Reddit has developed into a significant platform for political discussion among Millennials. In this exploratory study, we examine subscription trends on three political sub-forums on Reddit during the 2016 US presidential elections: /The_Donald, /SandersForPresident, and /HillaryClinton. As a theoretical framework, we draw from work on online communities’ group identity and cohesion. Concretely, we investigate how subscription dynamics relate to positive, negative and neutral news events occurring during the election cycle. We classify news events using a sentiment analysis of event-related news headlines. We observe that users who supported Sanders displayed no consolidation of support for Clinton after she won the Democratic Party’s presidential nomination. Secondly, we show that negative news events affected Sanders and Clintons subscription trends negatively, while showing no effect for Donald Trump. This gives empirical credence to Trump’s controversial claim that he could “stand in the middle of 5th Avenue and shoot somebody and not lose any voters”. We offer a number of explanations for the observed phenomena: the nature of the content of the three subreddits, their cultural dynamics, and changing dynamics of partisanship. We posit that the ‘death of expertise’ expresses itself on Reddit as a switch in persuasion tactics from a policy-based to an emotions-based approach, and that group members’ agreement on policy proved a weak marker for online communities’ group identity and cohesion. We also claim that strong partisanship coupled with weak party affiliation among Millennials contributed to the low levels of Democratic support consolidation after Clinton won the nomination.