Predicting Custom-Feed Returns for New Bluesky Posts: A Prospective Study

This research introduces a novel 'cold-start routing' task within the context of social media custom feeds, specifically using Bluesky posts. It focuses on predicting which independent custom feeds will subsequently return newly published public posts. The study proposes a method for ranking feeds based on their likelihood of featuring new content and is building a benchmark dataset for this purpose, covering 17.8 million posts across 5,000 monitored feeds.

Why it matters

This research is strategically important as it addresses a fundamental challenge in content discovery and distribution on decentralized or highly customizable platforms. Effective prediction of content dissemination pathways can inform strategies for content creators, platform operators, and advertisers, enabling more targeted outreach and improved user experience. Understanding these dynamics can also influence the design of future social media architectures.

Key insights

• Traditional cold-start recommendation typically addresses new users or items, but this research redefines it for newly published posts and independent custom feeds.
• The study introduces a 'cold-start routing' task where new public posts are queries and rankable feeds are candidates.
• Feeds are ranked based on the probability of them subsequently returning a new post.
• A benchmark dataset is being developed to support this research, which includes 17.804 million public posts and 1.865 million observable post-feed returns from 5,000 feeds.
• The dataset is characterized as 'collect-first, label-later' and is still evolving.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2608.13874