> Wouldn't a long string of failed predictions be indicative of failure to understand...? No, the number of failures is completely meaningless, just as the num

> Wouldn't a long string of failed predictions be indicative of failure to understand...? No, the number of failures is completely meaningless, just as the number of successes is. The only thing that is relevant is the ratio of success vs. failure, somehow weighted appropriately based on the certainty claimed for the prediction and to correct for duplicated predictions.