The "problem", as I see it is to cover areas where there is a risk of being unable to retrieve the recorders, an extremely scant probability. Coverage could be extremely limited to those regions that are historically difficult for data transmission. 447 was effectively out of range, an "acceptable" parameter that needn't occur at all, in my opinion. All flights over land, and most of the Oceans, present high probabilities of retrieval of the recorders.
That ACARS was successful (and disseminated), is a case in point. Simply because the data encapsulated in the company channel speaks to a possibility of capturing accident information, that would seem to be a starting point.
Four minutes of the final entry into this a/c "diary" was directed to maintenance issues. Fine. Even this information very nearly solves the riddle without the flight recorders. Linking this channel's ennable via transponder mayday code could disable the maintenance flow, and substitute flight controls data, loads, EPR's, etc. This would be seemingly at pilot's discretion, but flipping the character of the data from saving some Mx money to transmitting the a/c's information in an emergency (certainly to include GPS fixes) seems a valid deal to me.
We are talking about a miniscule vulnerability to Radio isolation compared to system wide needs, certainly this could be done at reasonable expense?
bear