Brad Wardell's views about technology, politics, religion, world affairs, and all sorts of politically incorrect topics.
The starting of the tweaking
Published on January 21, 2007 By Draginol In GalCiv Journals

The Korath are the warriors supreme of the Drengin Empire. When the Drengin Empire absolutely positively has to win, the call goes out and the Korath always answer. The Korath always win. The Korath were the ones who took the Torian home world thousands of years ago. If the Drengin are a bunch of ruthless, evil, conquerors, the Korath are that to the nth degree.

As the Terran Alliance was about to celebrate its victory over the Dread Lords, it was the Korath who led the charge to crush the increasingly powerful human civilization.  The humans had never dealt with the Korath.  And if they had, they would not necessarily have noticed the difference in prior times.  But now, in the past couple of years, the physical difference between the Korath and other Drengin is quite noticeable.

The Drengin (and Korath) take pride that the Drengin home world has been unified under one supreme emperor for over 100,000 years.  Until they met the seemingly benign humans from Sol 3, they thought this was a source of great advantage.  But the pleasant negotiators of the Terran Alliance have a blood-thirsty past. And the Drengin and Korath have come to rue the day that they dismissed the evolved apes of Earth as being a race of merchants.

The Terran Alliance and Korath fought a series of battles for control of key parsecs. This is one of those battles.

(AI note: This is a test of the Korath computer AI, notes and changes and errors will be documented)

Game settings:  Intelligent (AI playing at the same level as the human player)

Detailed look at the player setup. CPU usage is increased to allow advanced algorithms.

Only one will survive


Terran Alliance base world: Poseiden. Two very special tiles


Korath base world: Karza'Dun. One very special tile.

Terran Alliance First 10 tech choices:

  1. Xeno Research
  2. Planetary Improvements
  3. Xeno Industrial Theory
  4. Xeno Economics
  5. Uni Trans
  6. Xeno Entertainment
  7. Galactic Warfare
  8. Space Militarization
  9. Mass Driver Theory
  10. Basic Rail Guns

Korath First 10 tech choices:

  1. Xeno Research
  2. Xeno Communiations
  3. Basic Logistics (bad move)
  4. Xeno Biology
  5. Space Mining
  6. Uni Trans
  7. Planet Improvements
  8. Xeno Industrial Theory
  9. Extreme Colonization
  10. Radioactive World Col

 

 

The quadrant:


The center has two key worlds that the Terran Alliance will attempt to take. It has the advantage of knowing where the planets are (because this is a pre-canned map that I've played before whereas the Korath have to explore it).

End of Year 1 status report:

Even split in galaxy.

Very close overall still.

Korath base world:


More built up but no super projects.

Terran base world:


Manufacturing capital constructed. But one tile not filled.

Conquest

In 2227, both sides that military parity. But the Terran Alliance had invasion tech. In fact, the Korath were quite slow in focusing on their Spore weapon technology.

It was a walk over.

Their ships had an attack of 2 each, mine had 3.  But they had adapted to the changing circumstances but now they only had 2 worlds to my 5.

In the end, it was over.

Where did they go wrong?

Two things in particular:

Mistake #1: Planet Construction.

VS.

Their capital world had only a 34tp production on research with no research bonus additions. Mine was at 74 X 2X with the tech capital.

Mistake #2: Their research path

They should have focused more on their weapons earlier on. They didn't get their spore weapons until too late.  Though really, looking at what techs they researched, the main problem was the home planet.

The Korath played badly. But as badly as they played, a lot was learned and it was largely in the tweaking of values that the solutions can be found.

What they did right

  1. Managed to effectively control half the quadrant.
  2. Took control of 1 of the 2 galactic resources
  3. Tactically beat back my weak areas (including taking out my resource starbase and taking over the resource)
  4. Took control of weak positions of mine (unfortunately, their inferior ships and slowness to get spore ships out allowed me to recover easily)

Back to tweaking and adding new functionality to make it smarter.


