The Internet, smartphones, Social Networking Service, and other IT goods have improved the overall quality of life, but created an over-connectedness with extremely low transaction costs in our society, amplifying latent social problems. In light of this, we demonstrate the main mechanism of information cascades in various network topologies using computational model that can consider autonomous agents, the adoption of others’ decisions, and network topologies. Our findings reveal that: (1) lower transaction costs may amplify the occurrence of information cascades; (2) the network structure significantly affects the behavior of traders in terms of the individual and the whole market; and (3) highly spread trend-shift cascades can be observed in scale-free networks when the influence of a dominant agent’s decision significantly affects the connected agents. Such findings highlight how a highly over-connected network has its critical shortcomings.