That's correct. Structural steel will bend a little and "spring" back but it will stay bent past a certain point and is not capable of storing a great deal of "spring" energy.
Yes this is correct. I'll elaborate if I may, because I get that little spark when someone understands. I don't mean to be condescending but others might like to have more info.
Look at the following stress/strain (Force/Area v [elongation-original length]/ original length) graph for a typical metal.
http://www.ndt-ed.org/EducationResources/CommunityCollege/Materials/Mechanical/Tensile.htm
Ever remember doing Hooke's Law at school by hanging various weights off of a spring and measuring the extension of the spring v load? Well the linear part is the same thing, it's elastic. The metal will "spring" back to it's original dimensions once the load is released as long as the stress the stress remains on the straight line. Again, as C7 states, once the load reaches a certain point, called the yield point, then the metal will no longer spring back to it's original shape, but is plastically deformed; or for want of a better phrase: bent.
Now imagine as you increase the stress from zero; follow the curve up, past the yield point, then stop at any stress before the UTS (see link) and then imagine reducing the stress gradually to nothing. The downward curve will not follow the one one the way up, but will now drop at the same rate as the original linear (elastic) part landing somewhere to the right of zero on strain axis. That's difference is how much the metal has been elongated or strained.
The two straight lines are parallel, but the one further to the right has a higher yield point. The next time you start to apply a load the graph will follow the second linear part. That is you have increased the yield point of the metal by previously bending it, but you have also made it more brittle (reducing the elongation after the yield point). Therefore the part of the graph between the yield point and the UTS is the part where "work hardening" occurs.
If we take a brittle material like glass or a ceramics then the linear portion of the graph extends right up until the material breaks. There is little to no plastic deformation (yielding) and therefore no work hardening.
At most, gravity can cause the exterior walls to be pushed out of the way by falling debris but falling debris cannot rip these interlocking sections apart and hurl them 400-500 feet in all directions at speeds of 50mph or greater.
I think that one of the things that we just cannot visualise is the tremendous energy stored within the buildings (potential energy). These buildings were so massive and so high that when we think of 12 ton (tonne?) sections being propelled distances we tend to think in more every day terms. When was the last time anyone saw a truck fall from 1/4 mile? The scale is mind boggling, there is a huge amount of energy released. Each time I see a collapse I'm reminded of slow-mo videos of water drops impacting puddles. The original drop is small and only has gravity acting on it's mass yet it manages to propel droplets in all directions, some a significant size and proportion away from the radius of the size of the impacting droplet.
I don't know if this is a good analogy (probably not) but it does remind me how a purely downward force causes lateral motion.