Finding high signal people – applying PageRank to Twitter (thefourierproject.org)

A recent analysis applies the PageRank algorithm to Twitter to identify influencers within the AI/ML community, allowing for a more nuanced understanding of online reputation beyond mere follower counts. This method posits that powerful figures like Demis Hassabis and Ilya Sutskever signal "high-signal" individuals when they follow others, suggesting a network of informed voices. By calculating the "influence" of Twitter accounts based on their connections, the analysis aims to surface underrated but impactful contributors in the field. They introduce a "discovery score," where a high score is indicative of lower rank and fewer followers, thus emphasizing the importance of content quality over popularity. This approach carries significant implications for the AI/ML community, as it challenges traditional metrics of social media influence. It underscores the notion that high-profile figures often can't express themselves candidly due to their public personas, which can dampen their "signal" quality. Additionally, the use of a large language model to evaluate tweet content provides a cutting-edge tool for assessing the relevance of contributions, promoting a more informed dialogue. By sorting accounts by rank, discovery, and signal, the methodology encourages a deeper exploration of ideas within the AI landscape, helping users connect with thought leaders who might otherwise go unnoticed.