Entropy Avatar

Join Date 02/2006

Starbase Defence and DA Combat

It looks as though the starbase defence modules follow the same progression (5, 10, 20) as they did in the original, v1.0 GalCiv2. Given changes in logistics, defence research cost and the new DA combat system these values are really appropriate anymore.

Through the various patches, defence on ships has become more compact and easier to research, but I haven't seen any corresponding improvement to starbase defence. Similarly, logistics have been expanded, meaning starbases can face larger fleets of small ships with the same defences they had before. While in vanilla GC2 a starbase with say, 35 defence in the appropriate type could at least be fairly resistant to medium-tech small ships, in DA a fleet of small ships with reasonable weapons could wear through those defences fairly quickly.

Given how much construction resources these structures cost they should be defendable. In the vanilla GC2 combat system, you are playing a difficult balancing game, because you didn't want to make starbases immune to small ships. With the logistics and combat changes, I think it's time to revisit these defence values. I think doubling the values for defence would not be out of line.

Also: does a starbase get only three shots (one for each weapon type) or are they broken up by installed modules?

Locked Post

Advanced Search

Watch this post

Do not email me updates for this post Email me updates for this post

Successfully updated karma reason!

Also: does a starbase get only three shots (one for each weapon type) or are they broken up by installed modules?

Unless it was changed before release, starbases fire all their weapons at the same target (that is, they don't benefit from the new combat system at all).

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

What i did, was simply accept the fact that starbase defences don't exist.

There is no need to waste your time researching the technology or using the modules, just accept it is no longer part of the game and you will be much better off.

Successfully updated karma reason!

True. Starbase defenses are weak now. Although, I don't mind that much. When you go to war, both sides' infrastructures get ripped up and your economy can't boom. So IMO it's reasonable not to be able to build starbases in war-torn regions.

That said, it would be nice if at least military starbases had more D.

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

So IMO it's reasonable not to be able to build starbases in war-torn regions

true. I'm still jelous of 'deep space nine' being able to defeat a huge fleet of Clingon ships. The mining starbases in the game even look a bit like deep space nine which is cool!

Successfully updated karma reason!

What's really bad is how many constructors it takes to build up a starbase if you are foolish enough to waste effort trying. Add up what all those constructors cost, then compare that to a well armed medium or large ship. The ship is much cheaper and more versatile. You also have to pay for several of those upgrades, which will pay for maintenance on your ship for a long time. I think powerful starbases would make the game more challenging than the sitting ducks they are now. The civ that loses its resource starbases is most likely a goner, so we should have to earn their downfall.

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

I think powerful starbases would make the game more challenging than the sitting ducks they are now

I agree. having to work hard to dislodge AI starbases would be fun. Imagine if there were starbases so powerful that you would finally be able to have somthing to challenge that super engame fleet. The final culmination of all your efforts in the game, culminating into one hell of a final battle of the titans... AWESOME!

Successfully updated karma reason!

I agree. having to work hard to dislodge AI starbases would be fun.

This goes back to the AI again. It's easier to code a high-D starbase that's hard to dislodge that it is to code an AI which aggressively defends its starbases with fleets.

Successfully updated karma reason!

I agree that Starbase defense is laughable, and that it would be fun if Starbases were more powerful.

I station small fleets of ships on top of my Starbases to protect them before I go to war.

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

This goes back to the AI again. It's easier to code a high-D starbase that's hard to dislodge that it is to code an AI which aggressively defends its starbases with fleets.

Agree, but speaking of AI coding, i would imagine it would be even harder to program the AI to deal with enemy high defence starbases? The AI would have a tendancy to 'sacrifice' all it's ships in the attempt which would unbalance the game.

Successfully updated karma reason!

Right, that's what the AI did before, was to sacrifice its ships. I don't think it's too difficult to code an AI not to initiate a battle it doesn't think it can win. Or to go one step further, sacrifice ships only when you have a specific follow-on attack in mind. The previous DL AI just blindly sacrificed ships.

