Various forms of decentralized solutions for social media development have been proposed in the last years to address some of the open problems of centralized ones, which include security issues, micro-targeting, opinion polarization, and various forms of information pollution. Regarding the latter issue in particular, some decentralized social media have implemented reward systems based on the blockchain technology with the stated purpose of incentivizing users to generate quality content, which can be seen as a simple mechanism of bottom-up content moderation in decentralized environments. However, users can still cheat the logic of such reward systems, to maximize their economic gain, regardless of content quality. In this paper, we focus on this problem by studying the actual relationship between the quality of User-Generated Content disseminated in the context of blockchain-based decentralized social media platforms, and the rewards provided to users by their reward systems. To this end, we analyze distinct sets of features characterizing quality content in social media, in association with distinct classifiers to identify content actually worthy of reward. By exploiting Steemit, one of the most well-known decentralized social media platforms based on blockchain, we construct a labeled dataset to both train and evaluate distinct configurations of the different feature sets in association with the considered classifiers. The performed evaluation shows the effectiveness of some of these configurations in identifying reward-worthy content, demonstrates that current reward systems are indeed unable through their mechanisms to promote actual quality content, and suggests the need for the development of further reward strategies allowing for better reward distribution and, ultimately, help pursue the original purpose of reward systems.
Mancino, D., Guidi, B., Michienzi, A., Viviani, M. (2025). Striking the Balance: Evaluating Content Quality and Reward Dynamics in Blockchain Online Social Media. IEEE ACCESS, 13, 21927-21945 [10.1109/ACCESS.2025.3536205].
Striking the Balance: Evaluating Content Quality and Reward Dynamics in Blockchain Online Social Media
Mancino D.;Viviani M.
2025
Abstract
Various forms of decentralized solutions for social media development have been proposed in the last years to address some of the open problems of centralized ones, which include security issues, micro-targeting, opinion polarization, and various forms of information pollution. Regarding the latter issue in particular, some decentralized social media have implemented reward systems based on the blockchain technology with the stated purpose of incentivizing users to generate quality content, which can be seen as a simple mechanism of bottom-up content moderation in decentralized environments. However, users can still cheat the logic of such reward systems, to maximize their economic gain, regardless of content quality. In this paper, we focus on this problem by studying the actual relationship between the quality of User-Generated Content disseminated in the context of blockchain-based decentralized social media platforms, and the rewards provided to users by their reward systems. To this end, we analyze distinct sets of features characterizing quality content in social media, in association with distinct classifiers to identify content actually worthy of reward. By exploiting Steemit, one of the most well-known decentralized social media platforms based on blockchain, we construct a labeled dataset to both train and evaluate distinct configurations of the different feature sets in association with the considered classifiers. The performed evaluation shows the effectiveness of some of these configurations in identifying reward-worthy content, demonstrates that current reward systems are indeed unable through their mechanisms to promote actual quality content, and suggests the need for the development of further reward strategies allowing for better reward distribution and, ultimately, help pursue the original purpose of reward systems.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.