PaperlayerView exact revision on arXiv

Source-linked AI summary

What Makes a Dark Pattern... Dark? Design Attributes, Normative Considerations, and Measurement Methods

Arunesh Mathur, Jonathan Mayer, Mihir Kshirsagar

arXiv:2101.04843v1cs.HCcs.CY

TL;DR

Dark-pattern research has generated descriptive accounts of problematic interfaces but lacks a consistent conceptual foundation for identifying and evaluating them. The paper reviews this literature, connects it with related disciplinary scholarship, and develops normative perspectives with aligned empirical methods. It concludes that dark patterns are grounded in thematically related considerations rather than a singular concern or definition, while offering a framework for future research and policy-oriented measurement.

  • Problem

    Dark-pattern research lacks a clear and consistent conceptual foundation for determining what makes an interface a dark pattern and why it is normatively problematic.

  • Method

    The paper reviews dark-pattern definitions and taxonomies, synthesizes related disciplinary scholarship, develops normative lenses, and proposes aligned HCI methods demonstrated in a case study.

  • Results

    The literature reflects a family of thematically related considerations rather than a singular, unified concern or definition.

  • Takeaways & Limitations

    Future dark-pattern research can apply empirical methods grounded in normative perspectives, helping HCI practitioners and policymakers evaluate designs.

Abstract

from arXiv · show

There is a rapidly growing literature on dark patterns, user interface designs -- typically related to shopping or privacy -- that researchers deem problematic. Recent work has been predominantly descriptive, documenting and categorizing objectionable user interfaces. These contributions have been invaluable in highlighting specific designs for researchers and policymakers. But the current literature lacks a conceptual foundation: What makes a user interface a dark pattern? Why are certain designs problematic for users or society? We review recent work on dark patterns and demonstrate that the literature does not reflect a singular concern or consistent definition, but rather, a set of thematically related considerations. Drawing from scholarship in psychology, economics, ethics, philosophy, and law, we articulate a set of normative perspectives for analyzing dark patterns and their effects on individuals and society. We then show how future research on dark patterns can go beyond subjective criticism of user interface designs and apply empirical methods grounded in normative perspectives.

1 INTRODUCTION

Dark-pattern research has documented many objectionable interfaces, but lacks a consistent conceptual and normative foundation. This paper reviews the literature, develops normative perspectives, and proposes aligned empirical methods for future research.

  • Related work: Existing dark-pattern scholarship predominantly documents and categorizes objectionable user-interface designs across contexts such as privacy and shopping.The literature includes curated collections and studies of designs in specific settings.
  • Research gap: The literature lacks a clear, consistent foundation explaining what makes an interface a dark pattern and why it is normatively problematic.The paper identifies limited attention to the concepts and normative underpinnings behind these judgments.
  • Research gap: This conceptual gap contributes to limited dialogue within HCI and disconnects between research results and possible government responses.Legislators and regulators therefore face difficulty identifying standards for market interventions.
  • Conceptual synthesis: The authors find that dark-pattern research reflects a family of thematically related considerations rather than a singular, unified concern or definition.They synthesize definitions, types, and taxonomies from recent scholarship into cross-cutting themes.
  • Normative foundations: The paper connects dark-pattern themes to scholarship in economics, philosophy, ethics, and law to develop discrete normative lenses for evaluation.Examples include sludges, manipulation, and market manipulation.
  • Future research: It proposes aligning future HCI measurement methods with these normative perspectives and demonstrates the approach through a case study for practitioners and policymakers.The paper presents this alignment as a path beyond subjective criticism of interface designs.

2 A COMPARISON OF DARK PATTERN DEFINITIONS, TYPES, AND ATTRIBUTES

The scoping review finds substantial variation and inconsistency across dark-pattern definitions, while organizing prior work around recurring definitional facets and mechanisms of influence.

  • Review scope: The review compiled 20 academic papers and additional legislation and regulatory materials concerning dark patterns.The academic dataset covered HCI, security and privacy, law, and psychology venues.
  • Definition facets: The authors identified four facets of dark-pattern definitions: interface characteristics, mechanisms of effect, designer roles, and resulting benefits or harms.These facets were surfaced inductively from the reviewed definitions.
  • Mechanisms of effect: Prior definitions describe dark patterns through diverse mechanisms, including subverting intent or preferences, tricking users, manipulating them, and undermining autonomy.Some definitions specify multiple mechanisms of effect.
  • Definition variation: Nine definitions omit interface characteristics, four omit mechanisms of effect, eight omit designer roles, and ten omit benefit or harm elements.The omissions demonstrate substantial variation among the 19 definitions examined.

