Surely a pair og GPS's in agreement are more reliable than even 3 INS's?
But no. Now we discover that if the GPS doesn't agree with the INS's it will effectively vote itself out of the system
Some problems relying solely on GPS:
- Multiple GPS units share common failure points (e.g., the satellite signals) so they are not independent -- two units may simultaneously fail in the exact same way
- Common integrity monitoring algorithms can't detect all failure scenarios -- they can report acceptable integrity when the data is actually garbage
- GPS signals are subject to interference, jamming and spoofing
Also when the first aviation GPS systems were being designed back in the mid-1980s and early 1990s, we did not yet have the full GPS constellation launched in space, and the US DOD was actively degrading GPS signals via Selective Availability (SA).
So I think the engineers & regulators decided to take a very conservative approach in preferring the "proven" IRS (incidents like KAL007 notwithstanding). SA was not turned off until May, 2000.
What about FMGC Fuel prediction , or expected landing fuel at destination ?
Prior to takeoff, fuel prediction is only based on the entered flight plan & departure runway -- not the aircraft position -- so it would appear correct to the crew.
radar vectors to an old-fashioned ILS should have been easily doable once the attitude and heading information was restored
I believe they restored (partial) attitude & heading information only after the decision to divert to Melbourne was already made.