Source-linked AI summary
End User Accounts of Dark Patterns as Felt Manipulation
Colin M. Gray, Jingle Chen, Shruthi Sai Chivukula, Liyang Qu
TL;DR
Digital services can personalize influence to pursue monetary gain, while dark patterns can put stakeholder value ahead of users. This paper uses surveys and follow-up interviews to examine end users’ felt manipulation, documenting a temporal progression in perceptions and related emotions and judgments about creator intent.
Problem
Digital services can track users and personalize prompts for monetary gain, while research has limited direct accounts of how end users experience manipulation.
Source: §1 · pp. 1–2Method
Results
Users’ perceptions of manipulation include experiential characteristics that progress from initial perception through extended interaction, alongside emotions and judgments about designer intent and responsibility.
Source: §9 · p. 19Takeaways & Limitations
The end-user account is intended to inform public policy, user empowerment, and design ethics by connecting dark-pattern discourse with users’ experiences.
Source: §1 · p. 2
Abstract
from arXiv · showhide
Manipulation defines many of our experiences as a consumer, including subtle nudges and overt advertising campaigns that seek to gain our attention and money. With the advent of digital services that can continuously optimize online experiences to favor stakeholder requirements, increasingly designers and developers make use of "dark patterns"---forms of manipulation that prey on human psychology---to encourage certain behaviors and discourage others in ways that present unequal value to the end user. In this paper, we provide an account of end user perceptions of manipulation that builds on and extends notions of dark patterns. We report on the results of a survey of users conducted in English and Mandarin Chinese (n=169), including follow-up interviews from nine survey respondents. We used a card sorting method to support thematic analysis of responses from each cultural context, identifying both qualitatively-supported insights to describe end users' felt experiences of manipulative products, and a continuum of manipulation. We further support this analysis through a quantitative analysis of survey results and the presentation of vignettes from the interviews. We conclude with implications for future research, considerations for public policy, and guidance on how to further empower and give users autonomy in their experiences with digital services.
1 INTRODUCTION
Digital services can personalize influence at scale, while dark patterns prioritize business and shareholder value over users. This paper centers end users’ felt experiences to extend dark-pattern accounts and inform ethics, policy, and user empowerment.
1 INTRODUCTION
- Digital systems can track consumer behavior and tailor prompts to optimize monetary gain, continuing a longstanding tension between marketing and societal good.The paper links this capability to the “surveillance economy” and earlier concerns about advertising priorities.Source: §1 · p. 1
- Dark patterns and dark UX have emerged alongside growth hacking as ways to acquire customers and sustain a customer base through manipulative design.The paper connects these practices to a longer history of advertising manipulation.Source: §1 · p. 2
- The study examines end users’ felt manipulation as a way to connect dark-pattern discourse with user experience and potential policy and design-ethics responses.The authors use felt manipulation as a proxy because participants were not directly asked about dark patterns.Source: §1 · p. 2
- The paper documents users’ experiences and their perceptions of who creates manipulative technologies, including designers’ ethics and responsibility.These contributions aim to support further work at the intersections of user trust, designer intent, and public policy.Source: §1 · p. 2
2 BACKGROUND WORK
Prior work treats manipulation as influence over decisions, but its relationship to persuasion, nudging, agency, and social good remains contested. Dark-pattern research has broadened from interface tactics to questions of user agency, regulation, and users’ own perceptions.
2.1 Manipulation in Digital Systems
- Accounts of manipulation span philosophical, critical-discourse, and legal perspectives, including decision-making vulnerabilities such as anchoring, framing, and overchoice.The paper adopts hidden influence as a reference point but uses manipulation more broadly because links between felt experience and known dark patterns remain underdeveloped.Source: §2.1 · pp. 2–3
- In HCI and CSCW, manipulation is relatively underexamined, and its boundaries with persuasion, nudging, and coercion remain contested around agency and social good.The paper relies on Susser et al.’s account of covert influence targeting decision-making vulnerabilities, while using the term more broadly.Source: §2.1 · p. 3
2.2 Dark Patterns and the Optimization of Shareholder Value
- Dark patterns describe strategies that supplant user value with shareholder value, including nagging, obstruction, sneaking, interface interference, and forced action.The literature also distinguishes covert dark patterns from overtly coercive “asshole design.”Source: §2.2 · pp. 3–4
- Research extends dark-pattern concerns across design, gaming, privacy, security, and law, while policy efforts address transparency, agency, and consent.Maier and Harr’s interviews and focus groups found that users assess manipulation through impressions, balance, and acceptability; this paper builds on that work with English- and Mandarin-speaking users.Source: §2.2 · p. 4
3 OUR APPROACH
The study uses mixed methods to describe how users experience manipulative technology across English- and Mandarin-speaking contexts, without aiming to prove cultural differences. Its survey combines open-ended accounts with questions about reactions and perceptions of creators.
