The aircraft is not in the area we looked in and so, she continued flying for far longer than we are expecting.
No not really. The back-drift analysis from the locations where debris and bodies were found do strongly suggest that 447 went down "somewhere" fairly close to its last know position.
But why weren't any indications of parts of her found (as far as we know at this time), during the last search go-round? I think they were simply looking in the wrong areas.
I understand the drift analyses (thanks to mm43 and the others who have made these understandable), what I really don't understand is why they used them to determine the recent search area. *Some* parts of the airframe had to go straight down (or as straight down as possible in whatever conditions prevail out there). It would be nice if they could find, say, the engines or a couple of robust structural pieces in order to more closely pinpoint the actual impact location -- and then work from there -- the FDR and CVR might be closer than they think.
I guess that's why I mentioned using ROVs similar to the ones currently being used in the Gulf a couple of weeks ago. JD-EE and auv-ee had a nice discussion of their capabilities and provided a couple of links to similar kit. And yeah, it looks doable, slow, but doable.