Abstract: Under the background of Web 2.0, networks socialization generates collective intelligence which can enrich human beings wisdom. However, what is the main factor that influences the performance of this behavior is still in research. In this study, the effect of number of Internet users that is represented by quantity, quality and variety of User-generated Content (UGC) is brought forward. Regarding Wikipedia as a study case, this study uses Agent-based modeling methodology and real data of Wikipedia for about 10 years to establish and simulate the model. The results verify that the size of group is indeed a necessary condition to generate collective intelligence. When the number of participants in Wikipedia reaches about 400000, the quantity of UGC increases exponentially, the quality of UGC reaches a satisfactory level and the variety of UGC can be guaranteed. This insight gives significance to show when mass collaboration will lead to collective intelligence which is an innovation than before.