the calculation of the goal´s flight line for predictions and their transfers to cannons. 

Culture Filtering

The cluster, label, and compression process have a wide application. Because of its exceptional data processing ability, clustering is widely used for pattern recognition, feature extraction, vector quantization, image segmentation, function approximation, and data mining, among others.8 In addition, the algorithm which makes content recommendations, namely the Recommender system, plays an essential role in social media platforms such as YouTube and Netflix.9 Clustering is often the preliminary step used by various Recommender systems to increase the accuracy of the prediction.10

In the case of image compression, it helps reduce the size of the images without visibly degrading the compositional quality of the picture, which is crucial in industries like healthcare, where enormous amounts of medical images need to be archived daily. However, when reductionist algorithms are integrated into the core logic of communication platforms that hold power to steer cultural trends and influence ideologies, problems emerge. Machine learning techniques arise primarily from statistical methods invented based on information theory. Data points that do not offer significant explanatory power to historical patterns, but might significantly influence the potentiality of the system, are deemed unessential and deleted. Therefore, we rely on a mode of forecasting that determines the future based on the past, trapping us inside the perpetual loop of reduction and abstraction. As internet activist and author Eli Pariser puts it, “But what is troubling about this shift toward personalization is that it is largely invisible to users and, as a result, out of our control. We are unaware that we see increasingly divergent images on the Internet. The Internet may know who we are, but we do not know whom it thinks we are or how it uses that information. Technology that gives us more control over our lives is taking control away.”11

From System To Monoculture

First and Second Order Cybernetics: Observed and Observer

The term system itself does not necessarily imply cultural erasure. It is a method of studying a dynamic collection of interdependent parts, members, or agents. One of the pioneers of general system theory, Ludwig Von Bertalanffy, used the term to describe various organismic scientific phenomena he observed as a biologist.12 The human body is one of the most ubiquitous biological systems we encounter daily. When we feel cold, our muscles shiver to generate heat and warm up our bodies. When we are hot, we sweat and evaporate heat from our bodies to cool down. Without paying attention, our bodies automatically maintain a standard temperature range to keep us comfortable. Since this type of system takes actions to cancel out excessive effects and bring back the current state to its norm, we refer to the canceling process as negative feedback. Besides natural science, negative feedback is widely adopted in engineering processes and machines. For example, there is a cruise system built into cars. It uses control actions to ensure a stable driving speed without delay or overshoot. 

More specifically, cybernetics is the science of exploring regulatory, purposive, and normalizing systems. During WWII, Norbert Wiener was hired by the British military to invent an anti-aircraft system that could help improve the possibilities for ground soldiers to predict the flight path of enemy flying objects. Wiener adopted a mathematical modeling technique called Stochastic Processes, where the errors made in prior predictions automatically became an input parameter of the next round of calculation.13

Fig 5. From left, the Pursuit of a goal (one operator) and its localization (two operators) without radar. Pursuit of a goal with radar and localization, the calculation of the goal´s flight line for predictions and their transfers to cannons. 

This self-referential circuit becomes crucial for the control of information. Inaccurate target tracking is eliminated through continuous, automated data smoothing processes, which will not stop until the overall systematic objective of shooting down enemy aircraft is reached. Norbert Wiener achieved for actions initiated by a human to be measured, analyzed, and mediated by a machine, as long as the goal can be precisely defined and measured from a machine’s perspective. 

“When I communicate with another person, I impart a message to him, and when he communicates back with me, he returns a related message which contains information primarily accessible to him and not to me…. When I give an order to a machine, the situation is not essentially different from that which arises when I give an order to a person. In other words, as far as my consciousness goes I am aware of the order that has gone out and of the signal of compliance that has come back. To me personally, the fact that the signal in its intermediate stages has gone through a machine rather than through a person is irrelevant and does not in any case greatly change my relation to the signal. Thus the theory of control in engineering, whether human or animal or mechanical, is a chapter in the theory of messages.”14

Wiener’s cybernetic theory is simultaneously considered a theory of communication as well as control. This framework achieves control through communication, due to two factors: the highest level of systematic goal is deterministic, and the negative feedback forming a closed loop or automatically self-correct until the systematic goal is reached. 

From the perspective of system complexity, Winer’s theory is considered a relatively simple system of control, or first-order cybernetics. In the case of his ballistic shooter invention, The purpose of the inner agent, the soldier operating the machine, is aligned with the system-level goal of tracking enemy targets. The system’s design is relatively transparent to its inner workings. The obscurity of cybernetics deepens as contemporary systems’ complexity grows. 

