Decoding reddit memes virality

Abstract Understanding the elements that contribute to a meme’s virality is crucial due to the significant role memes play in digital culture and communication. This study analyzed 16,968 memes from Reddit, which are now available on the Open Science Framework. It aimed to identify predictors of virality, focusing on both textual and image-based characteristics. Our findings indicate that memes featuring multiple objects, humans, or animals, and those utilizing less saturated colors such as light gray and dark gray, tend to achieve higher virality. Additionally, smaller thumbnail dimensions and the avoidance of surprise facial expressions are associated with higher virality. Notably, memes posted between 12 pm and 6 pm UTC show a significant increase in virality. Combining these predictors enhances the area under the receiver operating characteristic curve of virality predictions, highlighting the complex interplay between content elements in meme dissemination. This research underscores the importance of a multifaceted approach in predicting meme virality and offers insights into effective social media strategies. Additionally, we provide an open-source codebase on GitHub to promote research transparency and reproducibility.