Mass media used to quantify public opinion through poll survey; nowadays, Weibo Hot Search has increasingly been considered as the barometer of public opinion. Drawing on the public opinion infrastructures model, this article uses the methods of text analysis, interview, and content analysis to investigate how Hot Search defines and measures public opinion and how it is framed and interpreted by the mass media. We find that Hot Search had combined “the aggregation of scattered individuals’ search traces” and “the engagement and communication between the publics” since the adjustment of the trending rules in 2018. While the former is individual, private, and egalitarian, the latter is relational, public, and hierarchical. This signifies a shift in logic from “one person, one vote” to “wise public opinion prioritized”. The media frames Hot Search as the synonym of public opinion, with the terms “rocketed to” and “top one” most frequently used as the benchmarks to evaluate the public response, thus legitimizing the knowledge logic of trending algorithms. This finding contributes to revising the previous conclusion, which held that “the party media held a critical attitude towards algorithms”. The future research could take more concrete contexts where “people and algorithms meet” into account.