I stand corrected, and granted it's been a year since I've read the story, and I don't have a copy. LH2 shielding was my first thought on as to why the accident wasn't considered plausible. Furthermore, I know the accident created a propellant leak that would have vented the protective LH2 as time went on.
Honestly, this whole tragedy could have been avoided if the test was unmanned, but RL NASA sent pilots on the first space shuttle flight, and they do sometimes toss safety aside (which played roles in both RL space shuttle disasters), so it's not implausible for this to happen in the story. Honestly, I never saw it as implausible except that I think NASA might have had more failsafes built into a test NERVA and fly it unmanned.
It's just hard to say.
Also, I meant to say "containment failure." I missed that error. What I was thinking about was an electronic failsafe that would turn the graphite moderators to a full position to try to shut down the core if temps exceeded a certain threshold (and a temp increase did precede the explosion IIRC). Containment failure would be the NERVA shielding blowing and spreading radiation out.
I have a limited knowledge of NERVA and nuclear reactors, so I'm prone to make mistakes.
It would be interesting to model the radiation field of an exposed NERVA core. IIRC, only part of the shielding was busted, and it busted out the side, so I wonder how the radiation would spread out.