Both upper and lower sections of the building were built the same way -- except that the upper would be built of progressively lighter material, and also suffered all the fires. Point being that one is not a "block" while the other is merely "floors". They are both one or the other. Take your pick.
If one portion cannot handle the impact, then neither can the other. We see the upper block largely disintegrating. The impact it experiences (if we follow the Bazant/NIST scenario, and suspend our observation that in fact the upper block gets destroyed in mid air) is not just 15 floors on 90, but 90 floors on 15. The rubble supposedly created (and the volume that isn't ejected laterally) does not have the power to crush down the rest of the building because it is disassociated material. Sure, you could have floor failure, perhaps over several floors, but this wouldn't be able to tear down the entire building symmetrically, pulverizing all material, and at the rate of six floors per second. You would have a more organic, asymmetrical collapse where you could pretty easily see or at least tell what was happening.* You'd probably end up having floors caving in like they did in other buildings; floor assemblies hanging from the core, and you would have a much more significant core remnant that actually stayed standing and didn't disintegrate in mid-air 20 seconds after the rest.
I'm sure the rubble principle would be pretty easy to model, using a scale metal structure with a comparable height and framework with "floors" attached at both core and periphery, and dropping some debris of similar composition, or perhaps a bunch of rocks, onto the floors. See what happens to the structure.