Social Networks and Cascade Behavior
Suppose you and your friends were snapchat users and a new application came to the tech market. Let’s call the new app TikTok. your friends start using the TikTok, then you change your entail opinion and begin using the new app and after a while you see a lot more other people join the app and so on. Here you change your behavior and choice based on inferences you make by observing your friends. Information cascading plays a vital role to lead you copying others even if it leads to a wrong outcome.
Information cascading is a key strategy for getting success and faster growth in a business. Organizations use cascade communication to disseminate information to people at all departments and employees. Imagining a large company with many branches. Finance department reports the company profit and expenses for a month. Decision makers in the main branch of the company came up with a final decision to bring changes in the manufacturing department and use another method of manufacturing to decrease costs of products x, y, z. if the manufacturing department does not act immediately to use the new method, the company will harm and face bankruptcy. Moreover, the employees are too busy to check their emails regularly or some of the employees are having another issue of not having access to their emails for that time. So, to ensure that crucial information from leadership reaches all these employees on time, information cascading is a very effective strategy to use and prevent bankruptcy. The decision can reach through cascading communication first to managers in each branch and then via direct reports to every individual in the manufacturing department.
Furthermore, most of the organizations use a cascading strategy in their marketing department to bring more customers and users. In online social networks, many companies have focused on information cascading in their marketing policy to maximize the number of influential users. Some have designed cascading models and algorithms which scales beyond million-sized graphs to increase in influence spread. In the table below, I have shown a summary of the quantitative description of the cascading information in a decision. H tell you to accept with high signal and L tell you to reject with low signal. It means when the correct decision is to accept with the probability P[H|A], you will see an H, and when the correct decision is to reject with probability of P[L|R], you will probably see a L signal. If you want to show this likelihood by q, then q > 0.5.
| Agent signal | True probability state | |
| Reject | Accept | |
| L | q | 1-q |
| H | 1-q | q |
Sources:
Influence maximization in social networks under an independent cascade-based model.
https://doi.org/10.1016/j.physa.2015.10.020
Scalable influence maximization for independent cascade model in large-scale social networks.
https://doi.org/10.1007/s10618-012-0262-1