These guys screwed-up, but they really had the deck stacked against them for a number of reasons...
1.) It was at night.
2.) The Relief Pilot was just coming off break and the first thing he noticed was that the Radar Scope was in the wrong mode, which made the thunderstorm activity look much less threatening that it actually was. This alone is a very distracting occurrence.
3.) Pilots that fly long-haul routes like these guys were doing get very little hands-on stick time and their basic flying skills suffer greatly over the years. The Relief Pilots often have to go back to recurrent simulator training due to not meeting the minimum number of take-offs and landings per 90-day period, which they HATE to do because it costs them 2 of their days off. In the U.S., allowing the aircraft to do an "AutoLand" allows all 3 (or for longer flights, 4 pilots) to get credit for a take-off & landing to meet the 3-in-90 requirement, but they get no actual stick time in the process.
4.) That model Airbus has a history of unreliable airspeed indication incidents. Note the entry "Then the sound of slipstream suddenly becomes louder." That's one of the most basic seat-of-the-pants indications that the aircraft's speed is increasing. The B-777 I flew would throw up an alert on the ICAS screen saying "Airspeed unreliable" to let you know that something was screwy with the Pitot system and you shouldn't rely completely on the airspeed indicator. Many airplanes still won't do that. So, when you couple the increasing slipstream noise AT NIGHT with no visual cues available, and the fact that a failing airspeed indicator will show a slowly
INCREASING airspeed that looks completely reliable with no other indications of a problem, it's easy the see how the guy could misread the situation and think the aircraft was moving toward an overspeed, when actually it was headed for a stall.
5.) Stall Warning. What most people don't realize is that an airplane can stall because it's going too slow OR because it's going too fast. Those guys were apparently flying in the "coffin corner" where the margin between high-speed stall and low-speed stall is only 20kts or so. Airlines generally want you to fly as high as possible for fuel economy, but the downside is that any amount of turbulence will cause the aircraft to bust either the high-speed or low-speed stall buffet boundary. If the turbulence is bad enough, the aircraft will alternate between busting BOTH the high and low speed boundary every few seconds. That's why you NEVER want to be at your max altitude if you anticipate ANY bumps at all. Going to Tokyo I always made sure I was back in the cockpit early before the Relief guys made the final climb to 41,000 for this very reason. The margin for error gets progressively slimmer the higher you fly.
6.) There have actually been a number of "jet upsets" as the FAA refers to them, due to this very thing, but the pilots usually get control of the aircraft before the situation gets serious. Northwest had a similar problem as Air France going to Tokyo
just last year in the same type aircraft, but they got the problem figured out before things got out of control.
The last actual crash I recall due to an unreliable airspeed indicator was Northwest Airlines Flight 6231
where crew was ferrying a B-727 with no passengers on board. They jumped in and didn't run the proper checklists, and missed turning on the Pitot Heat. They took off and had the same thing happen --the Airspeed Indicator was showing a slowly increasing airspeed, so they reduced power and pulled the nose up to keep the plane from overspeeding, so they thought. But, that was the exact opposite of what they should have done. Result was a big hole in the ground.
But the bottom line is, it was ruled pilot error, the insurance pays off, and everyone is happy.