You might be interested in this talk from Dan Schwarz (current Metaculus CTO, former Google employee) on "Prediction Markets at Google, and Lessons in Corporate Forecasting":
It gives a reasonably thorough history of prediction markets at Google, and some insight into the use of forecasting at Google in his time there. Probably my main takeaway from watching was the belief in the importance of forecasting questions resolving quickly (the section at 32:17) in producing high quality forecasters (who can learn from the fast feedback). It sounds very plausible to me but I'm not aware of anything in the forecasting literature that conclusively shows it to be the case.
You might be interested in this talk from Dan Schwarz (current Metaculus CTO, former Google employee) on "Prediction Markets at Google, and Lessons in Corporate Forecasting":
It gives a reasonably thorough history of prediction markets at Google, and some insight into the use of forecasting at Google in his time there. Probably my main takeaway from watching was the belief in the importance of forecasting questions resolving quickly (the section at 32:17) in producing high quality forecasters (who can learn from the fast feedback). It sounds very plausible to me but I'm not aware of anything in the forecasting literature that conclusively shows it to be the case.