Comments (Page 1)
2 Pages1 2 
on Jan 21, 2007
cool man,.. keeping us inform like that.
on Jan 21, 2007
Great to see the Korath and Krynn have new logos. Don't think I like the outline around the Korath one too much though but they are a massive improvement over there previous logos.

As to the performance of the Korath in my beta 2b games they have possibly been the most consistent good performer. They seem to fade after a strong start in most of my games though possibly due to alignment and economy issues.
on Jan 21, 2007
They do better in big games.
on Jan 21, 2007
When I play on 2b, the Korath and Drengin both seem to have the "split" event. It's almost amusing the way that the Drengin will have a civil war around the mid-point of a war-- which causes the civilization they're at war with to easily conquer them. This probably owes to the previously-mentioned "happiness bug"...
on Jan 21, 2007
The "Financial Resources" slider makes it look like the computer has a +100% bonus on top of their regular income, is that what it is supposed to be or is it just the regular default income?
on Jan 21, 2007
It means they get 100% of their tax income -- like the player.
on Jan 22, 2007
Also building that manufactory capital is not optimum strategy imo. A factory would be more beneficial a manufactory capital should only be built on a planet that has 4 or more factory buildings/bonuses increase to 4 or more worth. This is the case if the manufactory capital bonus is 25%.

The Terran homeworld has 300% factory production including the bonus tile another factory needs to be built to make the manufactory capital worthwhile (i'm not taking into account the small eco bonus).
on Jan 22, 2007
Lenius - you don't necessarily have to wait for 4 factories to make a manufacturing capital worthwhile. Remember, the Civcapital puts out 24mp and one of the factories is on a bonus tile. So you have a total of 24 + 6x2 + 6 = 42mp right there. 25% of 42 is 10, so building a manufacturing capital at this point gives you 5/3 the mps of a third factory. But not only that, this planet has 3 mining bases giving 10mps each, bringing the total output of the planet to 72mps (as the bottom of the summary shows). 25% of that is 18, three times the output of an additional factory! Plus, the 18 points of bonus production produced by the manufacturing capital only cost half the tax revenue, so there's a if he were maxed out production he'd be producing 90 points a turn for 81bc instead of 78 points a turn for 78bc. That's a 15% increase in production for a 4% increase in spending.
Of course, you could argue that the manufacturing capital would be better placed on another planet which could then be filled with factories, leaving more room for research centers on this planet (with the tech capital). In the long run, specialization is more productive, because of the multipliers (capitals, power plants, and starbases). But sometimes having the manufacturing capital up early in the game can give you the extra boost you need to pull out to a decisive lead. Makes you appreciate how hard it is to program the AI to take all this into account...
on Jan 22, 2007
Also building that manufactory capital is not optimum strategy imo. A factory would be more beneficial a manufactory capital should only be built on a planet that has 4 or more factory buildings/bonuses increase to 4 or more worth. This is the case if the manufactory capital bonus is 25%.


Not at all. Note that the civ capital tile generates 24mp on its own--that alone would make the manufacturing capital better than a factory (which produces 6mp). Plus the manufacturing capital enhances the output coming from the asteroid mines as well. When everything is considered, the manufacturing capital is providing 18mp.
on Jan 22, 2007
Good reply posts guys very thought provoking.   

It's possible it could be worthwhile to build on the homeworld early but I doubt it. The key thing with early homeworld management imo is getting enough production to produce colony ships once every two turns imo (100% mil). The manufacturing capital takes a long time to build/very expensive to buy. Just factories are more worthwhile imo also many races do not start with the right tech for it.

I tend to move my capital at the right time to a high PQ eco/trade world so the extra capital production wouldn't last long term ethier.

If it's small map/small amount of planets building it on the initial capital planet might be right idea though (esp using the AI 3 way split production method) so was perhaps right in this case.
on Jan 22, 2007

Also building that manufactory capital is not optimum strategy imo. A factory would be more beneficial a manufactory capital should only be built on a planet that has 4 or more factory buildings/bonuses increase to 4 or more worth. This is the case if the manufactory capital bonus is 25%.

The Terran homeworld has 300% factory production including the bonus tile another factory needs to be built to make the manufactory capital worthwhile (i'm not taking into account the small eco bonus).

