Scholars who study deliberative democracy put a lot emphasis on the importance of political talk. The way people talk about politics shape and reflect how they think about politics as well as how they practice politics. Talking politics is also a discursive act, which is closely tied to the discussions of political engagement. There is, however, an important concern when it comes to the analysis of political talk that people talk about politics with like-minded individuals creating echo chambers. Their boundaries are heavily guarded by group dynamics. As the scholar who support echo chambers argument suggest deliberative talk is quite rare especially in overtly political pages. However, we can also find deliberative talk in non-political pages. The literature on deliberation has a tendency to talk about it as an outcome, focuses on exclusively political venues and relies heavily on research settings like forums, focus groups. In doing so it understudies the political talk that takes place in non-political settings. Consequently in this paper I focus on the question of how we come across political discourse on a predominantly non-political platform on Social Network Sites. In order to answer that question I focused on top 50 most popular public Facebook pages from Turkey. I collected Facebook comments between September-December 2012 by selecting five random days. I located the posts made on these select dates and downloaded all the comments made on these posts allowing for a week lag so that I did not cut an ongoing conversation. I downloaded 430 posts in total and coded over 80.000 comments. If the comments were political, I coded these comments for the presence or absence of the following characteristics: encouragement of deliberative talk, civility, outrage, and us/them distinctions. For the analysis of this data, I pursued a mixed methods approach. In order to demonstrate the relationships and general trends among and within the Facebook page types (entertainment, sports and political) I used multinomial logistic regression and made a comment-level analysis. I aimed to demonstrate the characteristics of comments in relation to whether the comment is encouraging, discouraging or neutral in terms of deliberative talk. This quantitative analysis is followed by a qualitative conversation analysis of the data. Based on these analyses, I argue that contrary to the expectations in literature we might see a different kind of deliberative talk in some seemingly non-political venues, a rowdy one on sports pages.