To me the radar problems sound like pretty typical software programmes - pretending that experimental designs are the finished product and not being willing to accept the true cost of developing something to the point where it is practically useful.
My limited experience is that time and competition solve problems. Competition solves the problems of lack of imagination and of excuses. Time lets people learn and work through the problems.
I have often thought to myself that having 2 teams to develop any software might be more efficient in the end - continual pressure for each team to live up to the standard and thereby be finally selected.