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

follow-on attack

Brilliant concept!

Successfully updated karma reason!

Or to go one step further, sacrifice ships only when you have a specific follow-on attack in mind.

That's how it's supposed to work. But as you can imagine, getting the right balance between making it so careful it never attacks starbases at all, versus the ever-popular suicide train, isn't easy to pull off.

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

That's how it's supposed to work

Interesting!, i have really only noticed the AI choosing the weaker target, apart from that, it still seems to attack my defensive fleets with no follow on. However on occasion i have actually noticed it rally several fleets before attacking, but not very often.

But as you can imagine, getting the right balance between making it so careful it never attacks starbases at all, versus the ever-popular suicide train, isn't easy to pull off.

agree

Successfully updated karma reason!

I think it's reasonable simply not to choose battles you don't think you can win. Fleet up and improve your odds first. Doing suicide runs with a _specific_ follow-on attack in mind would be the logical next step in improving the AI. Even then, I don't think you should suicide run if you haven't maxed out your logistical points yet.

Successfully updated karma reason!

Mystikmind

Join Date 06/2004

+20

I think it's reasonable simply not to choose battles you don't think you can win

You do have a point... the AI thinks it can win because the game stats tell it that it is stronger than you without considering what technology your fleet is using.

I dunno if the AI is very good at reviewing battle outcomes? But that would be considerably difficult to code in.

Successfully updated karma reason!

I dunno if the AI is very good at reviewing battle outcomes? But that would be considerably difficult to code in.

Well, it would be pretty easy to size up it's odds for each combat. Simply simulate the battle a hundred times and look at the average result. Heck, if it is considering follow-up attacks on the same turn, simulate the 2 (or 3 or 4) chain sequence of attacks a hundred times and see how it turns out.

I also think the AI should use a similar method for evaluating an opponent's military strength, instead of using the military graph.

Successfully updated karma reason!

Or better, simulate a battle a hundred times during the game's development, and distill the results down to a polynomial equation as a function of weapons, defense, and hit points. Starbases tend to win early on because of their hit points.

Successfully updated karma reason!

I'm kinda mixed here. While I don't like it when the AI manages to wipe out a few key starbases I also don't want a situation where it takes me forever to dislodge theirs. What I tend to do is station a defense fleet on top of any important starbase. Back that fleet up with a nice military base (again with another fleet on top of it) and problem solved. Since the stacking rules force the AI fleet to attack mine first the starbase stays intact. And with the changes to combat rules its not as big of a disadvantage anymore to allow them to attack first. This keeps me from having to worry about trying to intercept a couple of fleets vectoring in from different areas.

As for getting the AI to follow this sort of strategy who knows. On the surface it seems relatively simple to say station a few *escorts* on the starbase for protection. But then you don't want to have it try to protect all of its stations since that isn't always needed. Also it means that those ships are normally special in terms of att vs def ratings. My defensive fleet is usually more defense then attack just because it needs to survive several battles with minimal damage.

Successfully updated karma reason!

That's probably what it does now. Cook up a formula to try to approximate the outcome. Given what I see in terms of AI ship design though, it can be a long way from accurate. It's tough to come up with a formula that really takes into account all the variables in terms of weapons, defences, ship size distribution and bonus damage. Despite my theory-craft post, it's not that simple once you get into specific matchups. There are a lot of corner cases. Even if you do come up with a workable formula, you tweak the combat system (or someone mods in different values) and it's back to the drawing board.

I doubt it would take more than 2 or 3 milliseconds to simulate a typical battle on a remotely modern computer, possibly quite a bit less. I'm always surprised in my own programs just how fast a modern processor can grind through stuff. Of course, it would have been much faster using the old, simpler, combat system. Though I expect results in the new system are less variable.

Even a dozen runs would give you a pretty good idea of how the battle will actually turn out and you aren't blind to some corner case. There's really no reason the AI can't have a pretty good idea of whether it's fleet is going to take out that starbase or inflict a few hitpoints damage and be eradicated.

Successfully updated karma reason!