Gordon Pask, an English psychologist, also a collaborator of Winer, is historicaly credited with expanding cybernetics from first to second-order by further introducing the role of an observer. In 1968, he demonstrated this logic via an interactive installation called “Colloquy of Mobiles” in the Cybernetic Serendipity exhibition in London. Colloquy of Mobiles featured a triangular panel with obtuse-angled corners and three rotating fiberglass bodies with organic forms. Inside inorganic mobiles were attached to a rotating oblong element under the triangle. Pask called these “Males” and the fiberglass bodies “Females.” The “Males,” photocells, and light-sending elements were installed to illuminate “Female” mirrors. “Male” photocells detected light reflected by “Female” mirrors. Rotating elements memorized and learned from each other. In addition, visitors could also influence the stimulus-response system by using pocket torches to direct light to the “Females” mirrors, acting as artificial “Males.” Pask’s experimentation added a new layer of the observer, expanding Wiener’s cybernetics model from first to second order.

Compared to Winer’s wartime machinery, Pask’s installation replaced explicitly defined goals with an observing party whose intentions are opaque to the inner agents. In such a system, the observed could only attempt to grasp and describe what the observer appears to be doing, knowing well that every model is provisional and may have as much to say about the observer as it does about the observed. The identification of a border is consequently accompanied by susceptibility to the speculative creations of both the other and the self.

Fig 6.  From left, Gordon Pask, The Colloquy of Mobiles, 1969 / 2018, installation view of the exhibition “Neurones” at the Centre Pompidou, 2020, without audience interaction. By Paul Pangaro. Gordon Pask, The Colloquy of Mobiles, 1969 / 2018, installation view of the exhibition “BioMedia” at ZKM Karlsruhe, 2021 ByFelix Grünschloss for ZKM Center for Art and Media

The Illusion of Freedom In the Control Society

The implicit conflict between the observed and observer has long been the subject of historical examination. In 1975, French historian and philosopher Michael Foucault published Discipline and Punish, a genealogical study of imprisonment – an effective means of social normalization. According to Foucault, observation is an integral method of confinement. While more explicit forms of conformity, such as physical torture, respond to specific actions, observation responds to the lack of actions.15 The Panopticon design is conceptually significant because it enabled a kind of invisible gaze from nowhere – an artificially constructed, godlike omnipresence. Under this gaze, inmates do not know precisely when they are being observed, so they act as if they are being observed all the time, even when they are not. The result is seemingly self-motivated and self-regulated behaviors – the illusion of freedom. 

In his 1990 essay “Postscript on the Societies of Control” (“Post-scriptum sur les sociétés de contrôle”), Deleuze builds on Foucault’s notion of the society of discipline to argue that society is undergoing a shift in structure. Where societies of discipline were characterized by discrete physical enclosures (such as schools, factories, prisons, office buildings, etc.), institutions and technologies introduced since World War II have dissolved the boundaries between these enclosures. As a result, Individuals have become “dividuals,” masses, samples, data, markets, or “banks.”16 The mechanisms of modern societies of control are described as continuous, tracking individuals throughout their existence via transaction records, mobile location tracking, and other personally identifiable information. The informatic society is over-coded with an optimizing algorithm, fed by data collected directly from individual users.

Fig 7. Diagram illustrating higher order cybernetics with individual and capitalism as the lowest and highest ordered layer. 

Through an onion-skin structure, the recursive logic of second-order cybernetics is expanded into higher-order cybernetics, where complexity unravels to the degree that is incomprehensible by the human purview. By participating in digital communication, the observer of lower-degree cybernetics is transformed into the observer of higher-degree cybernetics. Unable to comprehend the observer’s intentions at the higher level, the observed party at the lower level is immersed in the illusion of freedom while, through endless layers of outward recursivity, being covertly subjected to the utmost highest-order objective – the capitalism project of profit maximization. 

Control in the Age of Computing

Lying beyond the surveillant architecture of the panopticon is a network that continually attempts to manipulate the environments in which individuals exist. This ensures risk management by focusing on governing networks, as opposed to network governance, exerting hegemonic forms of social authority in new ways. One key outcome is smoothing out outliers, which may form a terrain for resistance and rupture. 

