Gameplay Example: Computer AI in transition

The journey to v1.1

from JoeUser Forums

In testing the computer AI's updates for v1.1, I'm also making use of some of the new features in v1.1.

I'm playing on a large galaxy.  In this game, I'm playing as the Dominion.

Back Story

The Dominion was a commercially backed colonization effort from the year 2178.  Over 10 million humans from across the globe took part in it and headed out into the galaxy.  Over time, humans and aliens migrated to the new colony called Haven.

The year now is 2225 and the quadrant Haven is located in now shares space with the Terran Alliance, Drengin Empire, Torian Confederation and Arcean Empire.

No technology trading is allowed amongst civilizations. Only the Drengin and Altarians are equipped with the updated AI.

Running in a window in case anyone wonders why there's a title bar:

The quadrant's control was something like this at the start of 2227:

The Dominion had a vast area of territory under its control -- the light brown (me). The Terran Alliance (blue) had the "north" and the Drengin had the red middle.  The Arceans (yellow) and Torians (green) were on the east side.  It would remain to be seen how things would progress.

AI Evaluation

Our spies have been reporting back to us on how well the Drengin, Terrans, Arceans, and Torians are managing their worlds. The Drengin have an updated social manager engine so in theory, it should do better than the others.

The Drengin have used their own planet very well.  Two factories, one on a mineral rich area ensures that they are maximizing their planet. One might quibble over whether they should have bothered to build more farms or not.

Their advanced marketing center combined with high taxes with native morale bonus has given them a very high amount of income for this planet when combined with the economic capital.

Our grade: A-

 

The Terrans have really focused on research for their planet. Their native morale is lower than the Drengin's resulting in a lower income due to lower taxes. It's a bit weak on production.

Our grade: B+

 

Arcea is still using the 1.0X based social manager.  No huge errors but one might argue whether Diplomatic translators were worth having on their home world.  Their lower morale ability means that they can't keep their taxes as high as they might want. However, at 70% approval, one might argue why they don't push their taxes up in order to be more competitive economically.

 

Our grade: B.

 

Toria is also using the 1.0X engine but did a decent, but not great job at its planet. It's in the same boat as I am overall in that it doesn't have a great natural morale ability or big economic ability like the others picked.  Three economic boosters  helped their planet some but I think they'd been better off with another factory.

Our Grade: B

So what's MY home planet situation look like?

So this is what my home planet looks like at this time.

The biggest difference, however, between the two groups of 1.0X AIs and 1.1 AIs is in what techs they have.  The Drengin and Terrans have a lot of miniaturization technologies which they now value far more.  The Torians and Arceans, meanwhile, don't even have basic miniaturization.

Another research weakness I've noticed is with life support. The 1.0X AI doesn't tend to value it very much. Hence, on large maps, wars become less likely and give the human leader a big advantage in being able to isolate players.

The various civilizations tend not to go to war with anyone they don't have in range unless they've been...paid off to do it.

..Time
    Passes...

I have used my distance from the major powers to stay neutral in their wars.  I've also been able to watch and modify their behaviors based on what they should be doing.

Now we're in December 2233.  Almost 9 years have passed since the beginning.  The Torians are gone.

The Drengin seem to be on the rise having defeated them. But not so fast. The Terran Alliance have been building up the whole time.

So why my civilization, the Dominion, watches on, the Terrans and Drengin have an all out war. It's a great way to see divergent strategies take shape.

The Terrans have built no less than 6 starbases that cover earth. Moreover, these aren't the crappy little starbase spam of 1.0X that you sometimes saw.  They're maxed.

By contrast...

The Drengin Empire has no starbases around their home world. Instead, they've been focusing on building up a huge military machine which took out the Torians with impressive lethality.

The Drengin are using the Nano Ripper to create very nasty ships. I think I agree with others who feel that the Nano-Ripper is overpowered.  I also think Good Races need to get a new power. I was thinking a 50% bonus to their ship defenses.

The Terran Alliance isn't using the full 1.1 AI either.  They can't upgrade their ships yet.  The difference is quite apparent.

