Societal Sorting as a Systemic Risk of Recommenders

Luke Thorburn; Maria Polukarov; Carmine Ventre

doi:10.1145/3640457.3688175

Societal Sorting as a Systemic Risk of Recommenders

Luke Thorburn, Maria Polukarov, Carmine Ventre

Research output: Chapter in Book/Report/Conference proceeding › Chapter › peer-review

Abstract

Political scientists distinguish between polarization (loosely, people moving further apart along a single dimension) and sorting (an increase in the probabilistic dependence between multiple dimensions of individual difference). Among other harms, sorting can increase the risk of conflict escalation by reinforcing us-and-them group identities and reducing the prevalence of cross-cutting affiliations. In this paper, we (i) review normative arguments for high or low sortedness, (ii) summarize the mechanisms by which sortedness can change, and (iii) show that under a simple model of social media recommender-driven preference change, personalized engagement-based ranking creates a systematic tendency towards sorting, while ranking by diverse engagement (sometimes called “bridging-based ranking”) mitigates this tendency. We conclude by considering the implications for those conducting systemic risk assessments of very large online platforms under the EU Digital Services Act.

Original language	English
Title of host publication	RecSys 2024 - Proceedings of the 18th ACM Conference on Recommender Systems
Pages	951 - 956
Number of pages	6
ISBN (Electronic)	9798400705052
DOIs	https://doi.org/10.1145/3640457.3688175
Publication status	Published - 8 Oct 2024

Publication series

Name	RecSys 2024 - Proceedings of the 18th ACM Conference on Recommender Systems

Access to Document

10.1145/3640457.3688175

Cite this

@inbook{193cac9e18904f3c86ae542ec82701de,

title = "Societal Sorting as a Systemic Risk of Recommenders",

abstract = "Political scientists distinguish between polarization (loosely, people moving further apart along a single dimension) and sorting (an increase in the probabilistic dependence between multiple dimensions of individual difference). Among other harms, sorting can increase the risk of conflict escalation by reinforcing us-and-them group identities and reducing the prevalence of cross-cutting affiliations. In this paper, we (i) review normative arguments for high or low sortedness, (ii) summarize the mechanisms by which sortedness can change, and (iii) show that under a simple model of social media recommender-driven preference change, personalized engagement-based ranking creates a systematic tendency towards sorting, while ranking by diverse engagement (sometimes called “bridging-based ranking”) mitigates this tendency. We conclude by considering the implications for those conducting systemic risk assessments of very large online platforms under the EU Digital Services Act.",

author = "Luke Thorburn and Maria Polukarov and Carmine Ventre",

note = "Publisher Copyright: {\textcopyright} 2024 Copyright held by the owner/author(s).",

year = "2024",

month = oct,

day = "8",

doi = "10.1145/3640457.3688175",

language = "English",

series = "RecSys 2024 - Proceedings of the 18th ACM Conference on Recommender Systems",

pages = "951 -- 956",

booktitle = "RecSys 2024 - Proceedings of the 18th ACM Conference on Recommender Systems",

}

TY - CHAP

T1 - Societal Sorting as a Systemic Risk of Recommenders

AU - Thorburn, Luke

AU - Polukarov, Maria

AU - Ventre, Carmine

PY - 2024/10/8

Y1 - 2024/10/8

N2 - Political scientists distinguish between polarization (loosely, people moving further apart along a single dimension) and sorting (an increase in the probabilistic dependence between multiple dimensions of individual difference). Among other harms, sorting can increase the risk of conflict escalation by reinforcing us-and-them group identities and reducing the prevalence of cross-cutting affiliations. In this paper, we (i) review normative arguments for high or low sortedness, (ii) summarize the mechanisms by which sortedness can change, and (iii) show that under a simple model of social media recommender-driven preference change, personalized engagement-based ranking creates a systematic tendency towards sorting, while ranking by diverse engagement (sometimes called “bridging-based ranking”) mitigates this tendency. We conclude by considering the implications for those conducting systemic risk assessments of very large online platforms under the EU Digital Services Act.

AB - Political scientists distinguish between polarization (loosely, people moving further apart along a single dimension) and sorting (an increase in the probabilistic dependence between multiple dimensions of individual difference). Among other harms, sorting can increase the risk of conflict escalation by reinforcing us-and-them group identities and reducing the prevalence of cross-cutting affiliations. In this paper, we (i) review normative arguments for high or low sortedness, (ii) summarize the mechanisms by which sortedness can change, and (iii) show that under a simple model of social media recommender-driven preference change, personalized engagement-based ranking creates a systematic tendency towards sorting, while ranking by diverse engagement (sometimes called “bridging-based ranking”) mitigates this tendency. We conclude by considering the implications for those conducting systemic risk assessments of very large online platforms under the EU Digital Services Act.

UR - http://www.scopus.com/inward/record.url?scp=85210525990&partnerID=8YFLogxK

U2 - 10.1145/3640457.3688175

DO - 10.1145/3640457.3688175

M3 - Chapter

T3 - RecSys 2024 - Proceedings of the 18th ACM Conference on Recommender Systems

SP - 951

EP - 956

BT - RecSys 2024 - Proceedings of the 18th ACM Conference on Recommender Systems

ER -

Societal Sorting as a Systemic Risk of Recommenders

Abstract

Publication series

Access to Document

Other files and links

Fingerprint

Cite this