Today’s communication system is faced with an imperative – the negotiation between functionality and openness.17 Digital communication is facilitated by computer software. Once designed, a computational device has a finite amount of total computational power that can only afford a finite number of signals to be transmitted in a given amount of time that is free from noise degradation. Therefore, we are faced with a dilemma, precision at the expense of variety. Machines with a high level of control of information over their surrounding environment offer low levels of choice and possibilities. Machines that offer variety and chance offer an ineffective rendering of the single immediate reality. If the history of information is seen as a process of balancing the informational and energetic outflow of the computational apparatuses, then this history has overwhelmingly favored efficiency and effectiveness at the expense of openness – a necessary condition for change. 

In mechanical systems, rules are implied by the relational design of the individual parts of the entire engine. In natural and social systems, rules emerge as a result of individual agents interacting with each other individually or collectively. For example, a computational system such as a videogame is composed of unit operations of electronic hardware and emergent cultural norms as a result of social interactions of human players. It has also been subjected to the Ghost in The Shell – a set of rules transcoded into computer algorithms, executed and reinforced by central processors of automata. 

According to media theorist Alexander Galloway, the protocol is how people are controlled from within a decentralized network. The protocol is a set of standards of practice that perpetuate control. There are three types of networks: centralized, decentralized, and airport system. to get to a more remote location, players must first fly to a bigger metropolitan airport and take a smaller flight. Distributed: all nodes carry the same weight. How federated networks (no moderation) function: take any technical protocol you use to access the internet, HTTP, for example, TCP, IP, DNS, promoted by institutions, become popularized. From this point onward, these proposals simultaneously represent the means to obtain necessary information about society and exposure to the network. In short, the price of not communicating equals complete elimination from the information society. Galloway proclaims it as the tragedy of interactivity – individuals must communicate whether they want to or not.18 Today, interactivity implies full participation and, as a result, global capture. The rise of digital capture has increased the significance of preventative measures such as blocking, obfuscating, concealing, suspending, and withholding.19

Beginning in the decade before World War II and accelerating through the contemporary Information Society, scientists have designed increasingly sophisticated mechanical, electrical, and algorithmic systems that acted as if they had a purpose. This turn toward “whole systems” led to an unexpected outcome in contemporary culture: stagnation. When a system is steered by a deterministic goal and reinforced by automated feedback, the potentiality of the outcome is reduced, and the entire system eventually loses its ability to adapt and evolve. In such scenarios, the individual agent can no longer learn and respond. In addition, not only is knowledge of the world knowledge increasingly removed through layered recursivity, but they are also no longer in touch with their aspirations. He is “opaque to himself, everything escapes him, as spontaneously co-operative, naturally empathetic, and fatally interdependent as he is.”20 The world is invisible to them, as much as they are completely exposed to it. 

Zone of Computational Instability(ZCI)

Where can one go once reductive patterns are assumed, established, and solidified in the geopolitical domain? The Man in the High Castle is a TV series based on Philip K. Dick’s 1962 novel of the same name. In the story, several travelers acquired the ability to temporarily travel between and affect parallel universes, either prompted by a rush of fear-based adrenaline or a brief burst of deep alpha-wave meditation. The book suggests that the fictional world, described twice, differently each time, could be a potential model of the world we are familiar with. 

In the television series of the novel, one of the travelers, Tagomi, was given the backstory as the Trade Minister for the Japanese Pacific States under the Kawasaki Government. Having witnessed the cruelty of war as a young air service pilot, he is motivated to prevent the deaths of more innocent lives. However, due to the Japanese code of honor, Tagomi could not speak out against the Japanese Government. Gradually, he turned to consult the I Ching, also known as The Book of Change, an ancient Chinese divination text that helped him sense the timing and forces of change. Through this ability, Tagomi gains access to the first concrete evidence of alternative possibilities – a token in the form of a recorded film strip that ultimately inspired more people to contemplate conditions for a different world. The key to Tagomi’s success lies in his ability to move freely between worlds, temporarily absent from the logic of any directly lived experiences of spacetime. However, one might ask, what is the comparable condition in the age of algorithmic surveillance, that would allow the observed to become unreadable to the observer? 

According to the Invisible Committee, when fleeing in the plain sight of the god’s eye of cybernetic control becomes increasingly challenging, we should focus instead on “increase[ing] the density of the communes, of circulation, and solidarities to the point that the territory itself becomes unreadable.”21 Modding, a slang expression derived from the English verb “to modify,” describes how this phenomenon is manifested within the realm of video games. More generally, the term modding refers to modifications of hardware, software, or anything else, to perform a function not initially intended by the original designer or to explore alternative effects or appearance. 