But they are making use of the improved ship design system.  They have a new type of Battleship:

14 moves!

So the question is, can the Terran Alliance's strategy of peaceful coexistence, weaker military but strong technology defeat the Drengin's further improved AI and ship upgrading?

That's what we were about to find out.

..and the answer...

The Terran Alliance crushes the Drengin Empire completely.

Twilight of the Drengin Empire

The Drengin Empire is nearly destroyed. The Terran Alliance, despite their weaker AI, is about to utterly destroy the Drengin.  This is an important thing to evaluate because it can mean one of three things:

1) The Drengin have a poor strategy

2) The Drengin just had some bad luck

3) There is a basic game mechanic that is non-optimal

I know #2 isn't true.

#3 the jury is still out.  The Drengin tend to build more, smaller ships. The Terrans tend to build more capital ships.

I've thought about making logistics go like this:

Tiny: 2
Small: 3
Medium: 5
Large: 6
Huge: 8

Anyone have any opinions on that? Basically two small vs. a large seems iffy as is. And right now a small is still worth 3 and a large is only 5. If I have a logistics of 15, I could have 5 small fighters vs. 2 capital ships. That seems more fair than 5 small fighters vs. 3 capital ships.

I think the real answer is this graph:

The Terrans simply were vastly more advanced than the Drengin were.  With the no-tech trade option on, the Drengin can't extort techs out of weaker civilizations like they'd normally do.

Let's look at the Drengin home world now:

Research Academies and Enhanced Factories.

Now for Earth:

Research Academies and Manufacturing Centers. The Terrans have better manufacturing capability.

Both reasonably designed.  I think basically the Drengin and others need to put more emphasis on technology research.

The Humans..are coming for us...

The Terran Alliance decided it was time to wipe me out. They also have an alliance with the Arceans so things are getting painful.

I do have a pretty good ship though against them.

It's a BIG ship. 

And now for the test.  4 of them vs. the Terran Armada.

Not enough.  My logistics is just too low to fight 6 ships of that level.

The Terrans dismantle my network of starbases around Haven.

Oh but I do love scaling.  My large sized ships against fighters.

 

And so the end comes..

Check this out :

The Terran Alliance destroyed 163 ships while only losing 38.  That's a 4 to 1 kill ratio. That's amazing.

Ironically, on the forums you hear people complaining that the AI always out techs them. I usually find myself out teching the AI but I'm more experienced.  I think we'll have to put in some effort into having it so that at lower levels the AI techs a bit slower but at higher levels it techs much faster because clearly the Drengin got smoked despite having all the advantages. It just doesn't research as well.

Next game: All AIs upgraded fully to 1.1 with tweaks from this game.  Stay tuned...

176,284 views 55 replies
Reply #1 from GalCiv2 Forums Top
Nice, interesting read. Looking forward to the next one.
I'm all for the change in logistics.
Reply #2 from GalCiv2 Forums Top
The first thing I thought when I saw the final screenshot with the humans' ship kill rations was, "that looks like a human player's stats."

This looks exciting.

I've only thought about it for a few minutes, but I'm in favor of your proposed Logistics change. What I've found is that I can rarely pit smaller ships against the enemy's larger ones if we are at the same tech and logistics levels. That makes the choice of going for many small or a few big ships less interesting unless I'm also chasing down Logistics techs.

Perhaps you could argue that wanting to field many smallers ships means I should be chasing more Logistics... Hmm.

I also like it because I think it will lend itself more to a capital ship system, in which each fleet gets its biggest ship, to have the ranks filled in with smaller support craft. As it is now I never pair Medium ships with anything else because the logistics cost of 4 almost always divides into my Logistics ability without enough left over for small or tiny ships.
Multiplicity - Control multiple PCs using one keyboard and mouse
Reply #3 from GalCiv2 Forums Top
I'd agree with the change in logistics. The jump to medium ships is a huge bonus right now. A fleet of medium ships can currently take out a fleet of small ships of the same tech level and logistics points. I'll be interested to see how the change goes.

