The elevator had sufficient authority to get the nose down - it was never stalled ... Had the stick been held forward the aircraft would have recovered, while the THS was returning to it's cruise position, (that's the behaviour we saw in the sim).
Right, and it's what we saw in the sim as well (my summary on the megathread is linked here):
http://www.pprune.org/tech-log/46062...ml#post6793521
The caveat was that it took between 5 to 8 seconds of constant full nose-down on the sidestick for the THS to return to a "normal" position. Release to pitch-neutral too early and you run the risk of the THS causing the aircraft to pitch up again. Of course, if you're certain of what you're doing that's no problem - but if the environment is as disjointed as the AF447 flight deck seems to have been, I'd argue it'd take quite a bit of nerve to follow through.
Yes, it is instinctual and must be, to the extent possible, "trained out".
Right again - I guess the angle I'm coming from is asking whether there may be more consistently effective methods of training it out than those we have now. And while it may not be productive to merely wonder, as you say, the idea that 80% of a random sampling of pilots instinctively pulled up when startled by a Stall Warning - i.e. the effect was such that it completely negated their training - worries me greatly. It also can't help but make me think we're missing something.