> (if they have a preference over which action will be taken) are incentivised to give more extreme answers.
With yes and no answers, how can answers become more extreme? If you are asked "Is this misleading? Yes/No" and it's only marked as misleading when "Yes" is the majority, then you are incentivized to answer with your true opinion. If you want the post to be marked misleading, then answering yes increases the chance that it is marked as such.
The bayesian truth sereum information score makes sense when you are trying to reward people for truthfully answering your questions; for example by paying them [0]. When asking "Is this misleading?", how do you use the information score to compute who won?
[0] - http://www.eecs.harvard.edu/cs286r/courses/fall10/papers/DW0...