Does Reddit have an astroturfing problem? What the data suggests
Summary
A detailed write-up by Peter Vijeh analyzes whether Reddit knife-subreddits exhibit astroturfing, using a data-driven approach with a named-entity model (GLiNER) to trace brand mentions. The study compares tail accounts that heavily mention specific brands to random reassignments to assess whether buying-thread recommendations are disproportionately coming from paid or highly aligned accounts, and reports an 11.3% tail share versus a 7.9% baseline in buying threads.