Chapter 5: Power, Knowledge, and Data

"Knowledge is not made for understanding; it is made for cutting." — Michel Foucault, The Foucault Reader (1984)

Chapter Overview

In the previous four chapters, we've explored what data is, where it came from historically, who claims ownership of it, and how the attention economy monetizes it. A thread runs through each of these discussions, surfacing in different forms: the question of power.

Who has the power to collect data? Who decides what gets measured and what doesn't? Who benefits from data analysis, and who is subjected to its conclusions? These are not merely technical questions. They are fundamentally questions about how societies organize authority, distribute resources, and determine whose experiences count as knowledge.

This chapter introduces the theoretical frameworks that help us analyze data power — not as a simple matter of "tech companies have too much power" (though that may be true) but as a structural phenomenon embedded in institutions, practices, and forms of knowledge themselves.

In this chapter, you will learn to: - Apply power/knowledge theory to data systems - Identify the forms of power that operate through data collection, classification, and analysis - Analyze information asymmetries and their governance implications - Evaluate data colonialism as a framework for understanding platform economics - Recognize strategies for resistance and redistribution of data power

5.1 Foucault and Power/Knowledge

5.1.1 Beyond "Power Over"

Most people think of power in straightforward terms: A has power over B. The employer has power over the employee. The government has power over the citizen. This is what political theorists call sovereign power — the ability to command, coerce, and punish.

Michel Foucault (1926-1984), the French philosopher and historian, argued that this understanding of power is incomplete. Power doesn't just operate through commands and prohibitions. It operates through the production of knowledge itself — through the categories we use to understand the world, the data we collect, the norms we establish, and the expertise we defer to.

Foucault's concept of power/knowledge (pouvoir/savoir) holds that power and knowledge are inseparable. Knowledge is not neutral information discovered by objective observers; it is produced within power relationships and reinforces those relationships. The psychiatrist who diagnoses a patient, the criminologist who classifies a criminal, the statistician who defines a demographic category — all are simultaneously producing knowledge and exercising power.

"This sounds abstract," Mira admitted during the fourth week of class. "Can you give me a concrete example?"

Dr. Adeyemi smiled. "You work in the Office of Institutional Research. Your office produces the data that determines which departments get funding, which programs get expanded, and which get cut. That data doesn't just describe the university — it shapes it. The categories you use, the metrics you prioritize, the questions you ask — all of that is power. Not the power to give orders, but the power to define what counts."

5.1.2 Disciplinary Power and the Digital Panopticon

Foucault's most famous analysis of power focuses on the panopticon — Jeremy Bentham's 18th-century prison design in which a central watchtower allows a single guard to observe all prisoners, while the prisoners cannot tell whether they are being watched at any given moment.

Foucault argued that the panopticon illustrates a broader principle of disciplinary power: when people believe they might be observed, they internalize the observer's norms and regulate their own behavior. Power operates not through actual surveillance but through the possibility of surveillance.

The digital version of this is pervasive. Consider: - Workplace monitoring software tracks keystrokes, mouse movements, screen captures, and application usage. Employees who know they're being monitored modify their behavior even during times when monitoring may not be active. - Social media self-censorship. Users modify their posts, opinions, and associations based on the knowledge that employers, family members, and algorithms are watching. The chilling effect operates even when no specific consequence has been threatened. - Smart city sensors. Residents of neighborhoods with extensive sensor networks — Eli's Detroit neighborhood, for instance — modify their movement patterns, association patterns, and even their conversations in public spaces.

"My grandmother won't talk on the phone about certain things anymore," Eli told the class. "She grew up in Jim Crow Mississippi. She says the cameras and microphones on the lampposts remind her of the informant networks the White Citizens' Council used to track Black community organizing. She doesn't need to know if anyone is actually listening. The possibility is enough."

Connection: We examined the history of surveillance in Chapter 2 and will explore its modern forms in detail in Chapter 8. The Foucauldian insight is that surveillance's power lies not primarily in what it discovers but in how it disciplines — the behaviors people abandon, the thoughts they censor, the associations they avoid, because they know or suspect they are being watched.

5.1.3 Biopower and Data Populations

Foucault's later work introduced the concept of biopower — power exercised not over individual bodies but over populations as biological entities. Biopower operates through statistics, demographics, public health measures, and population management.

In the data age, biopower manifests through: - Population analytics that identify "at-risk" groups, "high-value" customers, and "suspicious" communities - Predictive policing that targets neighborhoods rather than individuals - Public health surveillance that tracks disease patterns, vaccination rates, and behavioral risk factors - Insurance actuarial models that price risk at the population level

The characteristic move of biopower is to convert individual lives into statistical patterns and then govern the patterns. When VitraMed's predictive model identifies a population cohort with elevated diabetes risk, it is exercising biopower — translating individual health data into population-level knowledge that becomes the basis for interventions, resource allocation, and ultimately control.

5.2 Information Asymmetry

5.2.1 Who Knows What About Whom?

Information asymmetry — a term from economics — describes situations where one party to a transaction has significantly more or better information than the other. In data governance, information asymmetries are pervasive and consequential.

