For a couple of years, I've been working with the wonderful group of Swarm Cycle authors. It happens that I have a background with real-world weapons and battle management systems, and I'm one of those "hard science" science fiction/wargamer people that like to have as much consistency as possible.
There's more of an author consensus, I think, that the AIs, and CAP score, weren't quite perfect when they first arrived. This story works, in part, because I can assume some very talented people were not scored appropriately in the early testing.