On this map, there were only a handful of planets. The civ capital provides immense bonuses and it is the most secure.

on Jan 22, 2007

Lenius - you don't necessarily have to wait for 4 factories to make a manufacturing capital worthwhile. Remember, the Civcapital puts out 24mp and one of the factories is on a bonus tile. So you have a total of 24 + 6x2 + 6 = 42mp right there. 25% of 42 is 10, so building a manufacturing capital at this point gives you 5/3 the mps of a third factory. But not only that, this planet has 3 mining bases giving 10mps each, bringing the total output of the planet to 72mps (as the bottom of the summary shows). 25% of that is 18, three times the output of an additional factory! Plus, the 18 points of bonus production produced by the manufacturing capital only cost half the tax revenue, so there's a if he were maxed out production he'd be producing 90 points a turn for 81bc instead of 78 points a turn for 78bc. That's a 15% increase in production for a 4% increase in spending.
Of course, you could argue that the manufacturing capital would be better placed on another planet which could then be filled with factories, leaving more room for research centers on this planet (with the tech capital). In the long run, specialization is more productive, because of the multipliers (capitals, power plants, and starbases). But sometimes having the manufacturing capital up early in the game can give you the extra boost you need to pull out to a decisive lead. Makes you appreciate how hard it is to program the AI to take all this into account...

Thank you and well said.

This is exactly why it's challenging to write the AI. Because the AI really does look at the things you just mentioned:

  1. How secure is the planet
  2. How long will it takes to get other planets to the right manufacturing level (and how secure are they)?
  3. What bonuses are on the planet
  4. What asteroid fields are nearby
  5. When/if might I research more space mining technologies
  6. How soon might I be going to war?

It always pains me when someone comes on and says "The AI built a factory on a morale bonus tile! It's stupid!" because there are so many other factors that come into play. It's not just randomly picking stuff there's a ton of questions involved.

The problems the AI have usually come down to weighting. How much should security be weighed when deciding what to build? How accurate is the security rating for the planet? (a TON of work is spent on that question -- what makes a planet a secure planet to build things on?).

 

on Jan 22, 2007
Frogboy/Draginol - you often talk about the fact that the AIs (plural ) in GalCiv are not scripted and ever since you've been working on the AIs for DA you have mentioned that allowing "extra algorithms" lets the computer plan more turns ahead, so to speak. What does this mean for Planetary Improvements? Does the AI improve its planets with a long term goal in mind (so to speak, I know AIs don't have minds...)? Or does it try to maximize output for the (limited) number of turns it can see ahead?

What about star bases? On this thread, https://forums.galciv2.com/?forumid=346&aid=134469#1077863
some players (Fuels Chief and Mumblefratz) talk about getting 10-20,000 rp a turn from a single planet by arranging 16 starbases in four sectors around one planet with max research facilities, tech capital, etc. Would you consider this an exploit, or would it be possible for the AI to plan something like this?

Of course, I guess this comes back to the original point, security. Human players who get such high scores are playing from dominant positions in which they can rest assured that their expense networks of starbases won't get destroyed by the enemy.

A lot of humans I know (including myself, sometimes) would simply quit or reload to an earlier point in the game if the AI got itself into such a secure position.

Speaking of, it would be hilarious if at some point (as a joke) someone developed an AI that would quit (ending the game for you too) or reload the game to a few turns back if you took one of its best planets... or one that would hit Ctrl-N until it got a great starting position.

Of course, if you want that grief, you just need to add multiplayer.
on Jan 22, 2007
It means they get 100% of their tax income -- like the player.


Imho you should change the description then. As it is at the moment it reads like the cpu gets additional money, which could confuse some players.

(Especially german reviewers who are to lazy to check back with you if they're right about what they write. )
on Jan 23, 2007
A small issue, there is a slight grammer problem with the Korath description int he first two screens. It should read "has made them an even more dangerous foe".

These AAR's were my favortie posts in the months or so leaing up to DA, and they continue to not disapoint!
2 Pages1 2