Home of Ray the Wanderer
AI and Starbases
Published on September 17, 2003 By Ray the Wanderer In PC Gaming
Draginol suggested that I talk about the Galciv here on my blog page and that’s what I’ll be doing. First stop would be Galciv’s AI.

Galciv’s widely acclaimed AI is certainly one of the best I have played against for a space 4X game. The reported strength of the AI was the key factor that enticed me to shell out 50 bucks to download it direct from Stardock as opposed to waiting for it to appear at a lower price in retail.

I can’t remember the exact words but I read that Galciv is more a strategic empire building game rather than a tactical wargame. Consequently, I believe the Galciv AI was programmed to focus on developing its economy, planets and keeping its citizens happy.

Not surprisingly, due partly to the strategic intent as stated above, the area where the AI still falls far short of humans is in the area of tactical movement and in the conduct of war. It is oft explained that one of the reasons for not beefing up the AI further in this regard is that such additional AI routines would eat up CPU and thus put the game out of the hands of players with lesser machines. The fact that AI transports currently do not (even at Maso) scan for danger each turn to save themselves from becoming phaser cannon bait is an example.

I would like to propose that this is an area where Galciv’s AI could really stand out from the crowd. I believe TBS players would flock to a 4X game where the AI could emulate some of the better tactical moves made by humans.

As to the issue of alienating players of lesser machines, I would suggest that an option be created to allow players with better machines to activate “smart” routines. In this way, players would be given a choice to choose between the strength of the AI and the time lag between turns. I personally would gladly wait much longer between turns to play against a much smarter AI (on my machine, which is not top end, the only map size with a perceptible lag is Gigantic). I know Brad mentioned before that a suggestion that require an option in the setup screen usually implies a hardcore feature but I believe this would be an exception.

Let’s talk about starbases first.

SBs were inserted late into the game as a result of feedback from the testers. For such a late entrant, it’s tremendous ability to boost military power, influence, trade, production and empire abilities (through mining) elevates it to such critical importance that no Galciver can afford to ignore them.

Through months of gaming, Galcivers have derived ways to maximize the advantage that they derive from this powerful instrument. The AI’s ability to use them have not kept pace and the SB has become an avenue where humans extract tremendous advantage over the AI.

To preserve game balance, I would suggest that the AI needs to learn to use SBs at a level closer to humans, else the power of SBs need to be reduced.

Before I describe some of the SB models favoured by Galcivs, one critical imbalance needs to be addressed. SBs are currently not valued sufficiently by the AI. AIs need to cherish them at a level only second to planets. A human player would never dream of selling an influence SB to an AI if it meant impending assimilation by influence. The AI has no qualms in doing so (cheaply) even in the latest version (GCX)

A related issue is the lack of defense at AI SBs. The only means by which AIs defend their SBs is through the installation of defensive modules. The higher echelon defense modules could certainly provide adequate protection even against capital ships but most AI SBs fall easily under a deliberate assault in the early and mid game. These SBs are also helpless against AMMs. I would suggest that AI deploy ships of a sufficient caliber for defense duties at important and/or vulnerable SBs.

Here are some SB models favoured by Galcivers:

1. Close Space Support (CSS)

The attack, defense, speed and repair military modules available for installation on CSS SBs can buff up military ships in the SB sector considerably.

In older versions, a human player could park a CSS SB behind defended planets and watch as the AIs dash their valuable ships against buffed orbiting defenders. In the GCX version that I’ve tested, the AI ships do go for such SBs so such a strat is less effective.

I have seen AI build military modules on their SBs but frequently in uninhabited sectors where humans could easily avoid.

2. Party Palaces (PP)

The PP SB is a favourite instrument of the pacifist conqueror (oxymoron alert!). Multiple influence SBs have a multiplicative rather than additive effect (bug?) and no amount of cultural resist is going to save AI planets from a PP attack whether direct (in the target sector) or indirect (in adjacent sectors). The AI has been programmed to demand for such SBs if they detect them but coupled with the above-mentioned problem of the lack of perceived value, the wily human could always purchase the SB back the next turn. A couple of turns is all is needed for the human to convert all target planets using such an assault. Even Maso AIs are helpless against such an offensive.

I have seen cultural modules on AI’s SBs but never part of an influence offensive. The AI simply does not employ this offensive SB tactic.

3. Trade Centres (TC)

The key to unlimited riches. Unlike taxation which is capped, trade income is not. So every trade module built on a SB increases the income from trade freighters passing through the sector. Using speed modules to hasten freighters to sectors with trade modules or using shorter trade routes to keep them in those enhanced sectors, the human can extract huge amounts of trade income from those trade freighters. With a theoretical upper limit of 144 trade centre modules to be built in any sector, the sky’s the limit.

The AI does build trade modules but never at the intensity nor with the well planned design of humans.