3 OUR APPROACH
- The study asks what makes digital products feel manipulative, what emotions users associate with those experiences, and how aware they are of the technologies’ creators.English- and Mandarin-speaking participants are included to broaden interpretations across cultural and geographic boundaries, not to compare contexts directly.Source: §3 · p. 4
3.1 Survey Design
- A 22-question survey combined multiple-choice and open-ended items on technology use, manipulation experiences, emotional reactions, perceived user value, and awareness of creators.The survey protocol was piloted with over 50 participants, revised using feedback, validated in English, then translated into Mandarin and reviewed by a second native speaker.Source: §3.1 · p. 5
3.2 Survey Distribution
The survey targeted everyday users through researchers’ networks and social media, yielding 169 complete eligible responses, followed by nine interviews across English- and Mandarin-speaking participants. The sample included many heavy technology users.
3.2 Survey Distribution
- The survey was distributed through research-team networks and social media to reach everyday users, with a gift-card or WeChat-currency prize draw.The target respondents regularly used digital products but lacked detailed design or development knowledge.Source: §3.2 · p. 5
3.3 Summary of Survey Responses
- 169 of 253 survey responses were complete and eligible; participants came from China, the United States, the United Kingdom, Canada, and other locations.82.2% (n=143) identified as heavy technology users, reporting 5–14 hours of mobile and desktop use daily.Source: §3.3 · pp. 5–6
3.4 Follow-up Interviews
- Nine follow-up interviews included five Mandarin speakers and four English speakers, lasted about 15–30 minutes, and explored participants’ stories and views of designers.Interviews were conducted in participants’ native languages, recorded, and transcribed; Mandarin dialogue was retained with excerpts in the same dialect.Source: §3.4 · p. 6
3.5 Data Analysis
The analysis combines card sorting of open-ended responses, descriptive statistics for choice- and scale-based questions, and interview vignettes to examine users’ experiences and perceptions of manipulation.
3.5.1 Card Sorting.
- Card sorting mixed English and Chinese responses, organizing 609 data units from 169 surveys into 12 preliminary themes and a continuum of felt manipulation.Four researchers collaboratively sorted the cards, with Mandarin speakers focusing mainly on Chinese responses; the themes were later condensed.Source: §3.5.1 · pp. 6–7
3.5.2 Quantitative Analysis.
- Descriptive analyses examined emotion ratings, awareness of manipulation and blame by age, and how users felt valued.The analyses used frequencies of responses across emotion scales and age segments.Source: §3.5.2 · p. 7
3.5.3 Case Analysis.
- Three interview vignettes illustrate users’ felt experiences, whom they blamed, and how they wished to respond to creators.The cases frame findings and future-work opportunities; they are not intended to represent the whole dataset.Source: §3.5.3 · p. 7
4 RQ#1: DESCRIBING END USERS’ FELT EXPERIENCES OF MANIPULATION
The section examines how users identify manipulation in digital technologies and organizes relevant characteristics along a temporal continuum, from first impressions to longer-term interaction.
4 RQ#1: DESCRIBING END USERS’ FELT EXPERIENCES OF MANIPULATION
- The analysis first identifies aspects of technological systems that may help users recognize manipulation.This frames the section’s account of users’ felt experiences.Source: §4 · p. 7
- The authors organize characteristics of felt manipulation along a continuum that progresses over time.The continuum begins with initial impressions and extends to longer-term interaction with a product.Source: §4 · p. 7
- The progression links initial impressions with characteristics that emerge during longer-term user interaction.Source: §4 · p. 7
4.1 Perceptions of ‘Manipulation’
Users described manipulation through distrust, privacy and security concerns, tracking, explicit tactics, and freemium products. These perceptions were shaped by how products handled information and by users’ experiences with their interfaces and business practices.
4.1.1 Overall Perception of Trust or Distrust.
- Distrust (n=103) could make users feel manipulated even when they could not locate its source; external sources such as reviews, friends, and security alerts informed judgments.English-speaking participants mentioned cues such as foreign hosting or poor documentation, while Chinese participants were wary of bundled software.Source: §4.1.1 · pp. 7–8
4.1.2 Collection of Personal Information.
- Aggressive requests for personal information (n=72) felt manipulative when they exceeded users’ threshold, especially for sensitive data or location and contact permissions.Participants also raised concerns about phishing sites and long agreements that could hide unfair information practices.Source: §4.1.2 · p. 8