More specifically, modding in gaming refers to modifications by players or fans of a video game, mainly concerning how a game looks, functions, or structurally scaffolds. Often, this practice results in uninvited injections of structural change to a virtual world or environment via alternation of its algorithmic procedures. For example, the randomizer is a standard player mod, typical atop games of the 8-bit and 16-bit generations. It retains the primary form of gameplay but randomizes specific parameters to renew its challenge. Randomizers initially came out of the speedrunning community, a group of players who have exhausted the challenge of racing through the game. One early example emerged from The Legend of Zelda around 2015. Using statistical random distribution functions, the Zelda mod randomly moved the dungeons’ location and layout and the AI enemies’ location. Constantly reshuffling the parameters of critical in-game elements, this mod effectively created infinite subworlds for speedrunners and other players to explore.22

From Actions To Rules

Modding shares similarities with certain artistic gestures of transgressing boundaries of immediate reality but differs in one critical aspect. While artistic intervention frequently relies on a single action or a sequence of actions, practices inspired by ZCI instead focus on injecting or altering rules and conditions, which in turn enables new actions and behaviors. For example, in 2017, Chinese performing artist Ge Yulu surprised the art world in China. Utilizing the literal translation of “lu” as “road” in Chinese, he installed a road sign carrying his name on an anonymous Beijing street. Similarly, in 2009, german-based artist Aram Bartholl famously ran after a Google Streetview car in Borsigstrasse, Berin, which introduced a rare creative performance on the entire Google Map navigation service platform. Bounded by the enormous expense of altering physical realities and the difficulty of negotiating with city authorities, the experiments remained spontaneous gestures for both artists. 

However, in a game world, the front-end visual renderings of the streets and avatars are simulated by back-end rules. The modders do not alter individual player actions but rather conditions in the world environment that continue to invite new activities for all player-inhabitants. Even without being given access to the official software bundle released by the original game development companies, modders still manage to package these unofficial rules into mini-scale software packages or plugins and make them available for more players to download via open-source websites. For example, Civilization Five is a popular turn-based strategy series with an active modding community. Even a decade after its release, members have continued to create and upload new mods on the community website. While certain mods offer visual modifications like new skybox and player skins, others provide massive platform-wide changes. These mods can enhance the user interface and speed up aircraft acceleration, giving players an unparalleled advantage on the system level. One mod called Sukritact even changed in-game international diplomacy. It generates a graphic visualizer that allows players to see how every civilization in the world feels about other civilizations. It allows the player to strategize critical diplomatic moves, such as whom to invade versus forming an alliance. 

Chaos For Chaos Sake

In addition, it is also important to mention that modding differs from hacking, a hostile takeover implemented at the expense of other players. A world has its affordance – what price must each agent pay to enter and participate? The endurance of a world is correlated with “the degree to which its conditions of necessity (material and/or imaginary) compel, or more violently, force, members to affirm its configuration in practice, despite whatever dissenting attitudes may be held, such that worlds able to withstand and absorb such frictions are the most enduring.”23 In other words, a world may not last forever. Its vitality can be threatened by internal frictions, changes in rules, values, resource scarcity, and more. 

2b2t’s downfall demonstrates a virtual world coming to its end, unable to withstand the endless waves of chaos introduced by instability. Founded in December 2010, 2b2t (2builders2tools) is one of the oldest and vilest Minecraft servers alive. Since its inception, the server has hosted over 603,000 players over its procedurally generated map, creating over 9.5 terabytes of player-generated data. It is also one of the longest-running anarchy servers in the massively multiplayer game engine. As the only Minecraft server with no rule, authority, or reset since its inception, it is a perfect case study documenting player-initiated behaviors motivated by nihilistic intentions. On an anarchy server, novice players, also called “newfags” by seasoned players, cannot rely on the administrator’s help. Instead, they must gather food and resources while fending off well-armed, long-term players and be ready to die countless times before making it outside the central spawning zone, which lays barren after over a decade of bombardment. Experienced players reside in the far corners of the spawn area, well-stocked with armories and consumables. As a result, the map is less destroyed further away from spawn, allowing trees and animals to preserve. The server has no etiquette regarding ownership – any objects or structures can be destroyed at any time if found by other players. In 2018, two players discovered a technical loophole. They programmed an algorithm that simulated players traveling across the Minecraft terrain at hyper frequency. In order to avoid system overload, the Minecraft server responded by double-checking to verify that the signal came from an actual player instead of the malware. However, this fix accidentally exposed the coordinates of actual player hideouts to hackers. Using this information, they bombard structures years in the making with explosives in the blink of an eye. 

