Some changes I'd like to see:
1. Carrier modules. Lots of other people have put forth great ideas how this would work. A ship could contain multiple carrier modules, and interceptors would be player-designed.
2. Targetting computers. This is a technology line, with 3-4 technologies, that enable you to put better targetting computers/components on your ship. I imagine them being similar to sensors in cost and size. Without a targetting computer, a ship can only attack 1 target per round, but adding the computers would allow them to attack additional targets after the first ship was destroyed, if they still had weapons that hadn't fired.
3. In fleet combat, ships attack random targets. None of the old - everyone in the fleet shooting at the same target until it's dead, then proceeding to the next one. Due to the limitations of time, space, position, maneuvering, angle of fire, and dogfights, real battles require ships to engage different targets, because they cannot all target the same ship all the time at the same efficiency. Each round, each ship would randomly select a new opponent to attack. Perhaps technologies could be researched that allow you to give your fleet combat directives, such as "Attack larger ships first", "Attack smaller ships first", or "Each ship should attack opponents about the same size as them". Such directives would increase the chances that your ships would attack certain targets of the enemy fleet, but your ships would still have some chance of attacking any ship in the enemy fleet.
4. Ships need to take piecemeal damage, and this is how I would do it: Say I have a Frigate with 30 hp, and it has 3 engines, 4 weapons, 2 defensive shields and 1 sensors, for a total of 10 components. In battle, anytime it gets hit for more than 3 damage (which is it's HP/component count) then a component is destroyed. For example, during the course of a battle, it takes damage 3 times, for 2, 5, and 17 damage. The first hit doesn't destroy anything, but the second and third hit each destroy one component - which we randomly rolled from the list of 10. So, if a weapon and a engine are destroyed, the Frigate would now have 2 engines, 3 weapons, 2 shields and 1 sensors, and 6/30 hp. The hp could be regained relatively quickly even if the ship spends time alone in hostile space, but they would come back faster if the ship was in a spacestation, friendly planet, or fleet with uninjured ships. The component damage could only be repaired by bringing the ship to a planet with a spaceport or a military starbase with a repair module, and it would siphon off a small amount of the planets ship production to repair (ie, if the component costed 100, then it takes 20 shields to repair it... you get a discount)
5. Repair modules. These modules allow ships to repair other ships in the field from damage to their components. They would have a limited stock, which, once used, would need to be re-supplied at a planet. Preferrably these would be large enough that you would put them on their own cargo ship... you wouldn't want this taking up space on your battleship. In any case, it wouldn't allow a ship to repair itself, only other ships.
6. Starbase constructor module. These should have a constructor module that allows the starbase to construct modules. A starbase would start off 1 for free... Thus, it would require only 1 constructor ships to get a starbase rolling. Additional ships could be sent to build more modules and speed up the process if you wanted the starbase to be constructed faster. Of course, if your starbases are expanding themselves it would have to reflect a cost on your economy, just like improvements on a planet cost money.
6b. Persistent Constructors. This would work alot like freighters. When you setup a trade route with a civ, thereafter there's little freighter ships going to the planet. Similarly, when you create a starbase, thereafter there's little constructor ships going to the starbase from the planet the constructor was made at, ferrying goods and people there. Thus it would be advantageous to starbases with constructors on nearby planets, as you would have a shorter supply chain, with less risk of pirates and enemy fighters. Multiple freighters could be sent to a starbase, you'd want about 1 for every constructor module in order to keep it fully stocked with construction resources.
7. Interception control module. In the end game, some ships can move really fast... the fastest ship I ever made was a troop beamer that carried 2k troops and had 83 movement. In several games I've played recently, I've adopted a blitzkreig strategy where I declare war on an AI, move in several fast battleships to empty his plannets, then move in my troop carriers and take all of his planets in a single turn... his large fleets dissapear when his last planet is conquered. This module would provide a counter to this. Basically, a ship with this module can choose to engage ships during his opponents turn, as soon as they come into his sensor range. You could use these ships to man a blockade around an enemy civilization, or to form an outer defensive perimeter around your colonies.
8. Weapons, sensors, life support, and defenses shouldn't require more space on a huge ship than on a tiny ship. However, engines need to be different for tiny, medium, and huge craft. Huge engines are made to propel a large amount of mass at a relatively slow speed, reliably over long distances. Tiny engines are designed to move a small mass around at high speeds for short duration and distance. Basically, keep doing what you're doing now, but call the engines different names for different classes, and drasticly increase their propulsion for tiny ships, while drastically reducing their propulsion output for large ships. The same tech could enable an engine that moves my tiny ship at speed 18, but the engine for the huge ship would only increase its speed by 1. Currently, the bigger the ship I make, generally the faster it goes, as I have more space to pop more engines on the back of it, and engines are cheap. This would make small fighters naturally fast, but with limited range, while larger ships would be slower, but able to travel deep into enemy territory, giving us a good reason to use carriers to field fighters in enemy territory.
9. There needs to be more, subtle hints that a war is going to break out. You might be advised that the enemy civilization has started a propaganda endeavor that paints your civilizations in a negative light -- a sign that the regime is looking for an excuse to go to war with you, but hasn't gotten the people convinced yet. A civ could enact tariffs on another one as a way of goading them into war. A civ could authorize privateers to attack enemy shipping. A civ could be on the lookout for international incidents with another civ, which would greatly increase the chance of something happening that could be used to spark a war. (Examples: A. Sinking of the Luisitania, while it didn't bring the USA into WWI, it sure didn't help relations, or B. If you remember in 2000 when our spyplane was downed by china). Naturally, when the event happens, the other civ could choose how they react, whether they will make concessions to something that wasn't their fault (resulting in poorer galactic diplomatic standing) or whether they will take a hard line (greater chance of war).