2.2 Types of Dark Patterns

Prior scholarship has produced many dark-pattern taxonomies across applications, while measurement studies report frequent patterns in websites, consent notices, and apps; together, these efforts reveal conceptual inconsistency.

  • Taxonomy scope: Prior work developed taxonomies for games, mobile apps, physical-proximity systems, home robotics, privacy interfaces, and general web interfaces.Examples include Pay to Skip, Grinding, proxemic patterns, Bad Defaults, and Forced Action.
  • Privacy patterns: Privacy-focused taxonomies document patterns such as Bad Defaults, Hidden Legalese Stipulations, Ease, Framing, Impenetrable Wall, and Last Minute Consent.These patterns concern privacy-unfriendly defaults, hidden information, burdensome alternatives, and consent timing.
  • Measurement studies: Automated crawling identified 1,800 dark patterns on shopping websites, including Sneaking, Social Proof, and Countdown Timers.The finding illustrates measurement of taxonomy-defined patterns at web scale.
  • Measurement studies: Over 50% of cookie banners in a random sample of 1,000 popular EU websites contained at least one dark pattern.Identified examples included privacy-unfriendly defaults, hidden opt-outs, and preselected data-collection checkboxes.
  • Measurement studies: Nearly every one of 300 Scandinavian and English news-site consent notices contained at least one dark pattern, with 297 notices meeting that condition.Other studies similarly found dark patterns in more than 95% of 200 popular Android apps and only 11.8% of sampled UK websites without any pattern.
  • Conceptual inconsistency: Combining diverse taxonomies and definitions exposes a lack of conceptual consistency, including patterns that are not inherently misleading, deceptive, or nonconsensual.The authors therefore conclude that prior work lacks a clear and consistent conceptual foundation.

2.3 Attributes of Dark Patterns

The authors extend higher-level dark-pattern attributes by adding disparate treatment and group the attributes according to how interfaces modify users’ choice architecture.

  • Attribute framework: The authors build on shared higher-level attributes proposed by Mathur et al. and add disparate treatment as an additional attribute.Disparate treatment captures interfaces that disadvantage one group and treat it differently from another.
  • Asymmetric: Asymmetric patterns impose unequal burdens by prominently presenting service-benefiting choices while obscuring or complicating user-benefiting options.Examples include confusing language, guilt, and hidden privacy settings.
  • Covert: Covert patterns steer users toward outcomes while hiding the influence mechanism, using cognitive biases, color, or style.A decoy choice can make another option appear more appealing without users recognizing its influence.
  • Deceptive and information hiding: Deceptive patterns induce false beliefs through affirmative misstatements, misleading statements, or omissions, while information-hiding patterns obscure or delay necessary information.Examples include phony countdown deadlines, hidden subscriptions, and costs revealed late in a transaction.
  • Restrictive: Restrictive patterns reduce or eliminate the choices presented to users, including requirements to accept terms and marketing emails or difficulty cancelling subscriptions.These designs constrain the available choice set rather than primarily altering information.

2.4 Dark Patterns as a Thematic Research Area

The review characterizes dark-pattern scholarship as a family of related concerns organized by choice architecture rather than a single unified definition.

  • User decision space: Dark patterns that are asymmetric, covert, restrictive, or disparate modify users’ decision space by changing the choices available to them.Disparate-treatment interfaces can systematically disadvantage one group while operating covertly.
  • Information flows: Deceptive and information-hiding patterns modify information flows by manipulating or restricting the information available to users.Both themes concern how interfaces shape the information users receive.
  • Thematic organization: The authors show that choice architecture, divided into user decision space and information flows, coheres the current dark-pattern literature.These themes support a taxonomy linking broad concerns, attributes, and specific interface designs.
  • Conceptual foundation: The literature resembles a family of related concepts: its concerns share themes but do not collapse into one specific definition.This framing parallels Solove’s account of privacy as a Wittgensteinian family of concepts.
  • Normative analysis: The authors use cross-disciplinary scholarship to develop normative perspectives for studying problems dark patterns pose for individuals and society.The normative turn is intended to move analysis beyond merely cataloging interface designs.