4.1.3 Threats of Big Data.
- Unnotified tracking and use of interaction data (n=51) felt threatening when linked to targeted advertisements or search results shaped by interests rather than queries.Participants inferred that information could come from other digital products and be used for business purposes.Source: §4.1.3 · pp. 8–9
4.1.4 Barriers to get security.
- Security barriers (n=45)—including fraud or scams, viruses, and hacking—made users feel their property or information was threatened.Mandarin-speaking participants raised concerns such as illegal software and fake or unreliable websites.Source: §4.1.4 · p. 9
4.1.5 Awareness of Explicit Manipulation.
- Some participants (n=28) recognized explicit tactics, such as placing an agree button where users expect a continue button, as potentially manipulative.The placement can lead users to accept rules or policies without reading them carefully.Source: §4.1.5 · p. 9
4.1.6 Use of Freemium Products.
- Freemium products (n=21) raised suspicion when payment requirements or in-app purchases were not clearly disclosed, particularly when a product advertised as free later demanded payment.Participants described payment traps and mandatory fees during use.Source: §4.1.6 · pp. 9–10
4.2 Temporal Progression of “Felt” Manipulation
The paper presents felt manipulation as a temporal progression, from immediate impressions through inspection and interaction to harms recognized after extended use. The continuum includes both design cues and user judgments about a product’s usefulness and behavior.
4.2 Temporal Progression of “Felt” Manipulation
- The continuum organizes characteristics from initial and general impressions toward direct user interactions.Table 1 presents specific characteristics of manipulation (n=208), with its top focused on initial impressions and its bottom on direct interaction.Source: §4.2 · p. 10, Table 1 · p. 10
4.2.1 Initial Judgment.
- Initial judgments often drew on intuition that something was wrong, including an interface that looked unprofessional or had flashy design.Source: §4.2.1 · p. 10
4.2.2 Inspection of Details.
- On closer inspection, users scrutinized advertising quantity, quality, and format, as well as other interface details such as misspellings or links to social platforms.One participant inspected code or monitored traffic for signs of manipulation.Source: §4.2.2 · p. 10
4.2.3 Felt Persuasion.
- Users distrusted content they experienced as extreme, misleading, emotionally charged, too-good-to-be-true, or tied to incentives and limited choices favoring creators.Source: §4.2.3 · p. 11
4.2.4 General Conclusion.
- After initial interaction, users combined their first impressions with usability and usefulness judgments, sometimes linking complicated use to potential manipulation.Source: §4.2.4 · p. 11
4.2.5 Undesired Interactions.
- Undesired interactions—such as article clicks turning into ads, forced redirects, or glitches—could lead users to judge a product as manipulative.Some participants distinguished intentional designs from errors that also affected perceived quality.Source: §4.2.5 · p. 11
4.2.6 Negative Results from Interaction.
- Some users recognized manipulation only after extended use brought negative consequences, including unwanted downloads, data use, difficult exits, subscription traps, or repetitive gamified tasks.The account includes both direct harms and patterns that took time to discover.Source: §4.2.6 · pp. 11–12
5 RQ#2: EMOTIONS RELATED TO MANIPULATIVE EXPERIENCES
Users reported strong negative emotions during manipulative digital experiences, especially feeling upset, hostile, and irritable; guilt and shame were usually less strongly felt.
5 RQ#2: EMOTIONS RELATED TO MANIPULATIVE EXPERIENCES
- Users most often reported feeling upset (n=107), hostile (n=89), and irritable (n=86) during manipulative experiences.Distress was also frequently reported (n=82).Source: §5 · p. 12
- Users also reported scared (n=59), jittery (n=62), nervous (n=52), and afraid (n=56) responses to manipulation.These were described as negative but more neutral emotions.Source: §5 · p. 12
- Users most often rated guilt (n=123) and shame (n=121) as very slight or absent; 34 blamed themselves for manipulation.The authors note that age-group proportions varied, but the descriptive counts were not normalized for uneven group sizes.Source: §5 · p. 12
6 RQ#3: USERS’ AWARENESS ABOUT THE CREATION AND CREATORS OF MANIPULATION PRODUCTS
Survey responses and follow-up interviews describe users’ awareness of manipulation, their views of its creators, and how those views developed through experience. Users commonly recognized manipulation and blamed designers or company stakeholders, while the vignettes show delayed recognition, constrained choices, and concerns about creators’ priorities.
