Sociotechnical Audits: Broadening the Algorithm Auditing Lens to Investigate Targeted Advertising

Algorithm audits are powerful tools for studying black-box systems without direct knowledge of their inner workings. While very effective in examining technical components, the method stops short of a sociotechnical frame, which would also consider users themselves as an integral and dynamic part of the system. Addressing this limitation, we propose the concept of sociotechnical auditing: auditing methods that evaluate algorithmic systems at the sociotechnical level, focusing on the interplay between algorithms and users as each impacts the other. Just as algorithm audits probe an algorithm with varied inputs and observe outputs, a sociotechnical audit (STA) additionally probes users, exposing them to different algorithmic behavior and measuring their resulting attitudes and behaviors. As an example of this method, we develop Intervenr, a platform for conducting browser-based, longitudinal sociotechnical audits with consenting, compensated participants. Intervenr investigates the algorithmic content users encounter online, and also coordinates systematic client-side interventions to understand how users change in response. As a case study, we deploy Intervenr in a two-week sociotechnical audit of online advertising (N = 244) to investigate the central premise that personalized ad targeting is more effective on users. In the first week, we observe and collect all browser ads delivered to users, and in the second, we deploy an ablation-style intervention that disrupts normal targeting by randomly pairing participants and swapping all their ads. We collect user-oriented metrics (self-reported ad interest and feeling of representation) and advertiser-oriented metrics (ad views, clicks, and recognition) throughout, along with a total of over 500,000 ads. Our STA finds that targeted ads indeed perform better with users, but also that users begin to acclimate to different ads in only a week, casting doubt on the primacy of personalized ad targeting given the impact of repeated exposure. In comparison with other evaluation methods that only study technical components, or only experiment on users, sociotechnical audits evaluate sociotechnical systems through the interplay of their technical and human components.

[1]  J. N. Matias.  Influencing recommendation algorithms to reduce the spread of unreliable news by encouraging humans to fact-check articles, in a field experiment , 2023, Scientific reports.

[2]  J. N. Matias.  Humans and algorithms work together — so study them together , 2023, Nature.

[3]  Michael S. Bernstein,et al.  End-User Audits: A System Empowering Communities to Lead Large-Scale Investigations of Harmful Algorithmic Behavior , 2022, Proc. ACM Hum. Comput. Interact..

[4]  Paul G. Allen,et al.  What factors affect targeting and bids in online advertising?: a field measurement study , 2022, IMC.

[5]  Piotr Sapiezynski,et al.  Measurement and analysis of implied identity in ad delivery optimization , 2022, IMC.

[6]  Inioluwa Deborah Raji,et al.  Who Audits the Auditors? Recommendations from a field scan of the algorithmic auditing ecosystem , 2022, FAccT.

[7]  Ho-Chun Herbert Chang,et al.  Targeted Ads and/as Racial Discrimination: Exploring Trends in New York City Ads for College Scholarships , 2021, HICSS.

[8]  Deirdre K. Mulligan,et al.  Searching for Representation: A sociotechnical audit of googling for members of U.S. Congress , 2021, ArXiv.

[9]  Motahhare Eslami,et al.  Everyday Algorithm Auditing: Understanding the Power of Everyday Users in Surfacing Harmful Algorithmic Behaviors , 2021, Proc. ACM Hum. Comput. Interact..

[10]  Paul G. Allen,et al.  What Makes a “Bad” Ad? User Perceptions of Problematic Online Advertising , 2021, CHI.

[11]  Jack Bandy,et al.  Problematic Machine Behavior , 2021, Proc. ACM Hum. Comput. Interact..

[12]  James A. Landay,et al.  An Image of Society , 2021, Proc. ACM Hum. Comput. Interact..

[13]  Aleksandra Korolova,et al.  Auditing for Discrimination in Algorithms Delivering Job Ads , 2021, WWW.