Fig 8. Copper132. A render of 2b2t shows the destruction caused by 11 years of no rules on a Minecraft server. March 4, 2022.

Novice players of 2b2t have compared life inhabiting the server to descending into a drug cartel hideout or news reporters traversing war zones. Nevertheless, the rare times players go out of their way to bestow gifts and welcome others can seem almost shockingly heartwarming.24 2b2t is a manifestation of the State of Nature project by 17th-century philosopher Thomas Hobbes. According to Hobbes, in the state of nature, an anarchist state, everyone is subjected to disrupted work and essential belongings being destroyed by others. As a result, human energy is primarily exhausted on a basic level of self-preservation. No time can be devoted to value creation beyond the basic survival instinct. 


In contrast to hacking, modding is a reorientation of game mechanics with the ultimate aim being fostering diversity and multiplicity. Chaos is not the end goal but rather an instrument for achieving long-term adaptation and evolution. Cooperative play, also known as co-op, is a modification of a game’s single-player mode that encourages multiple players to work together as teammates, either against non-player characters(NPC)AIs or other teams of real players.

For instance, the Co-op mode in Elden Ring allows players to enter another player’s world and assist them in defeating a dwarf dungeon or boss by using a summoning token or a summoning pool. When one kills, one gains runes; if one dies, one loses nothing. Therefore, players can study bosses before facing them in their own worlds. They can safely assist other players in completing battles by learning to anticipate the boss’s attacks. It is a fun and low-risk method for learning the behaviors of demanding boss AIs. Many players also use this opportunity to test out extreme tactics that are too risky to try out in their game worlds, such as stripping off all armor and weapons to reduce one’s defense in exchange for a maximized level of dexterity and agility. Multiple players make manipulating the boss’s attention easier, which gains time to heal wounded players or interrupt AI’s attacks. Besides the gold summon symbol, which begins cooperative play, the game offers a red summon, which allows players to challenge each other with goodwill and cardinality. Challenging other players is a great way to learn about different builds and reflect on one’s tactics. Among the weapons, spells, and items in Elden Ring, most players will only try a small subset of them in the game. Playing against other players is a great way to try out completely different weapons and stat choices more suitable for duels than AI bosses. For example, Sleep is a status effect in Elden Ring that puts enemies to sleep, giving players a chance to inflict a critical attack. When enemies are inflicted with sleep, they are stunned, allowing players a chance to heal or prepare for a heavy attack. However, many bosses and enemies in the game are resistant or immune to sleep. Therefore, a better alternative is to use a sleep build for duels against other players. To summarize, the co-op mod allows players to self-organize whom to play with and against, effectively allowing them to design their own games within the original game. 

Nevertheless, co-op is not the only force of reorientation in the game world. As cybernetic communication systems, games are also subjected to the formidable forces of feedback loops. For example, in the multiplayer game Super Mario Kart, the game world drops random gifts from the sky, seemingly introducing uncertainty. However, the parameters do not seem completely random. To keep players engaged for a more extended period, players ahead of the crowd are more likely to receive trolling gifts, such as slippery banana skins that temporarily tamper with the driver’s control. By comparison, players who fall behind are more likely to receive boosting gifts such as flying wings, which helps racers to bypass obstacles and catch up. However, due to the close-loop determinacy of the cybernetic control mechanism, it is not hard for seasoned players to notice the pattern – randomizing gifts tend to average out player performances.

On the other hand, unlike positive and negative feedback, Co-op allows players to self-organize into almost infinite subworlds based on multitudes of playstyles and interests. As a result, modifications can have a much longer and diversified impact. For example, the 1998 first-person shooter game half-life has one of the most vibrant mod communities in gaming history. Over the years, popular mods of the original game have evolved into entirely new games, such as Counter-Strike, Cry of Fear, Black Mesa, Day of Defeat, Dear Esther, Dino D-Day, Nuclear Dawn, and The Stanley Parable, among others. These new games vastly expand on, deviate from, or even contradict the original game’s concept, narrative, and rules of play, fostering a community of players and cultures that’s unique from the other spin-off games. 

