GreatBear, you might also ask bearfoil about the apparent period of communications loss. A flat spin does not account for it with an ACARS antenna's beamwidth. A full roll could accomplish it.
In this regard I have thought of something to try on for size and see if it fits. There is the comunications outage. Presume messages are mostly kept in a queue by order of arrival. Just how many such messages that were really important were left queued at the time of impact?
And I'm still not sure I see in bearfoil's scenario the applied forces required to stop the plane apparently very suddenly then having it drop. Could a violent pitch up cause this? If so what would tend to rotate the plane to roughly horizontal for the fall and not continue rotating it to a nose down attitude from which recovery is possible?