Homophily and influence have been argued and tested to be the driving forces behind the consolidation of opinion cascades that cause divides and polarization. Existing research has shown little evidence to identify methods to either slow down or reverse the progression of polarization. The problem is a lack of attention to structural differences in opinion cascades in networked settings, while much focus was drawn to individual social psychology on decision-making mechanisms. This work proposes a new approach to combine the general model of opinion and network structure to see if the structural properties of agents (e.g., being in the position of a broker) influence the opinion cascades and hence cause polarization. The model constitutes multidimensional attributes that characterize nodes’ identities and opinions on issues. Agents update their opinions based on the other agents they have ties in the network, and they form ties with other agents based on their distance in opinion spaces. This work theorizes the introduction of three categories of brokers to incentivize depolarization: structural brokers, social brokers, and power brokers. Results demonstrate that brokers in general attenuate polarization but not reverse polarization. Among the three categories of proposed brokers, structural brokers perform the worst while social and power brokers perform better in certain contexts. The best-performing broker to depolarize is the optimal broker, a combination of all three fundamental categories of proposed brokers.