>I recently heard a history professor state that when events happen, everything should be documented and first-hand accounts taped by all who were there or had knowledge of it. Then it should sit for 20 years and after that time, all of the information gathered together and looked at again with the perspective of time and other events.
That's the ideal case, which almost never happens. The data are coming usually from the reports of those who had to report to their superiors - they are the poor guys who have to do some writing whenever it hits the fan, and their first worry is to save their own lower backs, then to satisfy the form, and the truth is somewhere lower in the list of priorities. They are also too close to events to grasp their importance, and often omit the really important stuff.
But a lot can be dug out of their slips :).