Behavioral Strategies in Online Forums with Different Feedback Types

Forums for deliberation, i.e., coming to decisions and solutions for community problems, play an important role in public life. Methods for evaluating the content of such forums are becoming more and more significant. A promising direction is to introduce feedback options of different types, i.e., participants can assess a post by someone else according to different criteria, such as agreement/disagreement, originality or relevance for the topic currently discussed. However, participants in such forums may have specific interests such as earning themselves a high weight or earning high scores for 'their' arguments. Feedback options of different types now give rise to new kinds of strategic behavior, to back up one's specific interests and to push one's opinion. For instance, a participant could label an argument that he does not like as 'not original', in order to bog it down. To study this kind of behavior, we have built a respective game-theoretic model. The model incorporates an evaluation scheme, as follows: (1) users are assigned weights based on criteria such as originality of their posts according to feedback by others; (2) posts are scored based on the rate of agreement/disagreement feedback they have obtained and the weights of raters and authors. The model lets us study the following questions: When exactly does untruthful rating behavior pay off, and is truthful behavior an equilibrium strategy? A core insight is that the strategy 'rate posts always truthfully' is an equilibrium strategy. Further, our evaluation scheme is robust towards untruthful behavior of participants in many cases.