Start by understanding that human beings have this interesting failure mode, in which they self destruct, and sometimes seek to destroy other humans in the process. We (because we are all at risk of this failure mode) may also seek to make a statement, make an impression, or even to vanish without a trace (usually achieved by, say, jumping off a ship at sea, rather than attempting to make an entire plane vanish in the Southern Ocean).
Now add in the fact that while self-preservation is impaired, the ability to devise ways to execute this failure mode, so to speak, is not. So for every clever preventative measure we can come up with, one of us who has the misfortune to suffer this failure mode will likely attempt to devise a workaround.
This failure mode happens at a very low incidence (though not as low as you might think) but since there are billions of humans, it is not uncommon.
Now recognise that at present we have a limited ability to recognise this failure mode before it emerges. In some cases it is moderately predictable (terrorism), but in many other cases there are few if any warning signs.
The warning signs that are available are purely behavioural; either the person takes actions appropriate to their objectives, or sometimes they might even discuss their intent at some more or less hypothetical level.
For example, this pilot is quoted as answering instructions to prepare for approach to the destination in an ambiguous manner ("we'll see"), and to have discussed at least hypothetically an action of this sort with his ex girlfriend, clues we recognise with hindsight.
The trouble is that this "failure mode" is so rare that few people ever recognise the warning signs, and that even where there are recognisable warning signs, the failure mode, at least on this scale, usually never eventuates. And many of us desperately want to believe that anyone who could "fail" in this way is fundamentally broken and not like us and our friends and colleagues. So we don't see the warnings.
Now consider that some of us will voluntarily step forward, recognising stress or distress in our lives, and seek help, with a resultant reduction in adverse outcomes, while others will recognise that seeking help will destroy much of our life and so attempt to tough it out on our own, increasing the risk of an adverse outcome, which, if the failure mode comes, will seem perfectly reasonable and logical.
So who do we trust?
The person with insight, who is prepared to ask for help, and to cooperate with treatment, or the person who insists all is well?
The problem is of course that as long as you believe that this failure mode we've been discussing is vanishingly rare, then the person who insists all is well is *almost* certainly fine, while the person who seeks help is seemingly a permanent, career ending, increased risk.
But once you recognise that mental illness is not actually terribly rare, especially if you include people who are at no risk of suicide but who are so preoccupied by their worries that their performance is impaired, you begin to see that identifying and treating mental illness, without making the personal consequences so dire that people will do almost anything to avoid being detected, is vital.
So, taking our hypothetical pilot with a failure mode, who has escaped detection and has resolved to act on his impulses, what can we do at this late stage to prevent disaster?
Suicide bridges are a well known phenomenon; many big cities have them. Often they eventually are equipped with anti-suicide measures such as barriers (West Gate Bridge in Melbourne, for example) and emergency phones (the Golden Gate Bridge in San Francisco), which reduce suicides at the site in question.
Some people take the entirely logical position that such measures are a waste of money, since people will simply go somewhere else to kill themselves. The Bolte Bridge in Melbourne is just as tall as the Westgate and just a couple of miles away for example. But the person who goes to a suicide bridge to jump is not logical; they have latched onto the concrete idea of jumping from that particular bridge.
It may well be that a cabin crew member in the cockpit, ostensibly only there to let the other pilot back in, may serve to force a suicidal pilot to act in the presence of someone else. Undoubtedly some will do so anyway, but I can well believe that some will be dissuaded, giving us some more time either to detect them or to persuade them to seek help themselves.
One of the difficulties is that this particular failure mode of humans is well beyond the expertise of pilots, engineers and ergonomists; the solutions will lie with psychologists and psychiatrists, and they will not be perfect.