Influencing network through partial incentives
ttps://arxiv.org/pdf/1401.7970.pdf
Further studying the network cascade effect from my previous blog post, I found this interesting paper about partial incentives, written by Erik D. Demaine. The topic of this paper is “How to influence people with partial incentives?” The topic seemed very relevant to what we are learning in class, because this idea of partial incentives can be applied to finding the most effective way of advertising certain products. In class we have studied effective search advertisements in Google or Amazon, and I felt that this paper explains a different perspective of why search advertisements are so successful, compare to other types of advertising.
Obsolete way of advertising involves influencing a few key nodes in a network, for example, providing free products to a limited number of influential people, and hoping this would cause a network cascade effect to reach as many as possible nodes in a network. If an advertiser only has enough budget to wholly influence one third of the network, it may be better to partially(⅓ in each node) to influence the entire network structure
The paper takes real network structure from Amazon and Facebook, and applies algorithms to test the effectiveness of network cascade, and proves that fractional influence of nodes is way more effective than integral influence of nodes.
Search advertisements are similar to this fractional influence model. Search advertisements are less costly compared to advertisements like free-product, and it can influence more people in the network. I think this is the reason why search advertisements are so attractive to many advertisers, and why companies like Google or Amazon gets huge amount of profit.