acm-header
Sign In

Communications of the ACM

Opinion

The Infrapolitics of Algorithmic Resistance


red wagon filled with cellphones

Credit: Simon Weckert

There is growing concern about the power algorithms play in contemporary societies, which has led to debates about how to deal with these artifacts of power.5 Regulation, governance, transparency, and ethical AI have become buzzwords in the agenda about possible restraints on the algorithmic power. Less attention has been paid, however, to everyday forms of resistance enacted by lay citizens. These forms of resistance are often seen as banal, rudimentary, and void of political strategy to face the overwhelming and ubiquitous nature of this power, but they are the way citizens often choose as a form of friction to algorithms.

Back to Top

The Power of Algorithms and the Infrapolitics of Everyday Resistance

It should be clear, from the start, that the power of algorithms is complex. It derives from their capacity to affect, in different ways, individual behaviors, shaping broader collective outcomes, in often opaque ways.1 Algorithms are somehow in the background of arenas of interaction, shaping the contexts of action in tacit ways. An important dimension of the power of algorithms is of particular interest here: It is clear that defining individuals accurately and situating them within clusters is central to the way the power of algorithms is enacted. This need to know/define individuals represents a specific type of computational mediation pervading humans' endless processes of becoming. Algorithms tend to foster forms of identification that paradoxically crystallize and reinforce who (they assume) we are.

uf1.jpg
Figure. A wagon with 99 cellphones was pulled through Berlin streets by performance artist Simon Weckert to trick Google's algorithms to create a virtual traffic jam.

If the existing scholarship has been fruitful in showing the political impacts of algorithms embedded in different data-driven technologies, there is a gap regarding the ways in which humans resist in everyday life to the power of algorithms.12 Here, we are particularly interested in the forms of resistance that are part of what James Scott calls the infrapolitics.10 Scott argues that the public demonstration of opposition is a historical exception. Most often, it is too dangerous to express dissent. Analyzing contexts of slavery and colonialism, Scott identifies hidden transcripts used by the oppressed, such as rumors, gossip, chants, jokes, gestures, sabotage and particular languages. Despite its concealed nature, Scott notices that infrapolitics provides grounds for the more visible forms of politics. It would be, hence, misleading to underestimate the power of these tactical forms of resistance throughout the history of humanity. Infrapolitics has shown to be particularly relevant in contexts of deep asymmetries, carving the possibilities for more open opportunities of dissent while hindering the way structures of power operate.

Michel de Certeau has also devoted attention to the arts of resistance that pervade everyday life.4 De Certeau argues that everyday practices—such as cooking, walking, and reading—are full of creative potentials. Individuals reinvent the structures and materials with which they interact when they use them and make them their own.

These ideas are useful if we conceive of many of the practices used to confront the power of algorithms. Algorithms operate in the strategic dimension discussed by de Certeau. They frame contexts of interaction between humans and machines and between humans and humans, setting the boundaries of agency in many different contexts of life. In the infrapolitics dimension, however, the digital world has witnessed the emergence of resistance against this power of algorithms. To reflect de Certeau's example, we can highlight how the algorithms of platforms such as Google Maps or Waze play a key role in constructing the city, when they organize its flows and structures. Individuals may, however, use these structures in innovative and creative ways. They often choose different paths or ignore recommendations. They make mistakes that may confuse algorithms. They make forbidden moves that create nonexistent shortcuts and may prefer busier streets to see the façade of a nice building.


Individuals may also purposefully play with algorithms.


Individuals may also purposefully play with algorithms. A famous example on this respect is the performance of the artist Simon Weckert, who pulled a cart with 99 smartphones in the streets of Berlin to deceive Google's algorithms, pretending there was a traffic jam.a Weckert's performance played with the algorithms to shed light on how they work. In Indonesia, drivers of app-based transport services have developed resistance tactics to improve their working conditions against rules and control enforced by algorithmic managers. In their social networks drivers share information about little hacks to make daily work a bit easier, creating a form of everyday resistance against algorithms.6

Back to Top

The Infrapolitics of Disidentification

Individuals play with algorithms daily.5 This is part of the hidden transcripts that constitute the infrapolitics of ordinary life. Among the many possibilities to game algorithms in the infrapolitics of ordinary resistances, we argue the tactics of disidentification play a key role. The notion of disidentification is at the heart of Jacques Ranciére's political framework.9

For Ranciére, politics emerges from the attempts to promote the principle of equality, by showing the existence of a damage in the way societies distribute their shares. Politics exists in contrast to the police, which is not limited to the institutions of coercion but encompasses the broader conception of order and the organization of the political community. Politics generates ruptures in these established orders. It emerges through the attempt to reconfigure the way societies organize and distribute their shares. Politics involves a process of transformation in which their parts become something else. It is not only a struggle between pre-established identities, but, instead, a transformative process that is linked to the transformation of these identities. It is in this context that Ranciére advances his notion of disidentification. Politics is grounded on the idea that the subjective transformations involve a form of self-displacement in which social actors refuse the places previously reserved to them.

Our argument here is that a relevant form of resistance to the power of algorithms emerges from attempts to deidentify. By negating the naturalness of a place attributed to oneself, one may resist the power of algorithms. Disidentification emerges as a powerful weapon of the weak. Disidentification is embedded in many practices of gaming, through which individuals deceive algorithms to escape from their continuous attempts to establish a clear definition of them. We claim there are different types of practices in the repertoire of disidentification in the face of algorithms.

Anonymization is the first and most frequent tactic adopted by individuals. To remain partially anonymous, some individuals use browsers in incognito mode and use the Dark Web to escape constraints imposed by platform algorithms. Just like de Certeau's urban wanderer, individuals may use alternative online routes to create their ordinary forms of resistance.