-Dewar
Reply #4 from GalCiv2 Forums Top
Great post. I'm with you 100% on the logistics change. I like fighter craft, as I tend to play mostly on small-medium galaxies...and in those games you rarely get past that point before someone wins.

If I really race to them I'll pick up a medium...but I only get to Large or Huge if I slow-play and get there on purpose. If I play to win I can do it with nothing but fighters. It's tough taking on large ships with fighters if your enemy has advanced logistics. Fortunately for me when I play the computer rarely gets that unless I sell it to them.
Reply #5 from GalCiv2 Forums Top
Yeap, the logistics seem to be "unfair"... In my very first game of GC2, on all Intelligent, I found myself dominating the battlefield (battlespace? ) only because I focused on Building techs.

I think the problem is in Linearity. The logistics system is LINEAR (2,3,4...), while if you think on a spaceship, I would say the logistics needed to support a growing ship should be proportional to volume, or at least to surface.

What that means is, that logisitics points should be based on at least square feet of ships. A simple representation of that would be to use the square points instead of the linearity as is now.

Example: tiny=1, small=2, medium=4, large=8.

I know, that calls for a complete redesign of the system. But I think it will work better in the end.

Maybe, hull space should be the baseline for logistics. Use the hull space as representation, instead of linearity.

Example: baseline, tiny hull= 1 log point. Then, for all others, use the ratio of hull space to the baseline to define log points. That would make sense, and is easy to implement.

My 2 cents.
Reply #6 from GalCiv2 Forums Top
I'm more into capital ships and prefer medium hulls and up. I don't want to have to build tons of fighters. I'm all for balance but I don't want fleets of tinys/small ships to have an advantage over mediums and up.

A medium ships is obviously going to be more powerful than a similar tech small or tiny, and IMO, 3 mediums should obliterate 5 tiny/small of similar tech.
Reply #7 from GalCiv2 Forums Top
Looks fantastic. Thanks Draginol, and thanks Stardock.
Reply #8 from GalCiv2 Forums Top
Looks and sounds good. Really like the better ship designs the AI are coming up with.

I noticed that the Terrans still built one of those annoying orbital fleet managers on a bonus farm tile though! In fact they all completely ignored increasing their population at all, which will mean that they are economically dominated, and easier to buy off (not to mention easier to conquer).

As for the logistics changes - sounds reasonable, but how about a logistics value of 1 for any ship with zero attack ratings - would really make shifting constructors around a lot less mind-numbing early-mid game.
Reply #9 from GalCiv2 Forums Top
I have a question about how you make these screenshots:

Do you use cheats/debug modes to "spy" on the AI? or are you doing this by spending tons on espionage. I don't usually spend much on espionage as I'm not that good at managing my economy yet...so I'm usually strapped for cash. I've gotten to the level you need to view planets on several occasions...but not on every single AI player as you have.
Reply #10 from GalCiv2 Forums Top
I've been messing around with the log values of the ships myself. As of now it feels like once you've researched med ships you never have to build a tiny ship again. I'd like to see value placed on each size of ship so that no matter what stage of the game your in, the ships all have value.

Tiny: 2 pts, Decreased range. Cheap local fighters for defense.
Small: 3 pts, unchanged
Med: 5 pts, increased hull and HP.
Like to see the above three used by a swarm/fleet type of player.


Large: 10 pts, increased hull and hp
Huge: 99 pts, vast increased hull, hp. Ship maintence cost high enough to restrict a player to owning only one.

The above two ships for the player who wants his big bad destroyers all alone in the night.

I'd also like to see the total amount of logestics reduced. I don't think a player should have more then 20-25 ever.
Reply #11 from GalCiv2 Forums Top
You should be rewarded for building ships of larger size since you took the time to research the hull tech. If you have a fleet of larger ships you should always be able to easily beat a fleet of smaller ships given the same logistics and weapons/shield levels.
Reply #12 from GalCiv2 Forums Top
I noticed that the Terrans still built one of those annoying orbital fleet managers on a bonus farm tile though!