6.1 Level of Awareness
- 79.3% of respondents (n=134) said smartphone applications or websites were designed to manipulate them, with affirmative responses exceeding two-thirds in every age bracket.Three respondents did not answer the awareness question.Source: §6.1 · p. 13
- Designers (n=97), other stakeholders (n=92), and developers (n=54) were blamed more often than users themselves (n=34) or technological issues (n=8).71.6% blamed only designers and company stakeholders (n=121), while 14.2% blamed all three creator groups (n=24).Source: §6.1 · p. 14
- Mistrust was common: 31.4% (n=53) frequently mistrusted smartphone applications and 36.7% (n=62) frequently mistrusted websites.Most respondents selected “sometimes” for both types of digital products.Source: §6.1 · p. 14
6.2 Creators of Manipulative Digital Products
- Most respondents felt valued more as customers than as people: 62.11% (n=100) agreed or strongly agreed they were valued as customers, versus 27.6% (n=42) as people.Even respondents who had not recognized manipulation still felt more valued as customers than as people.Source: §6.2 · pp. 14–15
6.3.1 Vignette 1: Trapped in Auto-Subscription.
- Amy recognized the face-cream auto-subscription only after it cost her money and time, then blamed herself as well as designers and stakeholders.She felt creators had hidden information and prioritized customer acquisition and monetary benefit over users’ and social values.Source: §6.3.1 · pp. 15–16
6.3.2 Vignette 2: Gamified “Stepping for Cash”.
- Liu’s use of Qubu revealed that advancing and earning more required marketing the app to others; she later feared involvement and viewed it as a variation of multi-level marketing.She criticized creators’ financial incentives but had difficulty assigning them full blame because she felt they had not broken laws.Source: §6.3.2 · p. 16
6.3.3 Vignette 3: Forced Collection of Personal Information.
- Wang felt forced to grant unnecessary permissions because the vehicle-rental app was his only option, heightening his concerns about data collection and privacy.He viewed stakeholders as especially responsible for discouraging manipulation, while designers and developers should balance stakeholder goals with users’ rights.Source: §6.3.3 · pp. 16–17
7 DISCUSSION
The discussion links users’ felt manipulation to dark-pattern strategies across a temporal journey, then considers how users’ perceptions and language might support advocacy and policy.
7.1 Linking Experiences of Manipulation to Dark Patterns Strategies
- Users assess context, past experience, trust, control, data risks, and changing freemium-product risks when balancing value against threats.These assessments extend prior accounts of user impressions, assessment, and acceptability.Source: §7.1 · p. 17
- 24 instances involved initial judgments, while 130 occurred after judgment but before interaction, often as a sense that something was “off” without precise language.Obstruction and forced action could emerge after negative or unwanted interactions; nagging required experience over time.Source: §7.1 · p. 17
- Users can identify suspected manipulative intent across the user journey even when they lack overarching language such as “dark patterns.”Their impressions and assessments may inform whether they use a system, scrutinize it, or seek protection, when choice is possible.Source: §7.1 · p. 18
7.2 Felt Manipulation as a Potential Platform for User Advocacy
- The authors propose felt-manipulation frameworks as a first step toward user awareness, advocacy, and possible recourse, while noting that crowdsourced examples may not themselves empower users.They discuss mainstreaming “dark patterns” language as one strategy, while warning it could dilute the term’s meaning.Source: §7.2 · p. 18
- A broader vocabulary distinguishing persuasion, manipulation, and coercion could help users connect unease with possible design intentions.The paper frames these categories in relation to capitalist goals and the availability of alternatives.Source: §7.2 · p. 19
8 FUTURE WORK
The authors outline future research on users’ felt manipulation, ways to assess sensitivity and product experiences, and paths from awareness to advocacy and policy.
8 FUTURE WORK
- Future research could characterize users’ emotions and sensitivity, test factors such as age and technology experience, and assess manipulation through product-lifecycle user testing.Other proposed directions include conceptualizing felt manipulation and supporting design students and practitioners.Source: §8 · p. 19
- The authors call for participatory or crowdsourced advocacy research and a transdisciplinary account of manipulation across disciplines and stakeholders.The proposed research agenda also includes identifying problematic strategies and contributing to policy development.Source: §8 · p. 19
- Further studies could test specific dark patterns across the perception continuum and explore whether user awareness can translate into policy recourse or collective action.The authors point to experimental assessment and governance structures such as GDPR as possible reference points.Source: §8 · p. 19
9 CONCLUSION
The paper reports a survey and follow-up interviews with English- and Mandarin-speaking participants to describe felt manipulation, its temporal progression, and implications for user advocacy and agency.
9 CONCLUSION
- The study documents characteristics of digital-product experiences that users relate to manipulation and dark patterns.The authors draw on a survey and follow-up interviews with English- and Mandarin-speaking participants.Source: §9 · p. 19
- Users’ perceptions of manipulation progress from initial impressions through extended interaction and include emotions and judgments about designer intent and responsibility..Source: §9 · p. 19
- The authors identify opportunities to connect manipulation research and policy with user assessments to support advocacy and agency.Source: §9 · p. 19