Effectiveness of information retraction

In this work, we study the effectiveness of information retraction in situations where information being spread requires recipients to make a decision or take an action. Consider the scenario where information is introduced into a network, advising recipients to take an action. If at a later time, the information is found to be inaccurate and the action is unnecessary, it becomes a concern to cease the information from spreading any further and stop people from taking the action. The spread of inaccurate information can lead to confusion and mistrust, and therefore it is important to be able to quickly impede or retract inaccurate information, if needed to at a later time. We investigate the idea of introducing counter messages into a network to interfere with an ongoing diffusion and stop the action that was prescribed by the previous messages. These counter messages are diffusive themselves and may spread through the network based on the recipient´s evaluation of the information. We present an empirical framework for modeling the spread of actionable information and information retraction. Using the framework, we perform preliminary experiments to investigate strategies for broadcasting the counter message, in particular, how to identify individuals that should receive the counter message directly from the information source. There is a trade off between a fast effective spread of actionable information and the ability to retract the information. Findings also suggest that alternate strategies will have to be explored to incorporate group structures and the distribution of trust in designing a useful abort mechanism.