10. Moral decisions shouldn't always have 3 options. Sometimes there should be only 2, sometimes you could give them 4.
11. Troop modules should be reduced to only carry 1/10th as many people as they do now, but these people would be soldiers. When a planet is invaded, only 10% of the population are soldiers, only they actually fight and die, the rest surrender if the soldiers are all killed. If a person wants to transport large amounts of population, they will need to use a colony module.
11b. Recently captured planets would be inefficient, as time went on, and the population grows, the planet would get back to normal. Let's say you capture a planet, and after their soldiers are dead you have a planitary population of 5Bil original inhabitants and 100 Mil of your marines. Every one of your marines, and every new person born on the planet would count as a loyal citizen. For every loyal citizen, that loyal citizen and 1 original inhabitant would contribute to production. Thus, on the onset it would produce taxes as if it had 200 Mil citizens, and it wouldn't be producting at 100% again until it had 10Bil population. Similarly, social and military production and technology would be double the ratio of loyal citizens to total population, or 4% at the onset in this example.
11c. Implications of this: Conquering planets would be very costly on your economy. Happiness would help your new colonies gain population faster, which in turn would help them become more productive faster. A colony that has achived it's population cap would convert original inhabitants into loyal citizens at the rate that it would gain citizens were the cap not there. New colony or troop transport being made on the planet could also only take loyal citizens, limiting your ability to use enemy populations in warfare.
12. Bigger weapons. Bismark, a WW II battleship had 8 380mm guns, 12 150mm guns, 16 × 105mm, 16 × 37 mm, 12 × 20 mm, and 8 × 20 mm guns. The bigger guns were for penetrating enemy battleships (the Hood, a british battlecruiser, was sunk from a single round), but the smaller guns were more maneuverable and thus better suited for anti-aircraft and smaller ships. By contrast, the largest gun on a destroyer would be about 100mm. Right now in GalCiv2 my tiny ships might have 2 harpoons, while my huge ship has... 10+ harpoons - but they're all the same weapon.
12b. Larger defenses. Each ship could only have 1 defense of each type, so this would mean that a battlecruiser would have a bigger one, and a fighter would have a much smaller one. Due to building costs, a fighter could probably only afford to carry 1 type, whereas a dreadnought would want 1 of each, albeit the most expensive one should correspond to the enemy's primary attack. Each defense type would be rolled individually then added togeter against an attack, resulting in a smaller standard deviavion for ships with multiple defenses. The following might be statistics for 3 end game shields:
A. Invulnerability field. Size: 30, cost: 650, defense: 55. -- Ideal for Huge ship's main defense
B. Superior Force Field. Size: 15, cost: 350, defense: 30. -- Ideal for Medium ship's main defense, Huge ship's off defense
C. Barriers. Size: 2, cost 150, defense 12 -- Ideal for Small ship's main defense, Medium ship's off defense.
Remember, we can get away with only 1 shield because each of the enemy weapons fire seperately against your shields.
12c. Accuracy would be relational to the size and speed of the ship you're shooting at. A big, slow carrier would be impossible to miss, but a small fighter could easily zip around, evading lazer fire. Bigger guns wouldn't need to be less accurate, as since they only fire once per round, with the odds of missing a fast fighter being high, it might take you several rounds to hit/destroy a tiny fighter. Good players will also equip their ship with a variety of mid size and small weapons to garantee that something is going to get a successful hit. Anyway, each weapon firing would have to defeat shields/defenses seperately, giving you a real reason to put that monster cannon on your ship, even though it would drastily ramp up the cost and reduce alot of size for other weapons. Also, since each weapon fires seperately, you can mix and match different types at different size ranges without it being a big deal.
13. Random number state preservation in saved games. This was present in Civ VI... if I saved my game, and then had a battle, then reloaded and tried again, the battle would happen identically. I'm not 100% sure how they did this, but it could be replicated by taking an array of 1000 pre-saved random numbers, and using them one at a time in sequence each and every time a random number is needed. When a game is saved, you save the sequence number. This stops people from invading a planet, seeing that they did poorly, and going back to their save and trying it again, hoping for better results.
14. Enemy troop transports, battleships, and fighters only contain military personel. Destroying a colony ship with civilians on board is evil! When attacking a colony ship, the AI needs to decide what terms it's willing to surrender with, and you need to decide what you will do with the PoW's... kill em, send em to your slave pits, send em to prison on your planets, put them on one of your planets, or send em back to their civilization.
15. Soft tech pre-requsite. Currently, each tech has a hard pre-requsite. Sensors II is the hard pre-req for Sensors III. But perhaps there's additional techs that would make developing Sensors III easier, and more time efficient. For example, perhaps Sensors III has Medium Scale building and Advanced Computing as pre-reqs. If you were trying to research Sensors III without knowing Medium Scale building or Advanced Computing then it would require 12 weeks, but if you knew Medium Scale building it would require only 10 weeks, and if you knew Advanced Computing as well it would only require 8 weeks. This could be used to make it easier to research your second and third weapons/defensive tech trees.
16. Race specific Hulls. What I mean is, not all Medium size ships would have the same base cost, capacity, hp, or even logistics. For example, when the Altarian's research Large scale buildings, they might get access to 2 new hulls, one of which costs 25 more, has 8 more capacity, but has 3 hp less than the other one. Meanwhile, when the Drengin research Huge scale building, one of the two hulls they get access to is a Super ship with logistics 9 which has 20 more capacity and 15 more HP than the largest Terran ship, but also costs 200 more.
17. Building prototypes should cost a very small amount of money, ie, 5% of the cost of the finished ship. So, to design a Medium ship that costs 1000 production, you'd end up paying 50 currency when you save the model to your list of available ships.