[14]  Christo Wilson,et al.  Building and Auditing Fair Algorithms: A Case Study in Candidate Screening , 2021, FAccT.

[15]  Jim Holmes.  Subprime Attention Crisis: Advertising and the Time Bomb at the Heart of the Internet , 2021, Journal of Telecommunications and the Digital Economy.

[16]  Ilya Sutskever,et al.  Learning Transferable Visual Models From Natural Language Supervision , 2021, ICML.

[17]  Jack Bandy,et al.  Problematic Machine Behavior , 2021, Proc. ACM Hum. Comput. Interact..

[18]  Alan Mislove,et al.  On the Potential for Discrimination via Composition , 2020, Internet Measurement Conference.

[19]  Karrie Karahalios,et al.  Auditing Race and Gender Discrimination in Online Housing Markets , 2020, ICWSM.

[20]  Ron Kohavi,et al.  Online randomized controlled experiments at scale: lessons and extensions to medicine , 2020, Trials.

[21]  Eni Mustafaraj,et al.  The case for voter-centered audits of search engines during political elections , 2020, FAT*.

[22]  N. Shadbolt,et al.  'I Just Want to Hack Myself to Not Get Distracted': Evaluating Design Interventions for Self-Control on Facebook , 2020, CHI.

[23]  Inioluwa Deborah Raji,et al.  Closing the AI accountability gap: defining an end-to-end framework for internal algorithmic auditing , 2020, FAT*.

[24]  Piotr Sapiezynski,et al.  Algorithms that "Don't See Color": Measuring Biases in Lookalike and Special Ad Audiences , 2019, AIES.

[25]  Piotr Sapiezynski,et al.  Ad Delivery Algorithms: The Hidden Arbiters of Political Messaging , 2019, WSDM.

[26]  Eric M. Schwartz,et al.  Native Advertising in Online News: Trade-Offs Among Clicks, Brand Recognition, and Website Trustworthiness , 2019, Journal of Marketing Research.

[27]  James A. Landay,et al.  Search Media and Elections , 2019, Proc. ACM Hum. Comput. Interact..

[28]  Brian W. Powers,et al.  Dissecting racial bias in an algorithm used to manage the health of populations , 2019, Science.

[29]  Catherine E. Tucker,et al.  Algorithmic Bias? An Empirical Study of Apparent Gender-Based Discrimination in the Display of STEM Career Ads , 2019, Manag. Sci..

[30]  Markus Luczak-Rösch,et al.  You can't see what you can't see: Experimental evidence for how much relevant information may be missed due to Google's Web search personalisation , 2019, SocInfo.

[31]  C. Lee Giles,et al.  Screenomics: A Framework to Capture and Analyze Personal Life Experiences and the Ways that Technology Shapes Them , 2019, Hum. Comput. Interact..

[32]  Inioluwa Deborah Raji,et al.  Actionable Auditing: Investigating the Impact of Publicly Naming Biased Performance Results of Commercial AI Products , 2019, AIES.

[33]  Tobias Meisen,et al.  Ablation Studies in Artificial Neural Networks , 2019, ArXiv.

[34]  David Lazer,et al.  Auditing Partisan Audience Bias within Google Search , 2018, Proc. ACM Hum. Comput. Interact..

[35]  Karrie Karahalios,et al.  Communicating Algorithmic Process in Online Behavioral Advertising , 2018, CHI.

[36]  Christo Wilson,et al.  Investigating the Impact of Gender on Rank in Resume Search Engines , 2018, CHI.

[37]  Bernard J. Jansen,et al.  Analyzing Advertising Labels: Testing Consumers' Recognition of Paid Content Online , 2018, CHI Extended Abstracts.

[38]  Aleksandra Korolova,et al.  Facebook's Advertising Platform: New Attack Vectors and the Need for Interventions , 2018, ArXiv.

[39]  Arunesh Mathur,et al.  An Empirical Study of Affiliate Marketing Disclosures on YouTube and Pinterest , 2018, arXiv.org.

