3) Another thing in AI Wars that is more restrictive than in other strategy games is the ability of the player to affect the AI's means of waging war.
Personally I would absolutely love to be able to smash up the AI economic and industrial infrastructure (and to see a separate economic, industrial, and research component to AI Progress, but that's just not going to fit the game)... but this is getting into one of the other big, huge, nasty problems that AI War's core design is dead-set-against: getting to the midpoint of the game and you've clearly won, it's all just easy mop-up. Granted, some players still get to a mop-up point by mid-game, but they have to be careful even during that phase or they can get torn to bits. So any exposure of the AI's infrastructure would have to either be really superficial or the destruction of it would have to shift the challenge to another aspect rather than just remove it.
The way I see it, there are roughly 3 'stages' in the game.
Stage 1: The start of the game. You have limited resources and are unable to react rapidly to incoming threats. You may not know what type of opponent you are fighting, which adds to the uncertainty. You also don't have a very good idea of what's around you. A mistake at this point, overestimating an AI force, not spotting an alarm post or raid engine, will end the game very fast. This is the riskiest part of the game, as your homeworld is most vulnerable to attack.
Stage 2: You have taken adjascent systems to your homeworld, and building up a reasonable resource pool and fleet. You are no longer directly at risk, the AI can no longer directly threaten your homeworld without fighting their way through buffer systems. You have a good idea where the AI homeworlds may be, and are planning a desirable expansion towards those objectives. You might not be nimble enough to stop the AI from destroying command centers in your outlying systems, but this is more of a nucance than anything. Rebuilding is cheap and easy. Your important structures are well secured and unlikely to break. You should be able to attack and clear Mk III and Mk IV systems at this point.
Stage 3: You have a solid, defended cluster of systems which the AI is unlikely to be able to threaten effectively. Resources are no longer an issue, as you easily build to ship cap and start storing up metal and crystal. Even if you lose all your ships, your resource stockpiles could allow you to rebuild thousands of ships in very short order. You are likely just a few hops from the AI homeworlds, or can easily extend your reach to them if you wanted. The AI waves are no threat, you are too heavily defended for an attack to penetrate your defenses. Even against a CPA, you have enough ships to protect yourself. If you lost now, you would not feel out played by the AI, as it makes no thoughtful moves against you, you would probably simply blame yourself for not reacting or preparing properly.
In my opinion, Stage 1 is obviously the most fun. There's a lot of unknowns and you have to change your plans a lot, especially on higher difficulties to not get overrun at this point. You have a very clear goal: secure your position against the AI. This isn't always easy, and how you achieve it varies quite a bit from game to game.
Stage 2 lacks direction, you have no real goals or objectives that you don't make up yourself. It might be by design, but the game does little to push your hand. You make yourself a shopping list of planets you want to get based on their value, and then you go out and take them. There's little urgency, the AI has no plans to get you, and you have no way of hampering the AI. You know the AI you're fighting, know how it will act and how it will react. This part is still sort of interesting, because there's a lot that can go wrong and managing AI attacks can still be difficult, but it doesn't have much direction. Stage 2 doesn't vary much from game to game.
Stage 3 is usually when I stop playing. It's just not interesting. Once you're at this point, every game plays out more or less the same. You've got your chokepoints with superfortresses and shield boosters and hundreds of turrets with a near limitless supply of fodder ships to back it up. Some players have shown resiliance against an AI progress as high as 3000+. At that kind of game, it's not really interesting. The AI has no new cards.
For that kind of reason I'm unsure some of the current endgame scenario ideas really help. The Avenger is an interesting idea, and quite a brutal ship to fight, but it doesn't really change the endgame. It's something that can be beat by throwing more time into preparation, not necessarially making it more interesting. And that's true for most of the ideas I've read. Having to kill objects to get to the AI homeworld, AI counterattack, they're interesting, but they don't make stage 3 any more interesting and if anything, it stretches it out.
I guess my main complaint, and the complaint of the OP, is that the AI doesn't do anything. Waves and CPAs are scheduled, the AI doesn't use them as tools, they are simply 'events' that occur randomly. A seasoned player will know exactly how the AI ships will move and react to the players, and know exactly the reaction the AI takes to specific events. Minor factions add variety but they still don't give the AI the ability to act.
It's maybe out of the scope of the game, and it may be unfeasible, but what I would kind of like to see would be:
1) AI Awareness of Opportunity. The AI should be aware of what the player has. It should have an idea where the player homeworld is, where his fabricators are, factories, power generators, and resources. It could assign a score (with some variance) to these systems, and make some kind of attempt to send wave attacks against the juicier targets rather than simply being random (although they should be somewhat random because being predictable isn't good either).
2) AI Awareness of Risk. The AI throws ships against the player seemingly at random, which isn't really effective nor is it particularly challenging for the player to overcome. Even if the AI has thousands of ships, it won't have much meaning if they throw it directly into your meatgrinder. The AI could decide which planets are more dangerous to attack, and weigh that against the opportunity available. It doesn't really have to know what ships you have there, it just needs to remember it's past history. The AI thought process could be "Well, I sent 500 ships here 20 minutes ago, and I only maintained a 10% kill to loss ratio. I should try striking somewhere else for now".
3) AI Desire to overcome Obstacles. If the AI is unable to make any headway, your systems are so bottled up that his attacks don't make a reasonable dent in your defenses, he should plan accordingly. If you've got a chokepoint and the AI can't get through it, then sending more waves won't really change that. In this case, the AI could hold back waves and launch a larger assault later. At higher AI progress, the AI could literally start building powerful ships such as Golems to use offensively to break these strong worlds, or at very high progress it could even begin production of say, an Avenger. These are objects you might go out of your way to stop, unlike regular minor factions that simply result in AI progress, these things result in widespread destruction of your faction.
Those are my two bits! It would be really cool if some of the AI logic were open in lua or something like that, there's some of us here that would love to come up with some crafty behaviors.
