I find it very hard to understand why almost a whole week's time and effort was spent searching an area 2500 miles from Perth, when a better analysis of the satellite communications data on Friday now suddenly puts the most likely site of the crash to be 700 miles closer to the coast.
There are only eight hourly points in a time series giving distance and speed away from the satellite. Given that the radar data in the early stages of the flight is - at best - of very modest use, why did it take a week to make the fix?
There are only so many possible speeds; follow a swarm of consistent Monte Carlo-ed paths to the best place to search, and don't be distracted by various random pictures of whitecaps taken by random satellites from random countries.
It would seem to be long past time to release the full set of distance and speed values from the Inmarsat system and allow the world's spring-breaking students to mail in a guess. They couldn't do much worse.
There's also the issue of the lack of information about the time of flight after 0811, other than to say that it didn't extend as far as 0911. It could be anywhere from ~100km to ~900km. None of the search box plots include this degree of uncertainty along the track. Is this reported partial call after 0811 being assumed to be a clear sign of the first engine running dry, or is it wanting to present an unduly optimistic picture to the press?
There is a pressing need to try to find the wreckage before the sonar pingers run out of power, but that goal is surely not served by doing lots of MPA flying in the wrong place.