GMDS, wrong again! The trust didn't increase, it just remainded on the last set trust setting!
If you would have left a TL in a Boeing on the last existing approach trust setting it would have provided trust as much as the TAM A320 did. There is no difference in this case. In such a conventional cockpit, it would be even worse, because the TL position would only be a little above idle, thus more difficult to detect. But the problem was not that the TAM captain did not detect it (most prabably), but that he left it on CLB deliberatly.
Correct, the non-moving TL are not the ideal solution, but in this very case it has nothing to do with it. Because there wasn't ANY trust change!
You can argue a lot that AI's TL don't provide any tactile feedback, yet I have seen too many pilots fighting against the moving ones. The classical case of acting against an automatic, instead of concentrating on the FMAs and other flight modes. This possibility of overriding an automatic function is exactly the wrong instrument to train a pilot of good airmanship, i.e. first check the mode you're in then change the parameters.
AI's cockpit philosophy has helped a great deal because it really urges pilots to do the correct thing, not just anything. Still some people (mainly in less developed countries) try do go around such procedures and invent their own - to a catastrophic outcome from time to time.
I agree with you that the easiest man-machine interface would be to fly everything manual, only basic instruments and no help from the system. But it's still not safer.
Dani