The power asymmetry inherent in these information gaps is profound. When one party knows everything about another party and the reverse is not true, the knowledgeable party can predict, influence, and exploit the unknowing party with minimal resistance.

5.2.2 The Transparency Paradox

An intuitive response to information asymmetry is to demand transparency: force platforms to disclose their algorithms, require data brokers to reveal what they know, mandate that governments publish surveillance statistics.

Transparency is necessary but insufficient for several reasons:

1. Complexity opacity. Even when algorithms are disclosed, they may be too complex for non-experts to evaluate. A machine learning model with millions of parameters is "transparent" if published, but it is not comprehensible.

2. Information overload. Privacy policies are technically transparent — they're published documents. But the average person would need 76 working days per year to read all the privacy policies they encounter. Transparency without comprehensibility is theater.

3. Strategic disclosure. Organizations can comply with transparency requirements while obscuring the most consequential information. Disclosing that an algorithm uses "over 100 factors" is technically transparent but practically useless.

4. Power-preserving transparency. Simply knowing how you are being exploited does not give you the power to stop it. If you learn that your insurance company uses your social media activity to price your premiums, you can either stop using social media (a significant cost) or accept the surveillance (no change in power dynamics).

Thought Experiment: Imagine a perfectly transparent surveillance system. Every camera, sensor, and algorithm is publicly documented. You can see exactly what data is collected about you, how it's analyzed, and what decisions it informs. But you cannot opt out — participation is required for basic civic functions. Is this system ethical? Does transparency alone resolve the power asymmetry?

5.3 Data Power in Institutional Contexts

5.3.1 Corporate Data Power

The concentration of data in a small number of corporations creates structural power that extends far beyond the consumer-platform relationship.

Market power. Companies with dominant data positions can erect barriers to competition. A new social media platform cannot replicate Facebook's social graph. A new search engine cannot replicate Google's years of search data. Data concentration creates natural monopolies that are resistant to market correction.

Epistemic power. Platforms shape what people know and believe. Google's search rankings determine what information is accessible. Facebook's news feed determines what events are visible. YouTube's recommendation algorithm determines which ideas get exposure. This is not censorship in the traditional sense — no one is prohibited from speaking — but it is a form of epistemic control that shapes public knowledge.

Labor power. Companies that control data about worker performance, productivity, and behavior have asymmetric power in employment relationships. Algorithmic management systems — explored in Chapter 33 — can monitor, evaluate, direct, and discipline workers through data in ways that were impossible with human management alone.

5.3.2 State Data Power

Government data power operates through different mechanisms:

• Surveillance: The capacity to monitor communications, movements, financial transactions, and associations (Chapter 36)
• Classification: The power to define categories (citizen/non-citizen, legal/illegal, normal/deviant) that determine access to rights and services
• Conscription: The ability to compel data production (census participation, tax reporting, identity documentation)
• Legitimation: The use of data and statistics to justify policy decisions, often selectively

The democratic challenge is that citizens need the state to collect certain data for governance (public health, resource allocation, law enforcement) while simultaneously needing protections against the misuse of that data for surveillance, discrimination, and political control.

5.3.3 Resistance and Counter-Power

Power is never uncontested. Foucault insisted that "where there is power, there is resistance." In the data context, resistance takes several forms:

Sousveillance — a term coined by Steve Mann — refers to the monitoring of authorities by the public, rather than the monitoring of the public by authorities. Body cameras on police officers, citizen journalism documenting government actions, and leak platforms like WikiLeaks all represent forms of sousveillance.

Counter-data involves communities collecting their own data to challenge official narratives. When community organizations in Eli's Detroit neighborhood conducted their own noise surveys to challenge the ShotSpotter data, they were producing counter-data — using the tools of data collection to contest the conclusions of institutional data.

Data activism encompasses organized efforts to change data practices through advocacy, litigation, and direct action. Organizations like the Electronic Frontier Foundation (EFF), Access Now, and local data justice initiatives work to redistribute data power through legal challenges, public campaigns, and alternative technology development.

Obfuscation — described by Finn Brunton and Helen Nissenbaum in their 2015 book — refers to the deliberate production of misleading data to confuse surveillance systems. Browser extensions that generate random search queries, strategies for polluting advertising profiles, and community practices of data noise all represent forms of obfuscation.

Ethical Dimensions: Is obfuscation ethically justified? One perspective holds that it is a legitimate form of self-defense against disproportionate surveillance. Another perspective argues that it undermines the integrity of data systems that also serve beneficial purposes (accurate search results, effective public health surveillance). A third perspective asks: does the ethical judgment depend on who is obfuscating and why? A political dissident obscuring their search history is different from a corporation obscuring its pollution data.

5.4 Data Colonialism

5.4.1 The Framework

Nick Couldry and Ulises Mejias, in their 2019 book The Costs of Connection, argue that the extraction of data by platforms constitutes a new form of colonialism — data colonialism. Just as historical colonialism appropriated land and labor from colonized peoples, data colonialism appropriates human life itself (through data) for capital accumulation.

