Social intelligence design (SID) is an essential concept for constructing electronic communities or computer-supported communication systems which effectively support various intelligent activities. This study is focused on one example of such systems, that is, online chat for real-time computer-mediated communication. We examined various dimensions of human information processing in online chat sessions from both the cognitive and social perspectives within psychology. A key variable which affects the process of a chat session is what we call communication congestion. We regard this quality as characterizing online chat sessions and are experimentally exploring its effects on chat performance and cognition processing by participants. Results of this experimental study suggest that medium-density congestion with a relevant topic might activate communication, particularly for experienced participants. Based on our findings, we discuss some suggestions regarding future innovations for chat systems.