Algorithms as Cultural Objects. Exploring Topic Modeling to Investigate How People Tweet About Algorithms

The current research interest around algorithms is not limited to unveiling their technical premises and effects, as scholars increasingly recognize how algorithms are relevant to society also as part of discourse, practices and experiences. This paper approaches algorithms as cultural objects, investigating people's spontaneous associations to the broad notion of 'algorithms', as expressed in a dataset of tweets. Combining topic modeling with qualitative content analysis, we answer two questions: 1) What do individuals observe when making sense of algorithms? and 2) How do people position themselves in relation to algorithms? Our analysis offers a descriptive account of the rich sense-making activity behind algorithmic imaginaries, as well as a typology of how people stand in relation to algorithms. Following a computational hermeneutics approach, our research contributes to the developing debate on algorithms 'from the bottom up', complementing existing studies that usually focus on the technicalities of specific algorithms (rather than on broader cultural understandings) and that tend to adopt elicited methods (rather than non-Intrusive ones). Our empirically-grounded categorization of what people observe and how do they stand in relation to algorithms can serve as a framework for future research interested in cross-platform, cross-cultural and / or cross-sectional replications.