Misophist, the point isn't how good the combat systems are. I never played MoO1, but the MoO2 tactical AI was so simple to outmaneuver it gave me no satisfaction after I saw its blind points, and yet it was stupid enough that I wasn't about to let it be in charge of my ships. THAT is the point. Unless the AI is good enough to be equal to the player, that isn't going to change, and saying that an AI of that quality "may be difficult to develop" is sweeping the garbage under the rug. It would be a major resource drain for something that might not even be good enough to use in the end. Brad would love to have another month of nothing but polishing the strategic AI in GC2.
Look at all the strategy games with a tactical AI, and of all those, how many had a tactical AI that wasn't easy for an experienced gamer to beat? We're just not to the point that an AI can handle all the variables in tactical combat, those sneaky humans are too good at finding loopholes in the rules. The MoO1 repulsor trick is a perfect example. The tactical AI would have to either be programmed to recognize that specific trick, and all other cheese tricks, or would have to have the analytical ability to figure out why it is loosing battles and figure out a way to compensate. The first isn't going to happen because there will always be new cheese tricks, and the second is the kind of AI that exists only in research labs and takes more CPU time than the average consumer computer would have available and still have decent response time.
Right now, the release-level AI is good enough that I find "challenging" against 9 opponents just that. I might win 1 in 3 games, and I used to be one of the higher rated GC1 players. Add tactical combat and the inevitable cheese, and "challenging" is going to loose a lot of its meaning.