Creative Speculation

The big question, for NIST, isn't whether the towers fell down. That is known.

Nor is it whether the towers were inevitably destined to fall down once the planes flew into them. (It might be of intellectual interest to know, if one could launch the Challenger 100 times under identical conditions, whether the O-ring would fail every time, but it isn't crucial to the analysis.)

Nor is it whether *therm?te was used, although NIST finds no need for that hypothesis.

The big question, as stated in the dedication, was: "How can we reduce our vulnerability to such attacks, and how can we increase our preparedness and safety while still ensuring the functionality of the places in which we work and live?" What can we usefully learn?

Many of NIST's critics seem to entirely miss that question. It's as if some agency did extensive modeling of Katrina in an attempt to understand how to protect New Orleans, and a critic complained that the models were rigged to obfuscate the likelihood that the levees were blown up. I doubt that it is possible to use modeling to prove that the levees were not blown up, but that is a strange way of construing the entire problem.

If someone were to present evidence that ?therm*ate was smuggled into the towers, then some security recommendations might follow. Or, for that matter, someone might offer plausible security recommendations even without presenting any such evidence. That would be refreshing. But it wouldn't obviate the value of NIST's analyses.

IIRC, Case A didn't give any bowing of exterior columns at all, and Case C didn't match the observed bowing as well as Case D did. So, there's a rationale for preferring B and D to A and C that doesn't depend on the fate of the towers. (William Seger made this point, too.)

I think some of the differences among cases are directly linked to estimated measurement error (for instance, the impact speeds of the planes), while others are qualitative and/or stylized (leaving intact the top 1.2m of the core wall). If it could be demonstrated that any of these assumptions were implausible, then we might be on our way to concluding that the observed bowing prior to the collapses cannot plausibly be explained without some additional heat source.

I don't think one has to posit any sort of confidence measure, although I suppose that is one way of thinking about the problem. Given arbitrarily large computing time, one could probably rig up some stuff that behaved like confidence intervals, but I'm not sure what the point would be.

I doubt I'm really getting at your question. Oh well.