Wait,,,, so the NIST theory has it that the girder was pushed/walked across the seat to the point where it failed/fell off the seat. Chris now wishes to argue that the girder sagged which would pull axially along the girder and shorten the depth of girder residing on the seat. It would also reduce the stiffness of the steel of course, including that portion which sat on the seat.
In either case we have less girder steel on the seat and as Chris wishes to include the effect of the heat on that steel, we have more malleable steel on that seat , the combination of which would serve to have the seat-girder connection fail.
Then there is the scenario in which the fire begins burning down and the steel cools and contracts in which if Chris is correct the sagged girder now shortens axially even more, or as the CTBUH brought up, the sagged beams cool and shorten and pull the girder off its seat.
IN ANY CASE the girder comes off the seat!