4. Production

The production SB is a powerful instrument to accelerate production on planets. Even though paid and free production are capped on planets, it is still very well worth your while to build one production SB especially in sectors with multiple systems or systems with key planets.

The AI is reasonably good at building production SBs at their sectors.

5. Resource Mines

Mining galactic resources gives empire wide benefits. A maxed out resource SB provides +68% to their particular ability whether military, economy, tech, morale or influence. All resources are useful but the military resource SB has often been singled out as unbalancing. A few well-built military resources would often enable the empire to put out an uber fleet, crushing the militarily resource-less. This would not be unbalancing if not for the fact that AIs don’t defend their military resource SBs adequately nor do they seek out to destroy human military SBs before engaging their ships. Things humans do as a matter of fact.

The AIs excel at building up resource SBs.

6. Speed Catapults and Dampers

With the introduction of SB maintenance (which was upped to 10bc for GCX), these SBs have fallen out of favour. The Speed Catapult takes advantage of the uncapped speed bonuses from speed modules on SBs. With a theoretical upper limit of 288 speed modules in any sector, a ship can have additional 288 movement points (which means it could travel 24 sectors in one turn).

Dampers involve the stacking of speed reduction modules in a sector. In sufficient numbers, this could result in a dead zone where ships cannot move. This model has been used before to trap space monsters, enemy ships or to protect minors, etc. Again, the increased SB maintenance cost makes this model unpopular.

As of GCX, both the Speed Catapults and Dampers are ineffective. Speed boosted ships revert to normal speed after leaving their sector while ships can be damped only to a minimum of 1 movement point.

7. Terror Stars (TS)

The mobile SB platform with the big bad gun. The TS has been applied in both peaceful and belligerent ways. The culture bomber builds party palaces or other influence modules on the TS and moves them around to bomb sectors. Some combine this with the TS speed-fleet exploit to move TS quickly around the map, converting AI systems.

The traditional TS user chooses to obliterate all of the AIs’ systems in a single turn by sending a fleet of constructors next to the target system. Once the TS is built, an alpha TS strike can be launched at the AI in the next turn without any chance of response. This strat is obviously overpowered. One suggestion is to restrict the installation of TS modules to one per turn. This could be coupled with another measure to make AIs extremely hostile to any races that are building TS. Alternatively, we could restrict each race to one TS (or once we have logistics, TS could be made to require a huge amount of logistics points).

8. Starbase Maintenance

There were long discussions on the issue of SB maintenance and the chosen mode now is a flat 10bc maintenance cost for each SB. What this does is to make SBs cost prohibitive in the early game for humans. However, it is noted that the AIs still goes about producing numerous constructors and spawning SBs (esp over resources) in the early game.

It is my opinion that the SB maintenance costs disadvantage the AI more than the human. Some AIs still have the unusual practice of constructing empty SBs (they are not always for range extension purposes as I have seen some of these right next to built up SBs) so the AIs incur unnecessary maintenance costs especially in the early game where every bc counts.

Well, this has been a very long post so I’ll stop here for now. Watch out for future installments!
on Sep 17, 2003
Great stuff!
on Sep 18, 2003
Good review, but I'd like to point out more points about the starbases:


Starbases can be used to extend range. This is extremely useful if either range is limited or you are in an isolated cluster and there is a huge set of empty yellow stars just slightly out of reach. AIs should send a constructor somewhere to build a starbase that will allow them to colonize those juicy worlds.
The current ai inability to do so prevents any fun modding of the colony ship range and often results in one ai cornered on the bigger tight clusters map, resulting in a game vs. 1 less opponent.

On ai stuff in general, I know all too well ai is CPU expensive, but I don't want the game to be slowed down for that. It's already enough on bigger maps to my taste. Escorting transports rather than checking if there is an opponent every turn would already be an improvement, and not very CPU intensive (have the ai fleet its transports with a number of non missile units before it starts moving). This would be particularly effective if the escort ships strength was appraised based on previous attacks on other transports or just simply the best ship buildable.
on Sep 18, 2003
I believe that speed bonus' now only apply in the sectors with the speed enhancements. If you move the ship out of the sector it's movement points drop instantly to normal. This tactic is therefore no longer applicable.

If I was to ask for 1 single starbase exploit to be fixed, it would be to cap the effects of multiple starbases. This was played with during pre release betas and I definitely preferred the case where only the best starbase enhancement was used. In one single stroke it removes the trade starbase exploit, the influence starbase exploit, the military starbase exploit and the previous speed starbase exploit.

on Sep 18, 2003
Thanks for your comments, guys!

I've actually forgotten to insert one of the most obvious SB applications - the Terror Star. I've edited that in now.

Will check up on the Speed Catapult, Paul. Yeah, both speed catapults and dampers are ineffective now in GCX. Edited the article.
on Sep 18, 2003
Great stuff. Thanks Ray.
on Sep 21, 2003

great article, helped me alot.
tomorrow im gonna read part two