Well, Boeing (McDonell) was made aware of their TOWS limitations after Detroit accident. And as a result they DID take action: They tried to convey that the TOWS was not a 100% trustworthy device to be relied upon and that at least it should be checked as frequently as possible, specially shortly before each attempt to take off.
Tons of airplanes fly everyday with TOWS that do not produce overly noticeable signs of failure.
It has happened with similar results (accidents leading to deaths) to 737's and even 747's (Lufthansa 540), i.e. !!!!
A better TOWS is always desireable, but in its absence, crew/training/manteinance/regulation discipline on exercising frequent tests would usually save the day (it would've in Spanair case, in theory, of course, because the crew could've just as well neglected to carry out a pre-engine start TOWS test, i.e., or the failure occurred after testing).
So only (a small) part of the blame, from my point of view, lies on the TOWS designers. But no point on not having a re-design for 2009. Nowadays it should be a cheap and easy endeavour and no point on risking lifes for a stupid $100 microcontroller re-do.
Making an alarm to warn on a malfunction of an alarm is, as you can guess, a "catch-22". The "lighted iluminator" or "alarm failure warning" could've just as well have failed. Not to mention that a diagnosis system might imply the TOWS is fine only to find out that the loudspeakers to sound the "horn" have blown up right around that time. So we would need an alarm for an alarm for an alarm for an alarm failure.
Nothing is really better than an "actual test" to see if the thing works or not for sure.
Except the better solution of a design in which the TOWS announces both, a working and a bad config state. Then "silence" can only means it is not working right and hopefully crew will recognize silence as something is wrong.
I think that human error can not be eliminated. It has happened before and after "commercialism" and will continue to happen. "Stupid" human error happens to INCREDIBLY top of line PERFECTLY trained pilots (i.e. chief test pilot for airbus A330 case, etc), on INCREDIBLY safe "culture" airlines.
It happens even more to "bad pilots" on crappy airlines, and it also happens more to crappy designed or maintained planes, of course.
But it just can not be eliminated.
So, in my view, rather than "training" (which is consistently ignored by us humans after a few years of repetitive tasks), regulation, etc, I vote for technology to try to tack these problems. Technology will introduce their own set of problems, DIFFERENT from human errors, but those problems (hopefully) CAN be resolved and eventually be done with. Human error CAN NOT be ever "fixed" and done with. Unless humans are substituted by "robots". Thus my point.
Let's say that 10 accidents a year are PROVOKED by pilots doing something "stupid" to perfectly safe/working planes/conditions.
Let's say that another 10 accidents a year happen because mechanical/electrical/machine failures that humans on board are not capable of overcome (i.e. humans were "useless" against a wild cabin fire, multiple engine failure, catastrophic structural damage, etc).
Let's say 3 accidents a year, due to severe malfunctions, would've ended it total catastrophy if a robot/computer was piloting. But because it was a human, human inginuity saved the situation from a "certain death" against all odds. The human did something no machine would've ever "thought of".
As a result, that year, we had 20 accidents.
If humans "weren't involved" in the equation, we would've had 10+3=13 accidents. We would've saved the lifes involved in the other 7.
Most of those accidents left could probably be "fixed" through even better technology, learning from the errors causing the accidents. But the 10 cases due to human errors, it has been proven through history, are fairly consistent in spite of countries/cultures/training/regulation/comercialism/you name it.
Anyway: the industry is already moving in that direction with airbuses etc. For a while, NEW accidents due to "technological" malfunctions will replace those "saved" by the computers not-allowing pilots to make "stupid" actions.
But eventually those problems will be fixed and, unlike humans that can not be fixed, the rate of accidents will hopefully go down.
I vote for, technology permitting, gradually phase out humans from the equation. It never hurts to have a trained human supervising the whole thing in case something goes major hiwire and he can save the day.
But I'm afraid that if a (redundant, fail-safe) computer can't deal with the situation, the human certainly better be smart and trained (and lucky) to stand a chance.
As it stands today, well over 60% of aviations accidents are the result of "trained" humans making really bad choices on "perfectly fine" (to fly) airplanes under not-extreme conditions (some minor "issues" or "malfunctions" which confused the heck out of the pilot when he *should've* known better).
And, of course, there is a large fleet of airplanes made before technology (redundancy, independently developed multiple-software-agreeable systems) was up to the par on seriously taking over human functions.
It's probably unavoidable. It may take 100 years or 1000 years, but machines will eventually fly us from A to B w/o much fuzz and as few failures as "God" allows. So no point in trying to avoid working on it right now.
Do you wanna know if I'll ever trust my daughter to be transported by a computer? Just ask yourself which you rather send your children with: a "fairly good record computer" or an unknown (but you trust well trained) taxi driver.
Last edited by justme69; 9th Nov 2008 at 15:35.