You want planets that cannot be taken by an equally teched enemy. Not "more difficult", impossible.
To defend 10 planets in this fashion, you have to build 10 separate defensive fleets. To take one planet, all you need to do is destroy one of those fleets.
The attacker has an innate advantage: being the instigator, they determine the time and place of the attack, and they determine how much force to bring to bear against a target. The point of this is to give the defender an advantage, thus reestablishing balance.
In modern warfare, the attacker ALWAYS has the advantage.
Modern warfare is not fun. Games are.
So you understand the way to use ships defensively in the proper manner, and want to completely unbalance the game to make the proper use of ships less effective?
We're talking about GalCiv3, not an expansion for GalCiv2. GC2 made lots and lots of changes compared to GC1. If it makes the game better, it should be on the table for discussion.
The point is that, in the game as it stands, there is a problem with combat: defending ones territory requires the same military that attacking does (the exact same ships that are good for defense are good for attack). This one-dimensional system creates the unfortunate circumstance that a well-defended nation is also one that can readily attack others.
I'm not suggesting a patch onto GC2. I'm suggesting that the design of all of the aspects of combat should be such as to eliminate this problems. A bonus when in orbit is a beginning.
Similar games (I will use civ3 again) it was close to impossible to win at the toughest difficulties, yet theorically possible.
I think this is less from the quality of the AI and more due to the nature of the rules. There are fewer ways to screw up in Civ-style games. Or, to put it another way, the path that a human would obviously see when playing GC2 is not how the AI was told to play.
Look at Civ-style city improvement. For any given square, there are really only 2-3 things you can do with it. You'll build a road if it's between two of your cities. You'll clear jungle unless there's something more important to do. Otherwise, you choose what you want that square of terrain to focus on: food, production, or trade/commerce. Depending on the terrain and the tech, you may not have certain choices available, but towards the end of the game, you can basically make any city high growth, high production, or high commerce.
Having only three city resources (growth, commerce, production) means that choosing what to do with a particular square is usually a no-brainer (make a square that is good at something better at that something). And even if you choose the "wrong" one, the worst that can happen is that your city is level 7 when it could have been level 9 or something. Fairly minor overall.
Compare this with what happens if an AI screws up building up a planet. On a PQ8 planet, the AI decides to give it a farm and some tax buildings, along with 3 factories. Now, instead of having a mid-grade production center (with 6 factories and 1 StarPort), it now has... a big drag on its morale. It has given itself an economic handicap. This planet, because of the farm, will be a constant burden on the global tax rate. The computer may even decide to try to mitigate this somewhwat by turining a factory into a morale building. What does it get? The planet maybe provides a net 30 BC.
In a Civilization-style game, no improvements are ever bad; they are only differing levels of good. Even when buildings cost money, or an improvement decreases the health of the city that uses it, it is a local effect. It hurts that city's overall strength.
In GC2, building a farm on the wrong planet can cripple your global economy by suppressing tax rates. The rules of GC2 has innumerable pitfalls for a player/AI that doesn't know how to build cities. Players, being human, can learn from their mistakes. The AI can't; if it isn't programmed to avoid the pitfalls (and it isn't), it will fall face-first into them.
There are basically two possible solutions: remove the pitfalls or give the AI better abilities to develop worlds. The later shouldn't even be that hard: the AI has plenty of computing resources. It should be able to min/max factory-heavy vs. money heavy vs. research heavy planetary builds. The current AI doesn't. Instead, it tries to develop a world based on some kind of algorithm that was written 2 years ago before people learned how to maximize the economy of the game.
Of course, here's the question: would such a computer that could execute near-perfect economic builds, ever be defeatable? Considering how biased GC2 is towards economy (money wins), I'm not sure.