Tell me, how many Entertainment Centers and stuff would've been needed to keep the morale of the population from going into the toilet? How many factories and research labs would've needed replacing? The Terrans did great, in my opinion -- I only build farms on Gold Farm Bonuses when they're on Class 15 or higher planets. The morale hit and the many entertainment improvements needed to counteract it are, in my opinion, far worse than any economic hit you'll take in the long run. Now, if they'd built an Orbital Fleet Manager on top of a Precursor Lab or something, then there'd be something to complain about.
Reply #13 from GalCiv2 Forums Top
This kind of shows me that tech trading really smoothed out the difference in racial traits from a bonus research race to an average or poor research race.You could equalize it with a high diplomacy skill for another race.Add a race with a high morality skill and research bonus and it would be seriously lopsided.It seems you may have to adjust the native traits of the races to equalize this out somehow.Tech skill may become too important without tech trading.Perhaps a finer adjustment of percentages in racial traits.
Reply #14 from GalCiv2 Forums Top
A large part of the advantage of larger ships is the defense. It's hard to fit much in the way of shields etc on less than medium ships. Due to how combat works defense can have a major affect on battle results. (As can ship speed since a faster ship will stand a better chance of being attacker than defender.)

As is logistics is a huge factor already. Be sure you balance increase in logistics with increase in ship size. (ie. Getting medium ships should be about equal in value to getting equilivant in logistics. Don't make one a "bad" research choice.)

Want to see a ship walk? Make a huge ship with 20 def in shields, missiles, and armor, perhaps 4 nano rippers, and a few engines. Then watch it fight AI fleets of tiny, small, medium, and large ships. I think it might give you a feel for how to balance.
Reply #15 from GalCiv2 Forums Top
I agree with looking at tweaking logistics, making it harder to make large groups of large ships. But the equation has more variables than that.

Logistics
Maintenance Cost
Build Cost
Speed/Range
Internal Space

Maintenance Cost
The maintenance cost for making a small vs a huge isn't even double. Since I can only dedicate so much of my economy to maintaining a fleet (15-20%), I really like big ships for their economy. The initial purchase cost is insignificant to me. I regularly destroy my old ships because I don't like the monthly payments.

For the same maintenance burden, do you want 20 Dreadnaughts, or 35 Scouts? IMO, the advantage you get for dreadnaughts ought to come with a large, ongoing financial burden. But the opposite seems true.

Build Cost
The cost of constructing each next level up of ship isn't that significant, maybe 15-20% more expensive. Why buy the smaller when I can buy the larger for not much more? Again this seems backwards. I would think you could get 3-4 small ships for the price of the large one, not 1.5.

Speed/Range
Maybe small ships should be fast with a short range. Large ships should have more range but drag tail a bit. Or maybe the fuel consumption required to push dreadnaughts around at high speed should show up in a big way in their maintenance cost.

Internal Space
This is where I envision the advantage really comes in for the bigger ships. But in game, they don't really have that much more space. I can build a small with 10 missle attack, or a huge with 20 (although with some defenses as well).


I am not really saying any one of these is out of balance. Out of sync with my view of a space world, perhaps, but who am I anyway? But all of these factor into finding the right balance between large and small ships. My vision is that small ships are more ideal for scouting and local defense. They are speedy with good offense & defense, but it should be difficult to buy range with smaller ships. Large ships are slower but inevitable.

Its your game, and I love it. Not sure if you want to head down the path I describe, but those are the factors I see to futz with.
Reply #16 from GalCiv2 Forums Top
I'm on board for the logistics changes its annoying to have 1-2 useless logistics points left when building a fleet.

And the AI upgrade looks good, even with the "misuse" of the farming squares...which I do as well--smaller populations are easier to keep happy...
Reply #17 from GalCiv2 Forums Top
The Terran Alliance destroyed 163 ships while only losing 38. That's a 4 to 1 kill ratio. That's amazing.


This story brought a curious question to my mind...

