The "corridors" are actually geometrically symmetrical arcs as you suspect, somewhere around which it's believed the plane traveled until the pings stopped, presumably because the engines stopped running.
Thanks! According to sources [1, 2] there was a period of time during which it was pinging multiple satellites. Surely this symmetry doesn't persist in a dataset comprising pings at different times to different satellites (which themselves are moving)?
They're in geosynchronous orbit, so they rotate with the earth, are always over the same spot on earth at the equator.
Don't know which satellite series is used for this service, but the Inmarsat constellations aren't particularly large: http://en.wikipedia.org/wiki/Inmarsat#Satellites, so it could be they only got solid data---and it's not very specific---from one satellite.
I also wouldn't put much confidence in mass media reports saying "satellites", and the Wikipedia plural use is probably based on them.
The map I directed to you was apparantly an enhanced one from the NYT; here's one with many more details figured in: http://www.nytimes.com/2014/03/17/world/asia/malaysia-airlin... although I wouldn't treat those boundaries as gospel, there's likely more slop in the real estimates derived from the satellite.
The "corridors" are actually geometrically symmetrical arcs as you suspect, somewhere around which it's believed the plane traveled until the pings stopped, presumably because the engines stopped running.