Social media in goblin mode: an NLP approach to post-COVID behavioral patterns

Abstract Goblin mode behavior, characterized by impulsive, indulgent, and self-gratifying actions in online environments, has become increasingly prevalent in the post-pandemic digital landscape. This study investigates the relationship between goblin mode behavior and users' emotional responses, focusing on anxiety, anger, and sadness. Utilizing a dataset of 35,926 social media comments, we applied Natural Language Processing (NLP) techniques to detect topics, emotions, and behavioral patterns. Regression analysis revealed that users exhibiting goblin mode behavior are more likely to express heightened levels of negative emotions, with direct goblin mode correlating significantly with anxiety and anger. Furthermore, emotionally charged topics such as news & social concern and diaries & daily life were found to trigger goblin mode tendencies, whereas professional topics showed a lower association. Our findings contribute to a deeper understanding of impulsive online behaviors and their emotional implications, providing insights for social media moderation strategies and digital well-being interventions.