In your experience of plotting AI vs AI, do you find the race X always beats race Y, who in turn always beats race Z etc? Or does it get mixed up a bit depending on circumstances? Considering that they all have different tactics, it would be hard to find a set that are diverse and also equal I would imagine
Reply #18 from GalCiv2 Forums Top
I would have been far more worried if he actually would have built a farm on that tile because that would have been really dumb on a class 10 planet. If he would have done so, his maximum population would have been at least 26 billion and he would have been forced to fill the world with morale-buildings to keep the population in check, leaving few tiles for more important stuff.

If I have one 100% morale-bonus tile like on Toria, I always build a morale-building and a farm (but only on 100% food-bonus or a generic tile) though.
Reply #19 from GalCiv2 Forums Top
Hi,

Looks absolutely fantastic. I'm curious though...will the AI algorithms take whether tech trading is disabled into account when determining how much to emphasize research? Seems like they should...

Cheers

h
Reply #20 from GalCiv2 Forums Top
Re: non-optimal game mechanics.

Would it be better if players were allowed to extort or gift techs even with the no tech trading option enabled? Seems like we're losing an interesting dynamic here, one that has little to do with the complaints about the AI trading too freely. Of course, it also seems like that would only benefit aggressive, "evil" players, while "nice" players would have to... turn on tech trading. *Scratches head in bemusement.*
Reply #21 from GalCiv2 Forums Top
Forgive the long post, but here are my thoughts on the points you guys brought up:

Technologies...
1. The AI changes are promising, but what about games that still allows tech trading? IMO, if nothing else, I think this particular example game proves how valuable techs can be. I think the AI needs to value techs a lot more than they do now so that they don't buy them all off of one another (if you trade one tech away, in 3 turns all the other AI has it). The reason why players have found AI out-teching them is that they trade tech like no tommorow, even military ones. I know there's a hardcoded denial of military tech when a player tries to trade with the AI, but the AI will OFFER their own military tech out, it's plain crazy (if you don't believe me, take a beeline for dimplomatic techs and watch all the AIs offer trades for it, they absolutely love dimplomatic techs - and rightly so). To take their words, "sell me weapons so I can crush them with it". Frankly, I'm all for more emphasis on research for most AI, and a much higher value for each tech, so that they don't pass around as often, but still possible (just much more expensive). Also put in some sort of tag to completely prevent tech trading (so that modders can change it if they want). With this, I think the core game will be much better, and a no tech trading option won't really even be needed. Taking tech trading out completely, IMO, severely weakens diplomacy as an ability because then it will just be a market.


Starbases...
1. I'm very excited to hear that the AI are making 'complete' bases now, frankly it was very annoying when the AI filled space with bases with really nothing in them. It's like they only put it there for the influence bonus, but they don't even bother using an influence Starbase. I just don't get it. I've had games where the Drengin will fill it's border with military bases, at first I thought "oh wow, this guy is smart!". As it turns out (when I maxed espionage), each one had like +1 ship attack, and barely any defenses at all. I just flew a fast sweeper fleet around and wiped his border sparkly clean. Imagine how much he paid to put up all those bases. From my experience, the Dregin wins the "best Starbase location award", but also the "I don't know how to actually use my Starbase award" as well. I've noticed the Torians do this sometimes as well. Also, the minor races will completely fill their space with influence Starbases, correct me if I'm wrong, but does that actually do anything for them? Not being able to on spy them means I don't know what they actually DO with those things... bleh.

2. Another thing I'd like to point out is the Starbase placement. When a human player goes to put up a starbase, you get this "can't put a Starbase within 3 parsecs of another" dialog. I'm pretty sure the AI completely bypasses this. Although they are programmed to "'space them out" to mimic this, I've seen the AI build bases right next to mine, within 2 parsecs all the time. It's really annoying that they can do that. IMO, there should definitely be some sort of moddable tag to hardcode limit this, both so that all players are the same, and so that players who thinks starbases are overpowered can raise and to prevent "Starbase clutter".