3 CHOICE ARCHITECTURE IN BEHAVIORAL ECONOMICS, PHILOSOPHY, AND LAW

Research on choice architecture across behavioral economics, philosophy, ethics, and law offers concepts for understanding how interfaces modify users’ decisions and the normative concerns involved.

  • Behavioral economics: Behavioral economics examines how informational and cognitive limits cause decision-making under uncertainty to deviate from classical economic models.
  • Behavioral economics: Nudges steer people toward particular choices while preserving the ability to choose otherwise, whereas sludges impose excessive or unjustified friction.Sludges can cost time or money, make activities difficult to navigate, and restrict access to goods or services.
  • Behavioral economics: Nudges and sludges describe tools that advance or detract from policy goals, but they do not themselves provide a normative rationale for choosing those goals.
  • Philosophy and ethics: Philosophical and ethical accounts characterize manipulation as subverting decision-making and denying individuals authorship over their choices, distinguishing it from coercion and persuasion.Scholars disagree about whether manipulation must be covert or can also operate through overt incentives.
  • Philosophy and ethics: Normative critiques of manipulation emphasize exploitation, unfairness, impoverishment, and threats to autonomy, especially when manipulation targets vulnerabilities.
  • Law: Legal scholarship describes market manipulation as exploiting non-rational behavior and cognitive biases for profit, with digital marketplaces increasing targeting scale and sophistication.Such manipulation can increase economic harm, reduce market efficiency, undermine autonomy, and cause privacy loss.

4 NORMATIVE PERSPECTIVES ON DARK PATTERNS

The dark-patterns literature is largely descriptive, while its underlying normative concerns remain underdeveloped. The paper organizes those concerns into four lenses to support a common language for research.

  • Dark-patterns research has focused largely on describing interface designs, despite nascent concerns about why those designs should matter.
  • The paper draws normative considerations from prior literature and organizes them into four lenses for analyzing dark patterns.
  • The lenses are not intended to appeal equally to everyone; they are offered as a common language for discussing problematic practices.

4.1 Individual Welfare

The individual-welfare lens evaluates dark patterns by whether they benefit designers at users’ expense. It covers financial loss, privacy harms, and users’ unnecessary expenditure of time, energy, and attention.

  • Under the individual-welfare lens, a dark pattern modifies choice architecture to benefit the designer at the expense of the user’s welfare.
  • Financial-loss patterns can push users toward unintended purchases, recurring fees, added products, or products they might not otherwise have bought.
  • Privacy-related dark patterns can expose data through defaults, obstruct privacy-respecting choices, or use emotional language to discourage those choices.The literature also describes privacy harms as involving confusion, hidden protective options, required registration, and opaque legalese.
  • Cognitive-burden patterns impose unnecessary costs in time, energy, and attention, including through difficult cancellation, repeated prompts, and hard-to-exercise denial options.
  • Evaluation: The individual-welfare perspective lacks a clear baseline because preferences are unstable, vary across people, and can require large-scale measurement.
  • Evaluation: Because marketing routinely shapes preferences, this lens offers limited guidance for distinguishing benign or tolerable practices from practices warranting sanctions.
  • Evaluation: The perspective also omits competitive-market dynamics, possible welfare benefits of shaping decision spaces, and the costs of imposing liability for marketing practices.

4.2 Collective Welfare

The collective-welfare lens evaluates whether interfaces benefit designers at the expense of society and markets. It broadens analysis beyond individual users to competition, transparency, trust, and unintended social effects.

  • The collective-welfare lens defines dark patterns as interfaces that modify choice architecture to benefit designers at the expense of collective welfare.
  • Competition: Dark patterns can impair competition by creating switching costs, reinforcing market power, tying products, and raising barriers to entry for competitors.
  • Price transparency: Opaque pricing can prevent comparison shopping, cause potential financial losses, and reduce marketplace competition by obstructing informed decisions.
  • Trust: Dark patterns can undermine trust in markets and harm honest companies when users become skeptical of interface elements and miss genuine deals.
  • Unintended effects: Collective-welfare analysis must evaluate unintended or cumulative side effects, including societal costs that may arise even when a particular user accepts a design’s trade-off.
  • Evaluation: The collective-welfare approach requires a mechanism for balancing the different costs and benefits produced by design choices.

