Re: AI DEATH MATCH 2
 

I guess, the "overriding mechanism" is only invoked if the AI really has a negative income. As the EA has no negative income (the maintenance setting keeps it just above that) it will follow the Ai files (i.e. too few radioactive planets for this map). Well, you cannot create an AI for every map. Sometimes it works, sometimes it doesn't. http://forum.shrapnelgames.com/images/icons/icon7.gif

Re: AI DEATH MATCH 2
 

<font size="2" face="Verdana, Helvetica, sans-serif">http://forum.shrapnelgames.com/images/icons/icon7.gif Thanks.
But it worked in the previous contest...
Don't understand how the low bonus can affect the AI.

Re: AI DEATH MATCH 2
 

<font size="2" face="Verdana, Helvetica, sans-serif">With no boni the AI can quite easily run in a deficit by just adding 1 or 2 ships (or a large expensive ship like a star creator) because at least in the beginning you normal have just a surplus of round about 5k to 10k per mineral. In the bonus game, you don't hit the deficit quite as fast as you have at least 10k to 20k surplus. If the AI now adds a expensive ship, it will still have a surplus but will not another ship because of the maintenance threshold. In the no-boni game it will already have a deficit. It won't add another ship either but you are already loosing money from your storage and it will assign the next planet for the deficit resource.
The other problem is the “only 2 radioactive planets per system” setting as nearly all the good radioactive planets are in one system on this map. Nevertheless I think this setting is useful as it prevents the AI in a normal game to only build one type of colonies. Ok, I stop *****ing now. http://forum.shrapnelgames.com/images/icons/icon12.gif

Re: AI DEATH MATCH 2
 

<font size="2" face="Verdana, Helvetica, sans-serif">http://forum.shrapnelgames.com/images/icons/icon7.gif
But still don't understand how a resource unbalance makes the AI to stop build ships... If this is as you said, it's a bug, not a feature.

At the turn 100 in the game Fazrah vs EA, the EA had this resource production: Minerals 475371, Organic 78489 and Radioactives 53580.
Why then, the AI should not build more ships?? Doesn't look very smart.
Now I'm running a game, then, can't test anything. But later will change a bit the EA vehicle_construction file, to see if they would like to build more ships.

Re: AI DEATH MATCH 2
 

Ok, I was a bit unclear. The AI does not stop building ships entirely. But when they come close to the maintenance limit, they will only put one or two ships into the build queue, "dancing around that limit”. If the AI loses a ship (and gain room to the maintenance limit) it will put one more ship into the queue. It takes around 5 turns to build one ship. It will not build another ship for 5 turns as it is (with production) now at the threshold. It's no bug at all, just an inefficient use of the resources.
Let's take these numbers: Minerals 475371, Organic 78489 and Radioactives 53580.

Assume that a ship costs 25k Minerals and 10k Radioactives.
The AI will only round about build 3 ships because this will set the radioactives to 20k and which point it hits the threshold. 400k minerals are not used.
If the AI had more radioactives and less minerals, it would build more ships and would efficiently used all resources. The EA AI has to few radioactive planets and therefore is wasting mineral production that would be used if the had somewhat less minerals and more radioactives.
You see, no bug at all, just something the AI is not doing very well (resource management). With nearly all the radioactive planets in one system, the EA AI has this problem more then some other AIs because it will only take 2 radioactive planets per system, wasting other minerals that are not used. Give them 2 more radioactive planets per system and they will use up their resource much more evenly. In fact I have dedicated only one more breathable planet of the EA for radioactive production on turn 110 of the Fazra-EA match and it changed to whole game for the EA. The game is still undecided in turn 130 but very even with a slight advantage for the EA at the moment.

Re: AI DEATH MATCH 2
 

<font size="2" face="Verdana, Helvetica, sans-serif">Now I understand and agree with you.
The problem was that the AI reached the 'maintenance limit' for radioactives, not a problem related with a production unbalance itlself.

I understand your idea to limit the number of radioactive planets (and other resource planets) into a systems, but I would suggest that you should raise this number, for planets with more than 130%.
What if in a normal game, the AI colonize 2 small planets or mooons with 130% radioactives, but after research Gas Colonization, had a huge planet available to colonize into the same system, with 140% in radioactives??? The AI will lose the opportunity to use the potential of this planet, great for a Refining colony.

