If the pilot of the EgyptAir Flight 990[2] could not keep his co-pilot from crashing the plane, what could a mob of random passengers realistically do? It seems to me like the DRM problem - you can't protect the pilot and keep them from hijacking the plane at the same time.
At least the doors will prevent any of the other random 150 passengers from taking the plane. That does seem to me like an improvement.
[1] https://www.schneier.com/essays/archives/2005/12/airline_sec...
About the only thing I could think of is some way for people to report a problem like this and air traffic controllers on the ground having a way to take over control of the plane from people inside the plane. But that too could have problems (what if the air traffic controller on the ground is the murderous one, and now he is crashing planes that the pilot can't control).
I think it's a near impossible problem to fully solve. The closest thing I can think of is positive train control (PTC) which sets safe boundaries and overrides the operator if the train exceeds those boundaries. In this case maybe the plane simply can not be flown into the airspace around the Alps no matter how much the pilot wants to.
Even then, there are considerations like not allowing the cockpit to kill all com links / power so that people outside the cockpit are still able to communicate with the ground to get the root pwd, and as mentioned elsewhere the person inside the cockpit could change their strategy to make recovery of control difficult or impossible even in this case (EDIT: e.g. throwing the plane about). Also, you'd probably want >1 ATC person to authorise to reduce the chance of the door being unlocked by someone at ATC in cahoots with hijackers.
We just told the guards to sod off as we had our own private transport plane we were waiting for. Who on earth would hijack their own plane? Especially as there were 30 armed people on board?
That's the safest way to travel!
Let's accept it: we may make things gradually better, but there is no silver bullet that would solve all these problems.
To prevent attackers from muting all but the required majority of crew members, sensors all over the plane must be able to record any emergency code words spoken in the event of an attack, which will enact certain security measures (depending on the code word could lock or unlock the possibility to enter the cockpit, notify air security etc). This system should of course not be able to be deactivated.
And if you really wanted to make things sure, you could install majority-authorization buttons on every seat setup to be pressed by a percentage of boarded people within a period of mere seconds, which would allow the people on the plane to make a decision. For an attacker it is not possible to make all those people press the button, since he would not know who is not complying.
Of course you can also have some remote mechanism of unlocking, although the security implications of remote authorization and transmission of commands would be complicated to get right and open too many attack vectors.
Thus rendering the doors essentially useless, as hijackers threaten to kill passengers one-by-one until the crew (who will generally outnumber the 1-2 people in the cockpit) open the door.
Any procedure that aircraft crew is supposed to handle also needs to be something that is trained, memorized, and practiced regularly. Certain simplicity is ideal.
For instance, to replicate someones fingerprints all you need is a high res photo of their hand from a distance.
Then the airline will move on in the arms race by responding with a fingerprint reader that wants to have a pulse in the finger, and then it doesn't work when the user is in distress and has an extremely high pulse. And then there is an accident and we the armchair security experts come up with more bright ideas.
Last time I had an airline meal, even in first class, the knife "blade" was hard rubber mounted on a blunt metal handle. It could barely cut the food cleanly, never mind someone's finger.
In hindsight a policy of always having at least two people does seem preferable. (And, just a little reminder here: Just because someone is willing to fly a plane with 149 other people on board into a mountain doesn’t mean they would be willing to do other awful things, too, like acting violently in person against another person, for example a flight attendant. Preventing this kind of behavior may well be about putting in place many small hurdles.)
I think it’s pretty clear that people in the cabin should never be able to access the cockpit against the express will of people inside the cockpit (preventing scenarios in which someone violently forces someone with knowledge of how to access the cockpit – e.g. a pilot coming back from the toilet – to divulge that information and gaining access that way), so realistically you can only throw people at this problem – and throwing people at this problem is thankfully in this case relatively straightforward and shouldn’t really have any other negative consequences. I don’t see a technical solution to this problem, but there might not have to be one.
It certainly was debated before (here a german article from a year ago, closing with the cynical words "Than you're also dead, but at least save from terrorists. How comforting." http://www.austrianwings.info/2014/05/verriegelte-cockpittue... )
"Was geschieht, wenn sich ein Pilot zur Toilette begibt, sein Kollege im Cockpit - aus welchen Gründen auch immer - das Bewusstsein verliert oder verstirbt (solche tragischen Fälle kommen durchaus vor) und der nun außerhalb des Flightdecks befindliche Pilot plötzlich keinen Zutritt mehr zur Steuerzentrale des Flugzeuges hat? "
What happens if one pilot visits the WC and his colleague in cockpit - for whatever reason - loses his consciousness (such tragic cases do happen) and yet the pilot outside of the flight deck has no more access to control center of the aircraft?
It almost foresaw what was to happen on flight 4U9525.