3. Also, not sure if this made it in 1.1 or not, but the AI should definitely stop giving starbases away for support of war (because starbases are usually their strongest unit/ship), I've had games where the AI would give me 2-3 economic starbases around their homeworld to support a war, not only does this make it prime for me to culturally subvert them later, but it SEVERELY weakens them because they have suddenly lost all those production bonuses. Bases should really be completely off limit for trading, unless it is a military base near the area of conflict, it doesn't even make any sense to give them away, except to weaken themselves.


Ship Sizes...
1. Some people have pointed out that huge ships don't COST as much as they should. I'm not sure where this assumption comes from, my huge costs a LOT more than a small, mostly because I tend to put defenses on larger ships (which costs a lot), I think the balance is fine, except that engines need to scale up in size more, meaning it should be HARD to make very large ships fly fast and be uber as well. Make the default size (for engines) smaller as well, so that tiny ships benefit, while larger ships suffer. Simply put, the size scaling needs to be better, at least when it comes to engines. For weapons and defenses, it's fine because they cost an arm and a leg, engines, however are insignificant cheap in comparison, so it's way too easy to make a 20+ speed huge, and the price difference probably would be insignificant when compared to a medium. Personally, I believe it would be better to make small ships to be used for fast offensive strikes (weapon + engines) while large ones as slow hulks of steel and death (armor and weapons). This way, small ship will have a use and a play style, while huge ships will still be better (to reward those spend time researching them), but has to be played a bit differently. I've modded my game to do exactly this, and have been decently satisfied with it, at least when it comes to my own design. However, right now the AI just doesn't emphasize engines enough for it to really matter, and it actually hurts them because they never put engines on small ships, so it's iffy. If nothing else, atleast it would explain why the Precursor Ranger has no engines...

2. To be honest, as far as I can see, the problem with large ships and small ship isn't really about the logistic points itself, but rather how experience affects them. Right now, when you gain a level you gain an amount of HP porpotional to your original value. A tiny ship, needs to gain 3-4 levels to gain a single point in HP. A medium ship only needs 1 level to do that (and sometimes they will gain more than 1 HP). A huge ship gains 4 to 5 HP (because you have +hull techs by the time you have huge ships) per level. Now I ask you, how easy is it for a tiny to survive a battle versus a huge? Now, put in the fact that after every battle, a huge ship will gain 4 HP after every conflict and you can easily see why huge ships are god forsakenly more powerful than tiny ones. Just a couple of skirmishes and you will have a ship that can tank an army of tiny ships. Personally I fear the day the altarians/humans can upgrade their battleships. I've seen some of those things with upwards 200+ HP. The only thing that could kill them was to load a couple of fleets of transports with nano rippers and go suicidal (I've never seen the AI used this tactic, so... they are usually completely screwed when one of those shows up, and they tend to show up in pairs). If they can upgrade those things? The races that use smaller ships is like a snow flake in hell, they don't have a chance.

3. In conclusion: Sure, I would be for your logistic changes, simply because it serves to weaken larger ships a bit, but a lot more needs to be done if you want to balance it. Although I think it would be better to further the gaps between sizes and then increase the amount of logistics by default so that you can add more small/tiny ships to a fleet as well (tiny=2, small=3, medium=5, large=7, huge=9, and add 2-3 logistic to all racial defaults). Maybe implement the scaling suggestions to engines I suggest in (1) will help as well, since that is a quick change, you can probably put that in, although changing how the AI uses the ship sizes might take more time. As for the experience and HP problem, I think that there should be a modable cap (for those who think it is too much/not enough) on the number of levels/HP that a ship can gain, just so you don't end up with invincible tanks (a 20+ level large/huge with a full set of armors) in the core game.


Hopefully nobody died reading the long post.
Reply #22 from GalCiv2 Forums Top
Here's another YES vote to the proposed logistics change. Can't wait for the beta...
Reply #23 from GalCiv2 Forums Top
One thing to note with the change in logistics: fleets of small ships benefit much more from military starbases than fleets of large ships do...

That said, I personally find that the larger hulled ships are actually not that much larger. When I can only put one more weapon on a medium hull than I can on a small... It's kind of unsatisfying already. It's nice to have smaller ships be a strategy, but larger ships SHOULD be advantageous. They're higher up the tech tree!

