[jarek]

In practice, that's just not an issue these days.

Most crashes are due to human error or complex system problems that often include already known issues rather than a single thing failing catastrophically. Airbus put out a maintenance bulletin about pitot tubes freezing before AF 447 crash, and cockpit/crew management, concentration and problem solving while startled, and computer mode confusion were all known. The most recent example of one solid hardware problem are the 787 batteries and thanks to QA systems we know they are problematic but, fingers crossed, so far they haven't gotten bad enough to kill people. If a 787 goes down you know this will be the first thing looked at even without instant access to logs.

A new airplane entering production with a flaw that will go unnoticed until it suddenly starts crashing planes en masse is just incredibly unlikely. On balance of probabilities, we're better off focusing on existing known problems rather than coming up with super high tech monitoring schemes.