4.3 Regulatory Objectives

The regulatory objectives lens evaluates dark patterns against the rules and standards of the applicable legal regime. It offers measurable criteria but does not independently explain why the protected values matter.

  • Under this lens, a dark pattern modifies choice architecture to interfere with specific regulatory objectives.
  • U.S. consumer-protection enforcement addresses deceptive or unfair practices, including obstacles to consumers’ free exercise of decision-making.
  • Legal regimes vary: EU policymakers tend toward detailed rules, while U.S. regulators generally use principles-based approaches.
  • GDPR-related research has treated consent interfaces as potentially unlawful dark patterns when consent is not explicit, easy to accept or deny, or uses preselected checkboxes.
  • Evaluation: The lens is easier to translate into empirical metrics because regulations often specify standards, but compliance alone does not supply a normative rationale for protected values.

4.4 Individual Autonomy

The individual autonomy lens treats dark patterns as interfaces that undermine users’ ability to make decisions on their own reasons. It captures many existing concerns but raises broadness, idealization, boundary, and measurement challenges.

  • Under the autonomy lens, a dark pattern undermines individual decision-making by changing, denying, obscuring, or burdening users’ choices.
  • Digital-addiction patterns are associated with psychological harms such as poor concentration and physical harms such as sleep disturbance.
  • Autonomy concerns span dark patterns that manipulate users into unintended actions and interfaces that promote digital addiction.
  • Evaluation: The autonomy perspective captures many prior definitions and taxonomies, but it can classify any decision-making interference as dark regardless of its outcome.
  • Evaluation: Open challenges include accounting for limited information and deliberation, distinguishing permissible burdens from autonomy violations, and measuring violation extent.

5 MEASUREMENT METHODS AND APPLICATIONS

The paper connects four normative lenses to established HCI measurement methods, arguing that empirical studies should choose methods and metrics according to the concern being evaluated. It illustrates this approach with a Trick Questions case study.

  • The four normative lenses provide perspectives for analyzing dark patterns, but researchers must still determine how to evaluate interfaces in practice.
  • Grounding empirical analysis in normative considerations can make findings about problematic design aspects more principled and defensible.
  • Empirical Research Methods in HCI: Surveys, interviews, focus groups, ethnography, lab studies, field deployments, and experiments offer different kinds of user-experience and behavioral evidence.
  • Empirical Research Methods in HCI: Lab studies permit controlled observation and rigorous measures such as gaze tracking, but their findings may not generalize to real-world interaction.
  • Empirical Research Methods in HCI: Experiments compare randomly assigned control and treatment conditions to estimate treatment effects, with lab settings favoring control and field settings favoring realism.
  • Measurement Applications: Individual-welfare studies can compare dark-pattern and neutral interfaces using outcomes such as monetary loss, preference satisfaction, privacy value, cognitive load, or frustration.
  • Measurement Applications: Because many measurements require a baseline interface, comparisons with a neutral alternative are central to interpreting whether a design functions as a dark pattern.
  • Applications: The authors recommend explicitly selecting a normative perspective and then choosing research methods that can illuminate it, illustrated through a Trick Questions case study.

6 CONCLUSION

The paper advances a normative and empirical foundation for dark-pattern research by connecting cross-disciplinary perspectives to measurement methods. It argues that principled metrics can support researchers, regulators, policymakers, and practitioners confronting an under-conceptualized problem.

  • The paper connects dark-pattern scholarship with other disciplines, develops normative perspectives, and grounds empirical research methods in those perspectives.
  • The authors argue that normative foundations can motivate research problems and methods while adding rigor to analyses of effects on individuals and society.
  • Principled evaluation methods can help HCI practitioners critically examine their own user-interface practices and support ethical design standards.
  • The paper acknowledges that describing exploitative mechanisms carries risks, but states that empowering researchers and regulators to examine those risks outweighs them.
  • Policymakers have struggled with dark patterns’ amorphous, under-conceptualized character, leaving government action largely within traditional consumer-protection and privacy law.
  • Developing metrics and methods for regulatory agencies could help address this policy impasse.
Loading 2101.04843v1…