There are many empirical studies on Online Social Networks (OSNs). While some studies on the growth of the system, few research on the decline in user population. In this paper, we focus on an Online Social Network (OSN), namely ”kaixin-net” (www.kaixin001.com) which is one of the biggest OSNs in China. Based on studying the most popular game ”Buying-a-house,” we investigate user behaviors. First, we made a model for ”Buying-ahouse,” then using the dataset that we crawled and traced (we crawled about one hundred and fifty user's profiles and traced forty users for about twenty days), we draw the following findings: (1) People who just join the game have much interest in this OSN, actively interact with his friends. (2) The interest goes down according to the time lapse for most of users, but there are also a few players whose interests never go down and enjoy this game very much. These can be interpreted as the population of kaixin-net is growing at the beginning state, but it will decline after the peak comes. So the organizers should think about how to slow down the peak's coming.