[m4x]

Real time telemetry via satellite could definitely be useful in some situations. The Air France crash could have been prevented if somebody on the ground had taken a look at the data from the plane, realised that the junior pilot has been pulling up the whole time, and made the crew aware of it. That was one of the worst crashes in history and there was never even a problem with the plane. All they needed to save the flight was for somebody to point out that the plane was stalling due to a bad control input by one of the pilots.

There are certainly cases where the telemetry link would fail as the aircraft ran into trouble, rendering it useless, but there are undeniably cases where it could save hundreds of lives

[hga]

That's demanding rather a lot from a human on the ground.

The Boeing (and I'm sure others) design philosophy is "unsynchronized joysticks ARE A TERRIBLE IDEA", and if Airbus and Air France had use the normal synchronized yokes, the problem of the brain wedged junior pilot would have been quickly apparent.

Can you think of other examples?

BTW, it's guessed that iced up pitot tubes caused the autopilot to disengage.

You're also assuming the pilots would listen to a kibitzer from afar while struggling to fix their plane. Plus a hell of a lot of expensive technology to support all that. And expensive kibitzers looking over the pilot's shoulders, which the latter wouldn't like.

[m4x]

I agree that a better fix would have been to have synchronised joy sticks, but that doesn't negate the usefulness of telemetry

I'm aware that the pitot tubes iced over, but they de-iced long before the point of no return

And I bet the pilots would have listened to a 'kibitzer'. The senior pilot actually told the junior one to stop pulling up, knowing it would cause problems, and would presumably have taken control more forcefully if he'd been aware that the junior pilot had resumed pulling up after bein asked to stop. He just didn't know. Better communication, syncd joysticks or telemetry could all have helped make him aware of that.

[jarek]

Presumably you've read the story of what went on in the AF 447 cockpit - do you suppose adding one more voice to the cacophony of warnings would really have helped? Two seasoned pilots that were in the cockpit didn't fully realize what was happening - there's no guarantee a remote overseer dealing with incomplete and inconsistent data would have either.

[mpyne]

> That was one of the worst crashes in history and there was never even a problem with the plane.

There were many problems with the plane, including the physical one of the pitot tubes freezing up (despite heating elements installed to prevent that), and the software and design problems relating to silent disengagement of parts of the auto-pilot, silent transfer to different flight modalities by the auto-pilot, and poor affordances in the physical user interface itself.

Certainly the subsequent pilot error did not help, as the flight would have been recoverable if proper action had been taken, but it wasn't just the pilots either.

Either way if you think some random shore sider would have been able to identify the exact pilot fault and clue them in within time parameters even with the inherent latency of satellite comms you more faith than I do.

[ekianjo]

> There were many problems with the plane, including the physical one of the pitot tubes freezing up

Not many, a single problem actually. Plus, the plane was not crashing after the issue occurred and the autopilot was off. The crew misinterpreted the signals and did not pay attention the Stall alarms repeatedly. The plane was flying fine even without the autopilot, it's a case where the crew actually crashed the plane by themselves.

[gvb]

a) Nobody on the ground knew AF447 was in trouble. The aircraft never broadcast a declaration of an emergency.

b) Even if there was telemetry that sent the aircraft control inputs and instrumentation to the ground, with thousands of aircraft in the air, nobody would have been watching AF447's telemetry, waiting for Something Bad to happen.

c) It was about four minutes between when the aircraft got in trouble and when it impacted the ocean. Even if the aircraft declared an emergency and someone was able to pull up the telemetry for that flight, it is highly unlikely they would have been able to identify the cause before the aircraft impacted.

Ref: http://visual.ly/air-france-flight-447-crash-timeline

[m4x]

So ask yourself why it never broadcast a declaration of emergency - the autopilot had disengaged due to an instrument malfunction, which should have generated an automated broadcast on the spot. Then somebody could start monitoring the telemetry from the the one (or five, or some other low number of planes) that were in trouble of some sort.

It would also be quite simple to be running a bunch of automated tests on the telemetry from every plane in the sky and flagging anything out of spec. AF447 fell into the ocean. It's pretty fucking simple for a computer to monitor the altitude of a plane and say "one of your planes has significantly deviated from it's target altitude".

And "highly unlikely they would have been able to identify the cause" seems highly unlikely. The plane was stalled, and somebody was pulling up as much as possible. Is it really hard to imagine a pilot on the ground being unable to spot what was going on?

Even the captain on board figured it out quickly once he was summoned - the problem was flagging the issue/asking for help was done solely at the pilots discretion and they chose not to tell any body what was going on until it was too late

[ekianjo]

> It's pretty fucking simple for a computer to monitor the altitude of a plane and say "one of your planes has significantly deviated from it's target altitude".

Look, read the AF447 case again and the reports of what happened in the cockpit! The pilots were not EVEN listening to the different alarms being triggered in the cockpit and did not realize that they were in danger no matter how many instruments they had indicating major issues in front of them. In such situations what would make you think they would give a shit at what an automated message coming from a land operation would tell them ?

[m4x]

If a pilot on the ground had checked the telemetry and seen that a) the plane was stalled and b) the elevator was at max deflection, he would have asked why that was the case and the pilot who wasn't in control would have realised what was going on. We aren't talking about giving the flight crew another automated message or bombarding them with unwanted input, just allowing somebody in a more sterile/low stress environment to monitor the telemetry for obvious problems

Sometimes a fresh opinion, or some input from somebody who is removed from the situation, is all that you need to set you on the right track or break an assumption that you were incorrectly holding.

[ekianjo]

The accident occured in minutes, there was no time for a potential operator to do anything about it anyway. Too many "if" in your scenario to make it sound plausible.

[tzs]

> The Air France crash could have been prevented if somebody on the ground had taken a look at the data from the plane, realised that the junior pilot has been pulling up the whole time, and made the crew aware of it

There are something like 10k commercial flights in the air at a given time around the world. How would that somebody on the ground know to look at that particular plane's telemetry?

[jarek]

This is the least difficult part of the whole scheme. A program looking at all incoming data and picking out extraordinary data (in AF 447's case, for instance the stall warnings) to flag for a human operator to inspect would help greatly. There'd be a lot of false alarms but still vastly less work than looking at telemetry of every plane ever. (And even if - 10k is around the number of employees of one mid-size airport - hardly undoable if you wanted.)

[ekianjo]

> A program looking at all incoming data and picking out extraordinary data (in AF 447's case, for instance the stall warnings) to flag for a human operator to inspect would help greatly.

Look at what happened in flight AF447. Everything occurred in a matter of minutes. Even if what you mentioned were to be in place, there would be virtually no time for an operator to do anything about it. And probably such operators would have to go through textbook questions because such regulations would be imposed on them by the FAA.

That's why we have several pilots in every aircraft, to mitigate the risk of human failure. Having more operators outside of the plane are not going to help much. If you want to put more engineering power, it's more software in the plane that is the right way to go for safety, as demonstrated so many times before.

[judk]

There are millions of web page changes on the Internet every day. How would a Google engineer know to look at one particular page to update the index?

[fnordfnordfnord]

By constantly comparing incoming data to known-good historical flight data and flagging instances that fall outside the norm?