YEEEEAAAAHHHHHHHHHHH
Ideally you'd have a door that is strong enough to be "people proof" for long enough for the entire passenger compartment to rush a dude, but not strong enough to survive the entire passenger compartment trying to get in and save the plane in this situation. Something that can survive a couple kicks but isn't a bank vault door. Pretty much a typical apartment door would be about right.
I disagree on this one. Bruce Schneier, who writes often, and passionately, on the ridicelousness of "security theater", cites doors (along with your your statement that passengers must resist) as having improved safety to a measurable degree.
But there isn't, sadly, an absolutely fix for every vulnerability. We can reduce the number of incidents and mitigate the risk, but some will always remain. It is the price one pays for such extraordinary freedoms and luxuries.
https://www.schneier.com/news/archives/2008/11/the_things_he...
I very much disagree. Previously, an attacker could get to the cockpit by surprise and by suddenly overwhelming cabin crew. That is now much. much more difficult.
If they are outside trying to get into the cockpit, there may be time for passengers to react and neutralize them?
Historically, hijackings are more deadly than rogue pilots. Neither one is a particularly big threat, but hijackings are the more important threat of the two. How many potential attacks from the past decade and a half got stopped in the planning stage because "Well gee, Bob, that sounds great, but how are we going to get into the cockpit?" While this stuff is rare, I bet it's not zero.
Beyond that, does not having a locked cockpit door actually save you from rogue pilots? I don't think so. There's a lot you can do to defeat other people when you're the only one belted in at the controls. A series of violent vertical maneuvers would suffice to defeat anyone trying to get back in. Imagine if, instead of locking the door, this guy had just pushed the nose over to negative two gees, then pulled up at positive three, repeat until everyone is dead? Or, pull the breakers for the higher-level fly-by-wire systems, point the nose at the ground, apply full throttle, and rip the wings off? The thinking seems to be that there was a calm 10 minutes or so in which he could have been stopped if only the door hadn't been in the way, but the only reason there was a calm 10 minutes was because the door was in the way.
Is this true? I can't remember a single other plane hijacking that resulted in the deaths of passengers outside of 9/11.
http://en.wikipedia.org/wiki/List_of_aircraft_hijackings
Lots of those were resolved without death, or with just a few deaths, but many others got lots of people killed.
There were definitely a couple.
As is the way of things, the current manifestations of asymmetric warfare can muddy up the discussion with modern-day politics, so please allow me to remind you that science fiction authors have been predicting the rise of it since at least the 1980s, just due to the ever-rising destructive power any one individual can wield. Who or what cause happens to pick it up is an accident of history, but you can be guaranteed someone will try. This is an incentive structure that those of us on the side of civilization are going to be doing continuous battle with for a long time.
Seems like a sensible idea with hindsight.
The pilot can override the code by locking the door in which case it can only be opened from the cockpit for some minutes (5-20).
(4) If one pilot leaves a pilot duty station of an aircraft when operating at altitudes above 25,000 feet MSL, the remaining pilot at the controls shall put on and use an approved oxygen mask until the other pilot returns to the pilot duty station of the aircraft.
Additionally, it's either US airline operating procedure (or perhaps FAA regulations, I couldn't find it though) for a second crew member to stay in the cockpit when one of the pilots has to leave.
If you're descending into an airport near NYC it's possible that you need to fly towards the city, turn and then continue descending to land. If you hit the autopilot external override at the right time you could be lower than the tallest buildings and get locked into flying into them without enough time to react.
The real solution would seem to be what other people in the thread are suggesting; you must always have two people in the cockpit. At least one pilot and possibly another crew member.
At this point, pilots are really there to take care of unusual situations that the computer can't (which do happen from time to time.)
Source: I have a pilots license.
"Oh yeah, we know we want to keep it level, so you pull back on the stick a bit, I'll push it an equal amount and we'll be okay." Yeeeah.
However, a potential problem with side-sticks arises because the sticks are independent -- the non-controlling stick does not mirror (i.e., no force feedback) the controlling stick movement, as would happen with the mechanically connected centre sticks/yokes. Meaning, the non-controlling pilot cannot tell what the controlling pilot is doing.
This was the problem on AF447. If you read the CVR log, one pilot was confused about how to handle the situation and, unfortunately, he took control several times during the recovery attempt. In the final seconds, he verbally states he is pulling back on the stick (attempting to climb), which revealed to the [then returned to the cockpit] captain what he was doing, but there was not enough time to correct (forward, nose down) to exit the stall before they struck the ocean surface.
(The stall warning itself was another disastrous UX problem as well, in my opinion.)
Completely agreed about the stall warning as well, I just didn't mention it. Don't warn about stalls unless there's actually danger of stalling!
And yes, there are emergency override codes, but there are also time override-overrides from within the cabin. Even if the protocol where two members of staff are always in the cabin was broken (or absent), again, that does not stop a rogue operator from despatching with the other person.
The question is (hypothetical at this stage, there are a lot of details still unknown): have these type of doors already saved more lives than the ~150 tragically lost this week?