Lecture 11: Data governance and accessibility

Data

Facts & statistics collected together for reference or analysis

Algorithms

Sets of rules or parameters that guide computer calculations

Machine Learning

An approach that allows algorithms to adjust their parameters as they take in new data

Data Governance

Policies, standards, and practices that control data ownership, access & sharing, security, and accountability; includes regulatory frameworks and ethical concerns

Data Ownership

A key concept in data governance concerning who owns the data

Data Stewardship

Responsibility for the care and ethical use of data

Right to be Forgotten

Concept dealing with how data removal intersects with freedom of expression, privacy, and digital identity

Why Data Governance Matters

Data is power; poor governance leads to inequity, misuse, exclusion; accessibility defines who benefits from data

Data for a Purpose

Data is collected to track trends, inform decision-making, support innovation, and understand global and local systems

Why is Data Political?

Data is shaped by historical and political judgments; categorization is not neutral

Who's Included?

Question of who is represented in data and who is left out; categories reflect power dynamics

Who is Counting?

Questions of who collects data, the validity of sources, and whose narrative is prioritized

Which Categories are Created?

Data categories can be politicized, misrepresented, and subject to unequal investment

Whose Interests are Represented?

Stakeholders such as states, industries, and individuals influence data formation and use

Are You a Human?

CAPTCHAs test cultural knowledge and may present biases; challenge-response authentication

Facial Recognition Case Study

Facial recognition algorithms misidentify black women more than white men, revealing algorithmic bias

Surveillance Capitalism (Zuboff)

An economic system where user data is harvested and commodified without full consent

Behavioural Surplus

Data beyond what’s needed for service, extracted for profit in surveillance capitalism

Prediction Products

User data is modeled to predict behavior for profit

Markets for Future Behaviour

Selling predictive user data to advertisers, insurers, and more

Surveillance Capitalism: Dangerous Idea

Shift from monitoring behavior to shaping it for profit

Facebook & 2016 Election

Data used for micro-targeted political ads; raised issues around transparency, privacy, and accountability

Information Asymmetry (Akerlof)

One party has more or better information; leads to power imbalance and exploitation

Strategies to Reduce Asymmetry

Transparency mandates, right to explanation, and data portability

Challenges in Reducing Asymmetry

Technical complexity, accessibility gap, and corporate resistance to transparency

Infrastructure (Bowker & Star)

A sociotechnical system that underlies and supports broader data frameworks and social coordination

Properties of Infrastructure

Embedded, transparent, learned through membership, built on installed base, visible on breakdown

Figure-Ground Problem

Infrastructures (ground) often invisible behind data technologies (figure); requires inversion

The Privacy Paradox (Brown)

People express concern for privacy but often share data freely; behavior doesn't match values

Reasons for Privacy Paradox

Cognitive biases (present/optimism bias), information asymmetry, limited control

Critiques of Privacy Paradox

Some argue tradeoffs are rational; critique blames users instead of system design

Power in Data Governance

Those who control data hold power over participation, decisions, and resources

Challenges in Data Governance

Issues include security, complexity, and international regulatory differences

Ethics in Data Governance

Informed consent, data minimization, and transparency are key ethical principles

What is Data Accessibility?

Making data usable for all, including those with disabilities or limited digital literacy

Digital Inclusion

Ensuring marginalized or underrepresented groups have equal opportunities in data use

Universal Design

Creating systems that are accessible by default, not retrofitted

Data and Advanced Computing Landscape

Large-scale data collection is resource intensive and dominated by powerful entities

Corporate Influence in Data Governance

Tech corporations have monopolistic power and shape regulation via lobbying

Why Data Accessibility Matters

Ensures social equity and ethical responsibility; bridges the digital divide

Critical Data Analysis

Asking who collected data and why, maintaining heterogeneity and multiple representations

Narratives in Data Analysis

Must be context-specific (heteropraxial) and polyvocal (multiple voices)

Influence of Power in Data Analysis

Power can be institutional, structural, or financial, influencing data practices

Participatory Process

Involving communities and diverse stakeholders in data creation and interpretation