How does social support coordinate agents in online communities?
The Role of Social Support and Influencers in Social Media Communities
October 8, 2024
https://arxiv.org/pdf/2410.04619This paper explores how social support (likes, shares, etc.) acts as a "currency" in social media communities, similar to money in a market economy. It proposes that social support guides content creators on what to produce and signals the perceived value of content. This mechanism also enables efficient coordination between community members and influencers, even with limited information.
The key points relevant to LLM-based multi-agent systems are:
- Social support as a coordination mechanism: The paper suggests that similar mechanisms of social feedback could be applied in other multi-agent environments, including LLM-based systems, to guide agents towards desired outcomes and efficient coordination.
- Role of "influencers": LLMs could function as "influencers" in multi-agent systems by aggregating and filtering information, helping other agents make more informed decisions. This highlights the potential for LLMs to act as central nodes for information flow and decision-making in complex multi-agent settings.
- Price of influence: The paper acknowledges that relying on LLMs as information proxies can lead to sub-optimal outcomes, but this "price of influence" decreases as the LLM's knowledge and understanding of the system grows. This points to the importance of training LLMs on comprehensive data to improve their effectiveness in multi-agent coordination.