[ March 20, 2003, 22:16: Message edited by: Master Belisarius ]

Re: AI DEATH MATCH 2
 

Master Belisarius' and my findings about design rotating:

If the AI uses different weapons for a certain ship or unit type sooner or later it will begin to rotate designs (using obsolete (smaller) designs even if much better ones are available). If all of your attack (or whatever) ships (or units) will use always the same weapon, design rotating will not occur.

To give some examples:
Units:
Small fighters DUC III, medium fighters TKP I. Rotating back to small fighters happened immediately after medium fighters were researched. DUCs III were discovered before medium fighters.
WPs: Small WPs CSM V, Medium WPs TKP III. Rotating back to small WPs happened immediately after medium WPs were researched. CSMs V were discovered before medium WPs.
It seems, that if the highest level for a certain weapon is attained the AI will not "forget" this achievement. On the other hand the AI tries to use the newest design for a certain vehicle type without considering the level of the weapon this design has. To solve this problem it rotates between the old "achievement" and the newer (bigger) design. This rotating will take place regardless of any other - more recently - discoveries. Even if large fighters with small TKPs III are available the AI will rotate back to the small one with small DUCs III.

Ships: CSMs V were discovered before Cruiser class was researched, at this time TKPs II were available. TKPs V were researched before battle cruiser tech.
Rotating occured only after DN class had been discovered (about 20 turns after that) and AI rotated back to LCs not cruisers.
Ship Designs: Escort to cruiser with CSM, BC to DN with TKP.
It also seems that the AI rotates back to the first design which had the highest level of the first weapon (again: never forgetting the "achievement", the LC was the first design coming with a CSM V.) Exactly what it did with units. The only difference is that - as far
as we know - rotating back only occurs somewhat after the DN tech is researched.

This bad habit can be avoided by adjusting the majority weapon picks.

Using
....
Majority Weapon Family Pick 1 := 34
Majority Weapon Family Pick 2 := 2
...
for all your your attack (or whatever)ships/fighters/WPs will prevent any design rotating. The drawback is that the AI will use the first pick immediately after it is researched or the first design with it becomes available. You have to live with low levels of the TKP for lots of turns...

There is another possibility. Create 2 different designs for DNs (different weaponry) and large fighters/WPs. The AI still rotates designs. However, it does not fall back to smaller ships/fighters/WPs but only rotates between the two DN designs. It is crucial that the designs come with different weaponry, it seems that design rotating is triggered by weapons. This approach makes it possible that the smaller ships still can use the best available weapon (and not the one which is most recently discovered), that means for the example above the first TKP-ships which appear are coming with a TKP V and not with the TKP I.
For any more details about this approach check MB's AIs. (As I said: no mercy http://forum.shrapnelgames.com/images/icons/icon10.gif ).

Re: AI DEATH MATCH 2
 

I did the Tie Break between the Gron/Khrel/Toron, and thanks God, the Toron defeated the Khrel in the Last game!!!! If not I would be forced to continue playing extra games!

Khrel vs Gron.
The Gron won at the turn 500.
Link: Khrel_vs_Gron_B6TB1.zip

Gron vs Toron.
The Toron won at the turn 500 (a boring game with a military alliance until the end).
Link: Gron_vs_Toron_B6TB2.zip

Toron vs Khrel.
The Toron got a deserved victory at the turn 380.
Link: Toron_vs_Khrel_B6TB3.zip

Then the standings in the Tie Break are:

Toron 2
Gron 1
Khrel 0

Re: AI DEATH MATCH 2
 

Third Round
===========

GROUP C1
Space Vikings
Namovans
Khrel

GROUP C2
Pyrochette
EEE
Sallega

GROUP C3
Fazrah
Aquilaeian
Rage

GROUP C4
United Flora
Cue Cappa
Earth Alliance

GROUP C5
Tessellate
Gron
Orks

GROUP C6
Toron
Narn Regime
Piundon

Re: AI DEATH MATCH 2
 

Great work on this contest MB http://forum.shrapnelgames.com/images/icons/icon7.gif . I am sure a lot of valuable data is produced that way. Hopefully I can find the time and review some games, once I get my new system running.

I am not much up to date with this colony bug. Is that the same thing that occured somtimes with SE4 v1.49, but was hard to track down? I don't know if you recall this, but we talked about this quite some time ago, when I was developing the Vikings. Hopefully these games can give the data to track this bug down.

cya,
Rollo