When Synthetic Actors Suggest: Interaction Order of Recommender Systems on Social Media Platforms
(2020) SOCM04 20201Sociology
Department of Sociology
- Abstract
- The prevalence of algorithms and machine learning technologies in society opens up a vast sociological territory. In this thesis, I investigate interactions that transpire between users and recommender systems which statistically predicts and suggests specific items based on their and similar profiles’ data traces in the context of social media platforms in everyday life. Fusing several data collection methods with 16 participants, my analysis is based on rich qualitative data on 138 different instances of prediction. My theoretical framework draws on Ervin Goffman’s interaction order as interpreted through Karin Knorr Cetina’s theory of synthetic actors. I also present conceptual developments in my analysis of the symbolic interactions... (More)
- The prevalence of algorithms and machine learning technologies in society opens up a vast sociological territory. In this thesis, I investigate interactions that transpire between users and recommender systems which statistically predicts and suggests specific items based on their and similar profiles’ data traces in the context of social media platforms in everyday life. Fusing several data collection methods with 16 participants, my analysis is based on rich qualitative data on 138 different instances of prediction. My theoretical framework draws on Ervin Goffman’s interaction order as interpreted through Karin Knorr Cetina’s theory of synthetic actors. I also present conceptual developments in my analysis of the symbolic interactions with algorithmic counterparts. The work of recommender systems is found putting symbolic objects into circulation which effectively frames the interaction. This brings forth alignment dynamics which can enroll users into micro-alliances or act as a negative reference point. Drawing on a water metaphor to enrich patterns inductively drawn from the data, users are found to dive into the micro-context of a suggested object, float together with a stream of objects, or soar from the platform context while bringing traces of the object, a symbolic surplus, with them. Moreover, they swim under specific suggestions to change the direction of the talk while remaining in the interaction, swim away from the platform context, or get up from the water and cease their digital activity. My findings contribute to the literature on algorithms by showing that interactions with recommender systems is immersive yet slippery due to users’ reflexivity, leading to several combinations in alignment dynamics and transpiring moves, which is important in further assessing the algorithmic grasp of meaning-making currently under debate. (Less)
- Popular Abstract
- On most of the popular social media apps, there are systems in place that monitor users’ activities to provide calculated, personalized content suggestions. These systems are becoming increasingly effectful and make up a layer of activity that departs from users’ communication with friends, relatives and alike. In this study, I show patterns from instances when users and such systems come together based on evidence from 16 different young adult users who provided visual reports, wrote and spoke about their experiences. I find that these systems are providing continuous texture to users’ daily experiences. While these systems incite users’ thoughts and behaviors, it can head into directions afar from what the purpose of the recommendation... (More)
- On most of the popular social media apps, there are systems in place that monitor users’ activities to provide calculated, personalized content suggestions. These systems are becoming increasingly effectful and make up a layer of activity that departs from users’ communication with friends, relatives and alike. In this study, I show patterns from instances when users and such systems come together based on evidence from 16 different young adult users who provided visual reports, wrote and spoke about their experiences. I find that these systems are providing continuous texture to users’ daily experiences. While these systems incite users’ thoughts and behaviors, it can head into directions afar from what the purpose of the recommendation seemed to be, since users also paint the recommendation with their own colors. I present proof on users’ diving into the universe of a specific recommendation, float along with series of recommendations which platforms bring forward, and other related moves. I identify variations in the interactional accomplishments since users can dive into items they enjoy as well as into those they mostly get upset about and so on. Moreover, I consider recommendations which users disregard to sometimes reinforce users’ already acquired opinions. This is important to recognize since it shows how users and the systems in social media platforms come together in ways that can affirm and intensify held beliefs on important topics, whereas calculative systems sometimes drive this process further. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9028044
- author
- Engdahl, Isak LU
- supervisor
- organization
- course
- SOCM04 20201
- year
- 2020
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- sociology of algorithms, algorithmic decision-making, surveillance, digital sociology, machine learning, recommender systems, sociology of technology
- language
- English
- id
- 9028044
- date added to LUP
- 2020-09-03 11:48:10
- date last changed
- 2020-09-03 11:48:10
@misc{9028044,
abstract = {{The prevalence of algorithms and machine learning technologies in society opens up a vast sociological territory. In this thesis, I investigate interactions that transpire between users and recommender systems which statistically predicts and suggests specific items based on their and similar profiles’ data traces in the context of social media platforms in everyday life. Fusing several data collection methods with 16 participants, my analysis is based on rich qualitative data on 138 different instances of prediction. My theoretical framework draws on Ervin Goffman’s interaction order as interpreted through Karin Knorr Cetina’s theory of synthetic actors. I also present conceptual developments in my analysis of the symbolic interactions with algorithmic counterparts. The work of recommender systems is found putting symbolic objects into circulation which effectively frames the interaction. This brings forth alignment dynamics which can enroll users into micro-alliances or act as a negative reference point. Drawing on a water metaphor to enrich patterns inductively drawn from the data, users are found to dive into the micro-context of a suggested object, float together with a stream of objects, or soar from the platform context while bringing traces of the object, a symbolic surplus, with them. Moreover, they swim under specific suggestions to change the direction of the talk while remaining in the interaction, swim away from the platform context, or get up from the water and cease their digital activity. My findings contribute to the literature on algorithms by showing that interactions with recommender systems is immersive yet slippery due to users’ reflexivity, leading to several combinations in alignment dynamics and transpiring moves, which is important in further assessing the algorithmic grasp of meaning-making currently under debate.}},
author = {{Engdahl, Isak}},
language = {{eng}},
note = {{Student Paper}},
title = {{When Synthetic Actors Suggest: Interaction Order of Recommender Systems on Social Media Platforms}},
year = {{2020}},
}