Another central tactic of contention involves deceiving algorithms. This is clear, for instance, when individuals choose to spell words differently to avoid automated responses that can range from censorship to unpleasant responses. Writing with dots between letters or with numbers in the middle of some words is a frequent strategy to remain below the radar. Nardine Alnemr notices this strategy in the attempt of Palestinians to circumvent censorship. The use of the watermelon emoji instead of a Palestinian flag was used to avoid surveillance.2 To bypass content moderation algorithms of social media platforms, people are adapting their language and creating new vocabulary to communicate in TikTok, YouTube, Instagram, Facebook, and Twitter. For example, to avoid punishment by moderation algorithms, some sex workers refer to themselves on TikTok as "accountants."7 To resist against profiling algorithms, some young people resist personalization by making use of scripts to replace old posts in social media with meaningless lines of code.

Deception is also clear when individuals pretend to have liked something or intentionally go after something they dislike, just to confuse algorithms about one's real interests and tastes.5 Sometimes individuals know how an algorithm operates and react intentionally by altering the streams of information inputs to alter the streams of outputs. In China, for instance, gig workers often play with algorithms either to increase their rates or to augment the price paid for deliveries.3


A relevant form of resistance to the power of algorithms emerges from attempts to deidentify.


Lastly, some forms of resistance hinder the collection of biometric data. States and corporations have used different technologies to establish control over bodies, using bio-metric data that allows algorithms to identify subjects and focus their power on them. Facial recognition systems use multiple cameras and different sensors in public spaces to build databases that allow the precise identification of individuals.8

Strategies adopted by individuals to disidentify bodies include the use of masks, caps, helmets, or even makeup and other artifacts which may "confuse" sensors of facial recognition.11 Another tactic sometimes used is the hindering or even the destruction of sensors. Individuals in public demonstrations in Hong Kong, for instance, confronted sensors with different activists painting cameras to avoid being identified.

In summary, we argue the power of algorithms depends on their capacity to identify and cluster individuals. In refusing to be fully known and transparent, individuals may increase the internal friction of algorithmic systems. Processes of disidentification represent ruptures in the feedback loops generated by algorithms and avoid their tendencies to prophesize our future on the grounds of our (constructed) pasts.

Resistance against algorithms will not always be organized but may be based on a diffuse network of hidden transcripts through which subjects avoid to be known and reinvent themselves. This is not a trivial matter, though. Algorithms are part of enormous and asymmetrical systems of power. We are not denying or reducing the centrality of this dynamic in contemporary societies. We just refuse to concede all the power to algorithms, as if there were no resistance or creative forms of action.

Resistance and creative forms of action against these power systems reinforce the need for new studies on the interaction between humans and machines shaped by algorithms. Inter-disciplinary efforts to study the inter-action between humans and machines must recognize infrapolitics as an essential element, with practical implications for the design and consequences of algorithms in society.

Back to Top

References

1. Almeida, V. et al. Algorithms and institutions: How social sciences can contribute to governance of algorithms. IEEE Internet Computing 26, 2 (2022), 42–46; 10.1109/MIC.2022.3147923.

2. Alnemr, N. Defending Deliberative Democracy in an Algorithmic Society: Building Deliberative Capacity through Contesting Algorithmic Harms. University of Canberra (Oct. 2021).

3. Borak, M. China's gig workers are challenging their algorithmic bosses. Wired (Mar. 14, 2022).

4. Certeau, M. The Practice of Everyday Life. Translated by Steven Rendall. University of California Press, Berkeley, CA, (1984).

5. Ganesh, M.I. and Moss, E. Resistance and refusal to algorithmic harms: Varieties of 'knowledge projects.' Media International Australia 183, 1 (2022), 90–106.

6. Hao, K. and Freischlad, N. The gig workers fighting back against the algorithms. MIT Technology Review, (Apr. 21, 2022); www.technologyreview.com/2022/04/21/1050381/the-gig-workers-fightingback-against-the-algorithms/

7. Lorenz, T. (2022) Internet 'algospeak' is changing our language inreal time, from 'nip nops' to 'le dollar bean', The Washington Post, (Apr. 8, 2022); https://bit.ly/3PYHNTX

8. Petre, C. et al. Gaming the system: Platform paternalism and the politics of algorithmic visibility. Social Media + Society. (2019); https://doi.org/10.1177/2056305119879995

9. Ranciére, J. Dis-Agreement: Politics and Philosophy. University of Minnesota Press, Minneapolis, MN, 1999.

10. Scott, J.C. Domination and the Arts of Resistance. Yale University Press, New Haven, CT, 1990.

11. Shein, E. Using makeup to block surveillance. Commun. ACM 65, 7 (July 2022), 21–23.

12. Velkova, J. and Kaun, A. Algorithmic resistance: media practices and the politics of repair. Information, Communication, and Society 24, 4 (2021) 523–540.

Back to Top

Authors

Ricardo Fabrino Mendonça ([email protected]) is an associate professor of political science at the Federal University of Minas Gerais (UFMG), Brazil. He is also a fellow researcher of CNPq, Fapemig and INCT. DD, Brazil.

Fernando Filgueiras ([email protected]) is an associate professor at the Federal University of Goiás, Goiânia and a professor at the National School of Public Administration, Brasília, Brazil.

Virgílio Almeida ([email protected]) is a faculty associate at the Berkman Klein Center for Internet and Society at Harvard University and a professor emeritus of computer science at the Federal University of Minas Gerais (UFMG), Brazil.

Back to Top

Footnotes

a. See https://bit.ly/46jTkCv


Copyright held by owner(s)/author(s).
Request permission to (re)publish from the owner/author

The Digital Library is published by the Association for Computing Machinery. Copyright © 2023 ACM, Inc.

 


 

No entries found