So, you are teaching a math class with 51 knuckleheads and one bright student. You give them a tough problem. Is the mean answer of the 51 knuckleheads really going to better than the bright student's answer?
Another way to phrase this....does statistical analysis really show that an ensemble mean is better than the operational, and if so, at what time range? I recall the pros on this site saying that four days-out is the magic cutoff....is that based on analysis mythology?