The parallel is not merely metaphorical: - Appropriation of raw material: Colonial powers extracted natural resources from colonized territories. Data colonialism extracts behavioral data from populations. - Unequal exchange: Colonial trade was structured to benefit the metropole at the expense of colonies. Platform economics is structured to benefit platform companies at the expense of users. - Ideological justification: Colonialism was justified by the "civilizing mission." Data colonialism is justified by the promise of free services, convenience, and innovation. - Erasure of autonomy: Colonialism denied colonized peoples' capacity for self-governance. Data colonialism denies users' capacity to manage their own data by embedding data extraction in the infrastructure of daily life.

5.4.2 Critiques of the Framework

The data colonialism framework has been criticized on several grounds:

• Overextension: Historical colonialism involved physical violence, forced labor, and genocide. Equating these with data extraction may trivialize historical suffering.
• Agency: Platform users are not colonized subjects — they can (in principle) choose not to use platforms. The coercion is real but structurally different from colonial domination.
• Geography: Colonial power operated along geographic lines (metropole vs. colony). Data power operates globally, exploiting populations within wealthy nations as well as between nations.

These critiques have merit, but the framework's proponents argue that the structural parallel illuminates dynamics that would otherwise remain invisible — particularly the way data extraction is presented as natural, inevitable, and mutually beneficial, just as colonial extraction was.

Global Perspective: The data colonialism framework has particular resonance in the Global South, where populations generate data that is processed and monetized by companies headquartered in the United States and China. We'll explore this dimension in depth in Chapter 37 (Global South Perspectives on Data Governance).

5.5 Epistemic Injustice and Data

5.5.1 Whose Knowledge Counts?

The philosopher Miranda Fricker introduced the concept of epistemic injustice — injustice that occurs in a person's capacity as a knower. Two forms are particularly relevant to data governance:

Testimonial injustice occurs when a person's testimony is given less credibility due to prejudice. In data contexts, this manifests when communities report harms from data systems and are dismissed — told they're being paranoid, that the algorithm is objective, or that they're misunderstanding the technology.

Hermeneutical injustice occurs when people lack the conceptual resources to understand their own experiences. Before terms like "surveillance capitalism," "dark patterns," and "algorithmic bias" existed, people who experienced these phenomena had difficulty naming and communicating what was happening to them. They knew something was wrong but lacked the vocabulary to articulate it.

"That's exactly what happened in my neighborhood," Eli said. "People felt surveilled, but when they complained, they were told the sensors were 'just for traffic.' It wasn't until we learned the technical vocabulary — data exhaust, secondary use, scope creep — that we could name what was happening and push back effectively."

5.6 Toward a Theory of Data Justice

Data justice — a framework developed by scholars including Linnet Taylor, Lina Dencik, and Seda Gurses — attempts to move beyond data protection (which focuses on individual rights) toward a structural analysis of how data systems produce and reproduce social injustice.

Data justice asks: 1. Distributive justice: Who benefits from data systems and who bears the costs? 2. Procedural justice: Who participates in decisions about data governance? 3. Recognition justice: Whose experiences and knowledge are represented in data systems? 4. Epistemic justice: Whose interpretations of data are treated as authoritative?

This framework connects data governance to broader social justice movements — civil rights, labor rights, environmental justice, indigenous sovereignty — and insists that data ethics cannot be practiced in isolation from these struggles.

5.7 Chapter Summary

Key Arguments

• Power operates not just through commands and coercion but through the production and control of knowledge
• Data systems create and reinforce power asymmetries through information inequality, epistemic control, and behavioral modification
• Information asymmetry between data collectors and data subjects is a structural feature of the data economy, not an accidental one
• Data colonialism provides a framework for understanding data extraction as a structural relationship of exploitation
• Resistance takes multiple forms: sousveillance, counter-data, data activism, and obfuscation
• Data justice extends data ethics beyond individual rights to structural analysis of how data systems produce social inequality

Key Thinkers

• Michel Foucault: Power/knowledge, disciplinary power, biopower, panopticism
• Shoshana Zuboff: Surveillance capitalism, behavioral surplus (extended from Ch. 4)
• Nick Couldry & Ulises Mejias: Data colonialism
• Miranda Fricker: Epistemic injustice
• Linnet Taylor: Data justice

Key Debates

• Is "data colonialism" an illuminating framework or an overextended analogy?
• Can transparency alone resolve information asymmetry?
• Is the appropriate response to data power redistribution (giving individuals more control) or structural change (changing the systems that produce asymmetry)?

What's Next

In Chapter 6: Ethical Frameworks for the Data Age, we'll equip you with the philosophical tools to evaluate the dilemmas raised throughout Part 1. You'll learn to apply utilitarianism, deontology, virtue ethics, care ethics, and justice theory to data governance questions — and to recognize the strengths and limitations of each framework.

Before moving on, complete the exercises and quiz to solidify your understanding of data power theory.

Chapter 5 Exercises → exercises.md

Chapter 5 Quiz → quiz.md

Case Study: Detroit's Project Green Light — Surveillance and Community Power → case-study-01.md

Case Study: Facebook and Epistemic Power — The 2021 Whistleblower Documents → case-study-02.md