Computational Haze

Boris Vian, a French poet, once envisioned the impact of a natural fog on existing relationships in his short story “Love Is Blind.” The city’s residents awaken one morning to find themselves surrounded by a “tidal wave of obscurity” that gradually changes their behavior.

In the video game universe, the observer sits in the higher and deeper order of its logic, armed with cybernetics, and demands autonomy alone. In response, players created the Zone of Computational Instability by injecting a stream of computational haze as a brief layer of independence for everyone, which was dispersed via an algorithm on the open-source platform. Parallels to our immediate environment, video game worlds are extensively mediated by computationality and hypercharged by interfaces and algorithms. With the discovery of the Zone of Computational Instability comes the task of identifying the material that fills the zone, the computational haze, which is actuated as a flux of injected code fragments, third-party textures, and unusual protocols. Computation haze comprises procedures that inoculate noise, hide certainty, and increase unpredictability. The machinery is reactivated, and potential is restored through the fog. The revised conventional perceptual coordinates are disrupted, blurring the boundary between what is seen and what is not. 


[1] Fortune Business Insights. “Machine Learning Market Size, Share, Growth & Trends [2029],” March 2022.
[2] Ananny, Mike, and Kate Crawford. “Seeing without Knowing: Limitations of the Transparency Ideal and Its Application to Algorithmic Accountability.” New Media & Society 20, no. 3 (December 13, 2016): 973–89.
[3] MacKenzie, Donald. An Engine, Not a Camera: How Financial Models Shape Markets. MIT Press, 2008.
[4] Eslami, Motahhare, Aimee Rickman, Kristen Vaccaro, Amirhossein Aleyasen, Andy Vuong, Karrie Karahalios, Kevin Hamilton, and Christian Sandvig. “‘I Always Assumed That I Wasn’t Really That Close to [Her].’” In Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems. New York, NY, USA: ACM, 2015.
[5] Browne, Simone. Dark Matters: On the Surveillance of Blackness. Duke University Press, 2015.
[6] Saltz, Jerry. “Zombies on the Walls: Why Does So Much New Abstraction Look the Same?” Vulture, June 17, 2014.
[7] “Exploring Clustering Algorithms: Explanation and Use Cases,” September 17, 2021.
[8] Du, K.-L. “Clustering: A Neural Network Approach.” Neural Networks 23, no. 1 (January 2010): 89–107.
[9] Ricci, Francesco, Lior Rokach, and Bracha Shapira. Recommender Systems Handbook. Springer, 2015.
[10] Beregovskaya, Irina, and Mikhail Koroteev. “Review of Clustering-Based Recommender Systems.”, September 27, 2021.
[11] Pariser, Eli. The Filter Bubble: How the New Personalized Web Is Changing What We Read and How We Think. Penguin, 2012.
[12] Bertalanffy, Ludwig Von. General System Theory: Foundations, Development, Applications. George Braziller, 2015.
[13] Galison, Peter. “The Ontology of the Enemy: Norbert Wiener and the Cybernetic Vision.” Critical Inquiry 21, no. 1 (October 1994): 228–66.
[14] Wiener, Norbert. The Human Use Of Human Beings: Cybernetics And Society. Da Capo Press, 1988.
[15] Foucault, Michel. Discipline and Punish: The Birth of the Prison. Vintage, 1995.
[16] Deleuze, Gilles. “Postscript on the Societies of Control*.” In Surveillance, Crime and Social Control, 35–39. Routledge, 2017.
[17] Qi, Zhenzhen, and Yang Wang. “Computational Haze.”, April 14, 2020.
[18] Galloway, Alexander. Uncomputable: Play and Politics in the Long Digital Age. Verso Books, 2021.
[19] Brunton, Finn, and Helen Nissenbaum. Obfuscation: A User’s Guide for Privacy and Protest. MIT Press, 2016.
[20] Tiqqun. The Cybernetic Hypothesis. MIT Press, 2020.
21, Comite invisible, and The Invisible Committee. The Coming Insurrection. Semiotext(e), 2009.
[22] Harris, John. “What Happens When You Randomize The Legend Of Zelda.” Kotaku, February 12, 2016.
[23] Reed, Patricia. “The End of a World and Its Pedagogies .” Making and Breaking. Accessed August 27, 2022.
[24] FitMC. “The Fall of Minecraft’s 2b2t.” Video. YouTube, July 24, 2021.