[40]  Leland McInnes,et al.  UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction , 2018, ArXiv.

[41]  Timnit Gebru,et al.  Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification , 2018, FAT.

[42]  Krishna P. Gummadi,et al.  Potential for Discrimination in Online Targeted Advertising , 2018, FAT.

[43]  Pete Koomen,et al.  Peeking at A/B Tests: Why it matters, and what to do about it , 2017, KDD.

[44]  Leland McInnes,et al.  Accelerated Hierarchical Density Based Clustering , 2017, 2017 IEEE International Conference on Data Mining Workshops (ICDMW).

[45]  Michael Luca,et al.  Racial Discrimination in the Sharing Economy: Evidence from a Field Experiment , 2016 .

[46]  Christopher T. Lowenkamp,et al.  False Positives, False Negatives, and False Analyses: A Rejoinder to "Machine Bias: There's Software Used across the Country to Predict Future Criminals. and It's Biased against Blacks" , 2016 .

[47]  Alex S. Taylor,et al.  Re-Making Places: HCI, 'Community Building' and Change , 2016, CHI.

[48]  Christo Wilson,et al.  Peeking Beneath the Hood of Uber , 2015, Internet Measurement Conference.

[49]  Sean A. Munson,et al.  Unequal Representation and Gender Stereotypes in Image Search Results for Occupations , 2015, CHI.

[50]  Helen Nissenbaum,et al.  Trackmenot: Resisting Surveillance in Web Search , 2015 .

[51]  Michael Carl Tschantz,et al.  Automated Experiments on Ad Privacy Settings , 2014, Proc. Priv. Enhancing Technol..

[52]  Catherine Tucker,et al.  Social Networks, Personalized Advertising, and Privacy Controls , 2013 .

[53]  Marc L. Resnick,et al.  Ergonomics Society Annual Meeting , 2013 .

[54]  Latanya Sweeney,et al.  Discrimination in online ad delivery , 2013, CACM.

[55]  Rong Yan,et al.  Social influence in social advertising: evidence from field experiments , 2012, EC '12.

[56]  Ian R. Kerr,et al.  Lessons from the Identity Trail: Anonymity, Privacy and Identity in a Networked Society , 2009 .

[57]  Vanessa M. Buote,et al.  Comparing to perfection: How cultural norms for appearance affect social comparisons and self-image. , 2006, Body image.

[58]  C. Pollitt,et al.  Performance audit and evaluation: Similar tools, different relationships? , 1996 .

[59]  David Flint,et al.  Philosophy and Principles of Auditing: An Introduction , 1988 .

[60]  Albert B. Cherns,et al.  The Principles of Sociotechnical Design , 1976 .

[61]  A. Acquisti,et al.  Behavioral Advertising and Consumer Welfare: An Empirical Investigation , 2023, SSRN Electronic Journal.

[62]  Charlton D. McIlwain.  Algorithmic Discrimination: A Framework and Approach to Auditing & Measuring the Impact of Race-Targeted Digital Advertising , 2023, Social Science Research Network.

[63]  Karrie Karahalios,et al.  Auditing Algorithms: Understanding Algorithmic Systems from the Outside In , 2021, Found. Trends Hum. Comput. Interact..

[64]  Michelle A. Gan,et al.  An Image of Society: Gender and Racial Representation and Impact in Image Search Results for Occupations , 2021 .

[65]  G. F. M. S..  Gender Advertisements , 2020, Revista Latino-americana de Geografia e Genero.

[66]  Arvind Narayanan,et al.  Characterizing the Use of Browser-Based Blocking Extensions To Prevent Online Tracking , 2018, SOUPS @ USENIX Security Symposium.

[67]  Karrie Karahalios,et al.  Auditing Algorithms : Research Methods for Detecting Discrimination on Internet Platforms , 2014 .