A comparative study of algorithmic–user classification practices in online dating: a human–machine learning process

ABSTRACT

Dating-app graphical user interface (GUI) structures for data collection contain categories that enable classifying users algorithmically and classifying users between each other to define their sexuality and find a date. Indeed, app providers define categories that mediate the users’ bodies and interactions to present themselves, and these categories ultimately serve app algorithms for recommending profiles. Categories establish a main reciprocated mediation between algorithms and users that is explored in this article to shed new light on the way Big Data shapes human–algorithmic interactions in online dating. However, online dating research pays little attention to classification processes from the perspective of the user, although classification is key to algorithm function for codifying sexuality. Using a qualitative analysis of 40 participant situated interviews, I examine the way dating-app users make sense of predefined categorical structures and their underlying classification processes, within 26 platforms. The results show that actors learn to integrate algorithmic logic into their common knowledge, as well as to challenge the algorithmic logic, and thus produce new conventions to classify their emotional states, physical attractiveness, and sexual preferences.

KEYWORDS:

• dating apps
• datafication
• classification
• matching algorithms

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 I extend Bechmann and Bowker’s (2019) work on classification practices on Facebook (now Meta). The authors limit their work to the developers’ perspective.

2 I developed this double perspective in Pidoux (2021), inspired by two main distinct works (Burrell 2016; Pasquinelli 2019).

3 I use sex to define sexual orientation because dating apps target users through male/female binary sex combinations (M–M, F–M, M–F, F–F or both F–M–F, M–F–M). Sex is a mandatory field on the registration page that predefines the categories available in the profile, despite the fact that some apps offer myriad gender identities at later stages.

4 Celibataire.ch is a dating website popular in French-speaking Switzerland and is very similar to OkCupid, one of the most popular dating apps worldwide (see downloads: https://wiki.personaldata.io/wiki/Item:Q3555). Like OkCupid, Celibataire is based on a quiz with a high number of questions to compute a score between two profiles. For the sake of familiarity, here will be given examples of OkCupid also used by this research participant sample.

5 It was particularly difficult to reach women who were willing to speak about their experiences with dating apps. This issue requires further study.

Additional information

Funding

This work was possible with the support of the Swiss National Science Foundation through grant Doc.CH No. 172379.