In this study, we explore the dynamics of conversations on Flashback, a Swedish online forum that has more than 1.5M registered users as of February 2023 and that claims to ensure freedom of speech and users’ anonymity. Specifically, we focus on the "Immigration and integration" [Integration och invandring] subforum, where a large part of immigration-related discussions takes place. Flashback’s case is especially interesting since previous research has demonstrated that biased and Islamophobic narratives are quite common on the platform (Kaati et al. 2022; Törnberg and Törnberg 2016). Using the dataset of 270k messages from the forum spanning the last election cycle in Sweden (2018-2022), we focus on the dissemination of links leading to right-wing alternative media as one of the potential drivers for the conversations on this subforum. In particular, our goal is to investigate the dissemination of such content on the subforum and to focus on the impact that such links (and, possibly, content in general) have on the dynamics of conversations and the users’ evaluations of the immigration issue in Sweden. Namely, using a combination of text and network analysis methods, we do the following: a) explore the differences in user stances and structure of the conversations depending on the dominant type of content shared, and b) study the evolution of users’ emotional states and subjective evaluations of the immigration issue on the macro-(forum) and meso- (conversation) levels.