Understanding who talks about what: comparison between the information treatment in traditional media and online discussions

www.nature.com/scientificreports OPEN Understanding who talks about what: comparison between the information treatment in traditional media and online discussions Hendrik Schawe 1, Mariano G. Beiró 2,3, J. Ignacio Alvarez‑Hamelin 2,3, Dimitris Kotzinos 4 & Laura Hernández 1* We study the dynamics of interactions between a traditional medium, the New York Times journal, and its followers in Twitter, using a massive dataset. It consists of the metadata of the articles published by the journal during the first year of the COVID-19 pandemic, and the posts published in Twitter by a large set of followers of the @nytimes account along with those published by a set of followers of several other media of different kind. The dynamics of discussions held in Twitter by exclusive followers of a medium show a strong dependence on the medium they follow: the followers of @FoxNews show the highest similarity to each other and a strong differentiation of interests with the general group. Our results also reveal the difference in the attention payed to U.S. presidential elections by the journal and by its followers, and show that the topic related to the “Black Lives Matter” movement started in Twitter, and was addressed later by the journal. The debate about the influence of mass media on social opinion has shown peaks of interest each time that a technological breakthrough modified the media ecosystem, mainly by increasing the amount of people that can be reached by broadcasters1. The first important one, the invention of the printing press by Gutenberg, has indeed played an important role in the rapid expansion of Calvinism in Europe2, although its general influence on the formation of social opinion was mitigated by the fact that most of the population was illiterate. Later, around the beginning of the 20th century, when the wireless radio transmissions appeared and rapidly became a popular entertaining medium, discussions about the foreseeable consequences of the popularization of this new medium were carried in the written press, which by that time had become a traditional one. A review in the New York Times from May 7th 1899 entitled “Future of Wireless Telegraphy” warned: “All the nations of the earth would be put upon terms of intimacy and men would be stunned by the tremendous volume of news and information that would ceaselessly pour in upon them”3. Needless to say that the same kind of debates took place at the arrival of TV b roadcasting4. The rapid growth of digital media certainly triggered again the same kind of discussions but this time, with a major difference: the massive data accumulated on social media platforms allows us to perform measurements about the opinion evolution of large amounts of people. A countless number of articles have addressed different aspects of opinion dynamics based on social networks. A few recent ones are the study of opinion evolution on different selected t opics5,6, and the characterisation of structural properties of the interaction networks that result from the different functionalities offered by the platforms (like mentions, retweets, follower-friend in Twitter)7,8. In particular, there is a recent interest on the formation of information bubbles and echo chambers—strongly connected clusters of people that communicate only weakly with others9–11. Special attention has also been given to the diffusion of rumours and fake news in relation to the COVID-19 p andemic12, to the extent that the term infodemics was coined to highlight the parallelism with the diffusion of the v irus13–15. 1 Laboratoire de Physique Théorique et Modélisation, UMR‑8089 CNRS, CY Cergy Paris Université, 95300 Paris, France. 2Universidad de Buenos Aires, Facultad de Ingeniería, Paseo Colón 850, C1063ACV Buenos Aires, Argentina. 3CONICET, Universidad de Buenos Aires, INTECIN, Paseo Colón 850, C1063ACV Buenos Aires, Argentina. 4ETIS UMR 8051 CY Cergy Paris Université, ENSEA, CNRS, 95300 Paris, France. *email: Scientific Reports | (2023) 13:3809 | https://doi.org/10.1038/s41598-023-30367-8 1 Vol.:(0123456789) www.nature.com/scientificreports/ Nowadays, it seems clear that if media exert an influence on social opinion it is mainly by setting the terms of debate or, in the words of B. C ohen16, the press may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about. This notion is known as the Agenda Setting Problem17 and is a long-lasting subject of discussion in Political Sciences, Communication, Social Psychology , Cognitive Sciences, and Media studies. In particular, an open debate concerns the relationship between the notion of issues –the subjects that are addressed– and that of frames –the attributes assigned to these subjects when they are addressed–18–20. In this work we investigate the agenda setting problem, by studying the dynamics of the different topics treated by a traditional medium, The New York Times (NYT) journal, and their relationship with the dynamics of the public discussions that take place in Twitter among its followers. Here, the term topic designates the subjects treated in both media without attempting to differentiate between issues and frames. This is the standard meaning given in textual corpora analysis which has also been used to address the agenda setting problem21,22. We center our study in the first year of the COVID-19 pandemics, which by its very nature can be expected to become an important driver of public attention. Several works studied the evolution of the opinion in Twitter (and other platforms) during this period, mainly focusing on discussions directly related to health issues, or public policies related with t hem23–25. Here, on the contrary, we aim at understanding how the different topics that interested the society during this period were addressed both by the media and by the public that is in direct relation to them, without assuming a priory the existence of any influence on either direction. While some recent studies have compared how traditional media and social networks treat a particular topic of discussion26–30, in this work we search for global patterns characterizing each of them. We have collected a large amount of tweets corresponding to a randomized sample of the over 46M followers of the New York Times (NYT) official Twitter account (@nytimes), during the first year of the pandemic, along with the metadata of the articles published by the journal during that period. This sampling guarantees that we are reaching the topics discussed by users that have expressed an interest in that journal by following its Twitter account. In order to compare with the behaviour of the followers of different media, we have also collected a sample of the tweets published by the followers of other important media of different kinds: written press, radio, television, press agencies. With this data, we build a semantic network repres (...truncated)