The black box is sufficiently rugged to survive anything short of a nuclear blast, including fire, crash, and falling to the bottom of the ocean. Real time broadcast telemetry might be the first to go in some instances, black boxes are certainly the last.

Addition: the NTSB is expert at retrieving data from severely damaged black boxes.

The best reply I've ever seen on HN.

Redundancy is good, always. Especially when the cost is negligible compared to the lifetime operating expense of the machine and the costs from any misinformation regarding the last moments of a doomed flight (a Crash that is the Airline's fault mega-screws them in the wallet).

A black box captures human perspective of extreme conditions. Telemetry, flight control, and maintenance data is more than enough to figure out what happened to an aircraft. But, they don't tell why it happened? Data in a black box answers a critical question: "Is it human error?"

Having said that, your question does resonate well with me too. How much data does a black box store that we can't send it over satellite (or even to other aircraft nearby-- think of a distributed storage) in real-time or near real-time? The purpose of such distribution is purely redundancy. I haven't seen any concrete numbers or engineering barriers against this idea, yet.

I still don't see why only a black box should be used for that. As long as a satellite uplink is powered by a separate external power source, none of these factors will stop a transmission.

Satellite transponder space is expensive, and you would need a whole lot of it. Even if you somehow got the capacity, who would pay for it, and who would collect the data? There's also the problem that satellite communication is difficult to get right (to the level of reliability required by FAA regs) with any degree of consistency. It's also worth mentioning that when a plane is in a state that you'd want the uplink to work, it's generally not flying in a level straight line - it will be experiencing huge turbulence, buffeting winds, and random dives and climbs - which would make keeping an uplink very difficult.

There's also a practicality argument: when a plane is flying along normally, you don't need consistent satellite information - you've got radio communication, radio data-links, and controllers watching. So the usefulness of a satellite uplink only comes when something is not right with the plane.

If we're in that situation - that there's something wrong enough with the plane that you'd want to know about it via the uplink - what makes you think the uplink will still be working? Air planes are designed with huge levels of redundancy and overengineered margins: are you going to take all that, double it, and then put it into a satellite transmitter (which probably won't have that much data to send anyways if the plane is in that state)?

So no, I don't think we need constant satellite uplinks. Black boxes (especially ones of the newer generation) are very well-engineered, and the NTSB guys are masters at getting things off broken ones.

or you are flying through an electrical storm...

Yep. Commercial airliners already have a number of redundant power sources, located in different parts of the plane, wired in different pathways for maximum operational redundancy. There were even studies to done to find the parts of a jet that are least affected by eddy currents due to electrical storms....yet the black box can operate even when those different redundant systems fail.

Really? I think "falling like a rock and spinning like a top" would cause problems with your satellite uplink.