To better reflect actual human interactions in social network models, the authors take a bottom-up, agent-based modeling and network-oriented simulation approach to analyzing acquaintance network evolution based on local interaction rules. Resources and remembering are considered in addition to common friends, meeting by chance, and leaving and arriving. Based on these factors, friendships that have been established and built up can be strengthened, weakened, or broken up. Using computer simulations, Results from a series of experiments indicate that (a) network topology statistics, especially average degree of nodes, are irrelevant to parametric distributions because they rely on average values for initial parameters; and (b) resource, remembering, and initial friendship all raise the average number of friends and lower both degree of clustering and separation. These findings indicate a strong need for a bottom-up, agent-based modeling and network-oriented simulation approach to social network research, one that stresses interactive rules and experimental simulations.