Reply #118: Here's a criticism by one of your community ...

[AlienSpaceBat]

This is from Arup, who you might have heard of.

(September 6th 2005)

NIST report on the World Trade Center Collapse – Arup View

(...)
NIST has carried out a huge undertaking in modelling the collapse which should not be underestimated, and the key responses described by NIST of events that led to the collapse are all plausible when considered qualitatively. However, having read in detail Report 6, which deals with collapse, Arup believes that NIST have not demonstrated satisfactorily their main conclusion – that had impact induced fire proofing loss not occurred, the towers would not have collapsed.
(...)

Barbara Lane Ph.D. ...Associate Director within the Arup Fire group in London, with several years experience in the New York office...leader of the Structural Fire Engineering group in Arup and is responsible for a developing a group of specialist engineers, which provide structural fire design solutions, as well as complex analysis of structural systems for fire.

Also, it's only fair to say that I am critiqueing the methodology of the NIST modelling process. I say that when a model's predictions do not fit known real world outcomes, the model itself should be iteratively refined, and that the inputs to the model should not be iterated through until the desired prediction finally results. This is a general point, and does not rely on anything specific to the NIST model. If this is not general good practice feel free to show me something to the contrary.

I don't think my opinion is a minority opinion. The main consumers of the NIST report are engineers, not software developers, so I'm not surprised they don't tend to express an opinion about this aspect. Conversely, as I am not an engineer, I wouldn't have read the report if I wasn't interested in the WTC collapse in the context of 9/11, so most professionals in the area of software simulations won't read it and so won't comment. I think if opinions were canvassed as to whether their modelling process was well carried out I think that they would support what I'm saying.