Baseline trust, which refers to the personality aspect of trust and varies with different individuals, is essential for understanding the development of trust and cooperation in a team. At the same time, informal, non-work-related conversations (aka, cheap talk) have positive influences on the diffusion of trust and cooperation in global software engineering (GSE) practice. This paper seeks to develop an understanding of the influences of individuals' baseline trust on the diffusion of trust and cooperation, in the presence of cheap talk over the Internet. We employ a novel approach, designing a virtual experiment that integrates abstract agent-based modeling and simulation with realistic, empirical network structures and baseline trust data from two large open source projects (Apache LUCENE and Google CHROMIUM OS). The results highlight the significant impact of baseline trust on the diffusion of trust and cooperation, for instance, the emergence of non-traditional diffusion trajectories. The results also demonstrate that proper seeding strategies can improve the effectiveness and efficiency of diffusion of trust and cooperation.