Do people posting here read Patrick Smith's "Ask the Pilot" column in Salon magazine? He deals with questions like this all the time and always has sensible advice. In a recent column about airline safety, he says this:
Considering the rarity of crashes in general, engaging in these sorts of hyper-analyses is mostly a waste of time. You can drive yourself crazy poring over the fractions of a percentage that differentiate one carrier's fatality rates from another. If you feel more comfortable picking Northwest over United, or Lufthansa over Korean Air, go for it. Will you actually be safer? On some minuscule statistical level, possibly; on a practical level, not really. (This same line of reasoning extends to equally popular aircraft vs. aircraft debate. Which are more trustworthy, 737s or A320s?)
As a pilot (private, not airline), I've followed the AA587 investigation, and it seems to me that the fault, if any, is not with the
design of the A300, but with how pilots were trained to handle wake turbulence encounters. After a couple of crashes in the 1990's, the airlines started doing upset recovery training that emphasized being quick and fairly aggressive with the rudder if you encountered wake turbulence. What they didn't say -- and maybe Airbus should have told them they should have made this clear (I think this issue is still in litigation between the airline and Airbus) -- is you shouldn't apply back-to-back rudder inputs in opposite directions. But this doesn't mean the A300 has a design flaw relating to the rudder -- either the composite structure or the relatively light touch on the pedals needed to do a full deflection of the rudder.
Smith's point in the column I quoted is that it's kind of pointless to make comparisons between (I'm making these numbers up) a 0.0001% chance of crashing on a particular airline/airplane combination and a 0.00008% chance of crashing on a different one. So choosing planes on the basis of IFE, legroom, or whatever actually isn't irrational after all.