Good idea Brian. Better than impractical idea of broadcasting it for ground storage.
Why is that impractical?
Data rates these days are far in excess of what is required, and storage costs absolute peanuts.
Not only could it be used for accident investigation but in these days of big data crunching with machine learning and artificial intelligence, that data could be used for failure prediction. That sort of technology is already used in other operational technologies. In the long run it could pay for itself if predictive maintenance prevents "unplanned events".
And imagine if all of those sequences of events that lead to aircraft loss are combined and analyzed, and pilots could be warned that a seemingly harmless alarm taken in context of other data indicated that a catastrophic failure was in the offing?
Added: And automatically gave him a solution?
https://analyticsindiamag.com/how-to...-data-science/