Newt,
Sorry, busy last (& next) 24 hours.
I agree that the model suggested captures none of the detail of columns-to-truss connections & interactions. And therefore, what happens within a floor, how the various columns are tied together, etc. None of those critical details.
But it does seem that it gives an accurate GROSS model of a real phenomenon: the load shifting that happens amongst the columns with respect to the sum total of the loads from all the floors above the damaged floors. And it seems that it matches the results from NIST.
For example, NIST gives the results for WTC2 in NCSTAR1-6D.
Here is the pre-impact loading on all the columns:
Pre-Impact Loads
[Aside: Note that the parts with the larger loads are physically larger columns. The important parameter that determines how much you are asking of a column is its stress level, or force per unit area. You can see how the engineers beautifully equalized the stress in this "Pre-impact strain" figure. Note that this figure is for WTC1, but is just as applicable to WTC2. Note also that, even tho it gives you "strain" (i.e., specific deformation), since the columns are in their linear elastic region, the strain is a direct reflection of the stress in each column. Simply multiply the strain by E (28,000,000) to get Stress. So you can just as validly interpret the size of the dots as "stress".]
[Obviously, the aside is for all the Electron Jockeys, Newt. Not for you.

]
Pre-Impact Strain (and proportional to stress)
So you can see how the stresses are nicely balanced between all the columns.
And here is what NIST shows for the loads (compare with the "Pre-Impact LOAD" chart above) 41 minutes after the impact:
Post-Impact Loads (41 minutes post impact)
You can see all the effects that Wierzbicki talks about in the load redistribution.
Within the central zone of tilt, the load numbers go up. Above & to the right of that zone, they go up, but less than within the zone. Some of the "outer fiber" columns in the upper right actually decrease in load.
And the columns below the axis of symmetry really have to start carrying excessive loads. (It does not explain the reduction in the loads of the East external wall, tho.)
Again, I agree with you 100% that all the cross bracing & floor trusses and the floors themselves are going to mitigate this effect to some degree. The biggest mitigation factor, it seems to me with cursory examination, is that differential creep is going to tend to re-equalize the loads somewhat. But that doesn't seem to me to be able to eliminate this effect altogether.
Simply summing the moments & forces requires that the forces (i.e., loads) on the side towards the tilt be higher.
Again, I'll be interested in how you see it. It's one of those things that takes some experience and fiddling with the real numbers to see. And frankly, this is far more in your baliwick (and Wierzbicki's) than mine.
Tom
PS. But again, I'll be busy today...