We’re currently running online experiments to try out more effective formats for group forecasting. We’ve found that forecasters working together has generated more accurate predictions, by effectively sharing relevant information and breaking down questions into sub-questions, as well as revealing more insight into the reasoning behind the forecasting.
As part of these experiments we’ve developed a new, standardized format for online communities. An example output from one of our 1.5h sessions with 10 attendees can be found here. We’re currently working to make this format more scalable and widely useful, as well as trying out other experimental formats.