NEUTRAL TECHS. Why aren't we hearing about them? Don't mess with good unless you fix neutral!
Reply #24 from GalCiv2 Forums Top
The Drengin are using the Nano Ripper to create very nasty ships. I think I agree with others who feel that the Nano-Ripper is overpowered. I also think Good Races need to get a new power. I was thinking a 50% bonus to their ship defenses.


I have a couple of thoughts on this. First of all, while the Nano-ripper is very powerful, I have yet to see any player (even myself) manage to obtain it exclusively for more than a few turns. It seems that as soon as it has been developed, everyone makes sure they have it. It is also worth noting that currently this is the only counter that neutral and good races have to the evil races' Psionic weapons - and it's still not as powerful or cheap.

Second, I would be in favour of increasing good races' ship defenses, but I think a +50% bonus would be excessive.
Reply #25 from Stardock Forums Top
3. Also, not sure if this made it in 1.1 or not, but the AI should definitely stop giving starbases away for support of war (because starbases are usually their strongest unit/ship), I've had games where the AI would give me 2-3 economic starbases around their homeworld to support a war, not only does this make it prime for me to culturally subvert them later, but it SEVERELY weakens them because they have suddenly lost all those production bonuses. Bases should really be completely off limit for trading, unless it is a military base near the area of conflict, it doesn't even make any sense to give them away, except to weaken themselves.

Off limits? Well if you're a nice ally, sure - I often play as the peacekeeper with a price: You give me that mining starbase, or you might as well surrender to the Drengins now, because a) no-one else is going to stop them and b) I'll take it anyways if you don't, so you might as well get to survive while I do it. The same could be true for extortion - sure it hurts to get rid of the base, but my fleets will hurt more.
Maybe a "crisis" flag in the AI or something that allows larger things (starbases, maybe planets on really large maps) to be traded in exchange for something that will contribute directly to solving the crisis, either a declaration of war, ships, huge sums of money or weapons tech. A "crisis" might not even be hard to determine scripting-wise - if you're alone in a war with someone who's three times your strength and have already lost a planet or two, *boom*, crisis flag. Admittedly a few things to make sure it isn't easily abused would be nice - far-flung colonies don't count as lost planets, Spin-Centers don't count for this calculation (or the AI takes into account ship positioning, but I imagine this would be significantly more difficult ), ships sold that are far away from the fighting don't count as "crisis" ships, etc - but I think it would help the diplomacy model a lot. I want to see civilizations get desperate when dying looks probable, not give up without a fight because they wouldn't part with a lousy technology or something (or, for that matter, have them give away a valuable starbase for just about nothing compared to what the base could give long-term).

As for ships, I guess my opinion on logistic costs depends on what the final ship classes end up being in terms of comparisons. I think a best case would be where a fleet consists of one large ship, a few mediums and some escorting smalls. Larger ships tend to be ill-equipped to deal with smaller ones due to their small size and maneuverability, meaning that many small ships would eventually defeat one larger one but take heavy losses in the process. The medium ships are equipped to deal with the smaller though, and quickly make mincemeat of them. The capitals can obliterate the mediums, making a kind of rock-paper-scissors where no single class is "ultimate", even in a single role (combat, in this case - certain duties such as transporting can throw things off a bit).
This exists in the real world as well - despite the massive value of aircraft carriers in the seas during World War II, a carrier battle group consists of many other ships due to its vulnerability to other surface ships or submarines. Likewise, a navy consisting soley of smaller ships would have been defeated easily, even if the investment was equal. Granted this is a science-fiction theme, but any other sci-fi settings I can think of have a variance in ship sizes as well, a fleet consisting of a vast range of ships due to their varied uses. I don't see why GalCiv would be any different.

For speed, what about engines taking up a comparitive size (a fixed one-tenth of the base, for example) and an exponential size cost per engine (first booster costs five, then seven, then twelve, etc)? Not that steep of course, but you get the idea - no ships traveling across the map in a turn, larger ships not being twenty times faster.