Prediction markets seem useful in many decision making contexts where there are experts or crowds of people who knowledge, as opposed to large piles of machine-learning accessible data.
In many industries and across the world more generally, prediction markets are not well-known and they are treated with skepticism and a fog of illegality due to their association with gambling.
I think that launching more trendy prediction markets could help remedy these issues, e.g. royalwedding.metaculus.com
Another way to approach the goal of improved decision-making is using other mechanisms to improve predictions. We're trying a simpler variant at Empiricast. This is similar to Metaculus, but (currently) for internal use in organizations.
https://www.predictit.org/