I have since seen a drawing of what was supposed to be the finished bridge. There was supposed to be a tall vertical tower in the center island with multiple tension braces going down from the tower to the top flange of the bridge.
I do not see a center support or a tall center tower. That means the compression dead load just before it failed was many times higher than what it would have been when completed. I now have no doubt that the engineers did not check if the partially completed top flange was strong enough to take the compression before buckling. If they did, they made a mistake (fairly easy to do with buckling as a failure mode).
It is impossible to tell how much more compression was actually in the top flange compared to what it would have had when completed, but I believe it was a lot. At the VERY LEAST, they should have had the center pier (from the island to the bottom flange) in place. That would have cut the load by a factor of 4 from what it had when it failed. This looks like an engineering failure to me.
I think you’re right, except I wouldn’t call it an engineering failure, I’d call it a construction mgmt failure. As drawn (and I probably saw the same dwgs you did) the design is proven and there is a fine 12x over budget example spanning the SF Bay about 35 miles away from me. There are plenty of examples of this sort of quasi suspension bridge design in place (where there is a single central support tower; instead of dual tower designs like the Golden Gate or GWB) and those are fine.
Someone in management must have given the OK to yank the center temporary construction support without any part of the central tower or as you said, even the stanchion in the middle, from the ground to the lower flange in place. That’s simply unimaginable. Had the tower and stanchion been in place, there’s no reason to think the bridge wouldn’t have held just fine.