Gen. Partin's strongest point in that article seems to be that objects which are equidistant from the blast site should have equal levels of damage. For a building of non-uniform construction, that notion is nonsense.
The entire weight of even-numbered posts A4-A8 was borne via transfer beam to odd-numbered posts A3-A9. This created a structure in which the failure of even a single post would severely compromise the strength of the remainder, if not collapse them outright. That the entire area supported by the transfer beam collapsed should hardly be considered surprising.
Imagine, for example, that rather than using a truck bomb McVeigh had only used a cutter charge to take out column A3 at ground level. What would have happened?
Although it might have been theoretically possible for the transfer beam to support column A3, it would have been cantilevered by twice the normal span distance. Stresses on the beam would thus be over ten times the normal loads. Clearly, then, the transfer beam would have failed somewhere between A3 and A5. Most likely it would fail near A5, since that would be the part under greatest stress, and because even if it failed between A3 and A4, letting A3 drop free, the cantilevered stress from A4 alone would probably be enough to break it.
The failure of the transfer beam near A5 would only be the beginning, though. The weight of A6 on the transfer beam will try to "tilt" the end at A5 toward it. Normally, this tendency would be counteracted by the weight of A4 (doubling the strength of the A5-A6 segment of the beam). Since A4 has dropped free, however, there is now nothing to counteract this moment force. Worse, as the beam sags it will apply considerable lateral force on column A5. While lateral forces from the beam would normally be borne by columns A1, A2, A3, those columns are no longer connected. Consequently, column A5 will experience lateral forces far beyond design parameters and column A5 may likely topple.
If it does, this sets the stage for a progressive collapse as column A5's failure will collapse A6-A7, then A7's failure will collapse A8. A9 will not be collapsed, because column A10 is supported from the ground rather than by transfer beam.
As to whether an undamaged A5 pillar would have collapsed if there were a clean cutter charge on A3 I can't be certain; likewise for A7 collapsing from A5's failure. Those columns would certainly experience loads far beyond those for which they were designed, but if otherwise undamaged it's possible they might remain standing. On the other hand, post A5 was almost as close to the bomb as A3, and A7 was still close enough that it would likely have received significant damage. Indeed, A7 is almost as close to the bomb as the nearest undamaged pillar, B4, whose lateral loads would have been much better balanced.
If Partin's credentials are as people say, I don't know why he would ignore something as major as the building's alcove which totally change the structural nature of damage and collapse. The only thing I can figure is that perhaps he was in a hurry and did not use full care in his analysis.