Existing research on net-centric attacks has focused on the detection of attack events on network side and the removal of rogue programs from client side. However, such approaches largely overlook the way on how attack tools and unwanted programs are developed and distributed. Recent studies in underground economy reveal that suspicious attackers heavily utilize online social networks to form special interest groups and distribute malicious code. Consequently, examining social dynamics, as a novel way to complement existing research efforts, is imperative to systematically identify attackers and tactically cope with net-centric threats. In this paper, we seek a way to understand and analyze social dynamics relevant to net-centric attacks and propose a suite of measures called SocialImpact for systematically discovering and mining adversarial evidence. We also demonstrate the feasibility and applicability of our approach by implementing a proof-of-concept prototype Cassandra with a case study on real-world data archived from the Internet.