|
Re: AI DEATH MATCH 2
Orks vs Praetorian.
An interesting game. Again, the Praetorian expanded very fast in the early game and pushed the Orks to the ropes... and again the Praetorians lost the game! Think they need some ligth fixes in the contruction_vehicles.txt or change the setting to allow them to send the ships through knew minefields. The Orks won at the turn 320. Here the link: Orks_vs_Praetorian_B4.zip |
Re: AI DEATH MATCH 2
Tessellate vs CueCappa.
A long game. The Tessellate defeated the CueCappa before the turn 380. During the first part of the game, the CueCappa performed better than the Tessellate and was at the offensive. Also they did several holes into the Tessellate's territory. But when the Tessellate got the DN with Quantum Reactors, the game changed to their side. Here the link: Tessellate_vs_CueCappa_B5.zip |
Re: AI DEATH MATCH 2
Quote:
However, to give credits where credits are due, even without this dead end the EEE never would have won. The much higher expansion rate of the Fazrah coming with a decent and well-balanced AI is just too much. Hi Blade, Once I wrote that intel cannot decide a game. I still believe that this is true for a standard game but in this head-to-head matches the Fazrah's intel efforts are much more than only annoying. I didn't check the exact number of empty EEE colonies throughout the game (emptied by food contamination), but there are far too many of them. http://forum.shrapnelgames.com/images/icons/icon10.gif Quote:
Quite a disadvantage when the war started and probably the reason for it. Guess it was a MEE war, scores were 3.0M to 1.3M at turn 190. Not much left of this advantage at the end of the game... |
Re: AI DEATH MATCH 2
Quote:
Really I'm not sure what's the reason. It mean that the results of this contest, can't be considered 100% fair, because when 2 AIs are both strong, the starting place would decide the game. For the next rounds, the AIs will play at least 2 times against their opponents changing sides in the starting places. Think it would help to keep the fairness, but still I dislike this fact... |
Re: AI DEATH MATCH 2
The AI will search in a clockwise manner for planets to colonize IIRC. Maybe there is some bug with this but I have no clue what it could be.
|
Re: AI DEATH MATCH 2
Toron vs Gron.
The Gron defeated the Toron at the turn 330. The Toron's expansion was hurted by their starting place... but anyway at the turn 200 they had a decent game to play. But the Gron's ships with Talisman decided the battle. Here the link: Toron_vs_Gron_B6.zip |
Re: AI DEATH MATCH 2
Khrel vs Colonials.
After some turns, was pretty obvious that the Khrel had problems to colonize all of their available planets... but the Colonials were unable to take advantage of this. Then, the Khrel killed the Colonials slowly, planet after planet... The Khrel won at the turn 410. Here the link: Khrel_vs_Colonials_B6.zip |
Re: AI DEATH MATCH 2
Teslik vs Narn Regime.
The Narn Regime won at the turn 270. The Teslik had problems to colonize all the available planets, but still they has fought well, but the Narn had not problems to colonize... Here the link: Teslik_vs_Narn.zip Vikings vs Sallega. The Vikings won at the turn 290. They had not problems to defeat the Sallega. Here the link: Vikings_vs_Sallega_B1.zip Then the standings in the group B1 are: Space Vikings 3 Narn Regime 2 Sallega 1 Teslik 0 Then, the Space Vikings, Narn Regime and Sallega will play in the next round. |
Re: AI DEATH MATCH 2
Quote:
Unfortunately it does not work this way and never will against the better AIs. The Khrel learned about the punishment for having a too low expansion rate in their game against the UF, annihilation. |
Re: AI DEATH MATCH 2
Quote:
I'm not sure if this is a good idea: Entry 3 Type := Colonizer Entry 3 Planet Per Item := 300 Entry 3 Must Have At Least := 5 In my view 300 is a so high number. It mean that during a long time, the Khrel will have only 5 colony ships. It would be ok for some game settings, but in the average game, IMHO it will not help the Khrel: think they will colonize slowly. |
Re: AI DEATH MATCH 2
Quote:
That's why I like your contests so much. They open my eyes for further improvements or show me the flaws (all of my AIs have this colonizer entry). Some more work to do but work I'm looking forward to. |
Re: AI DEATH MATCH 2
Quote:
[ March 18, 2003, 23:04: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Namovans vs Pyrochette.
A game between 2 Temporal races and for the first place. The Pyrochette got a deserverd victory before the turn 289. Here the link: Namovans_vs_Pyrochette_B2.zip Rage vs Drakol. Honestly, I expected to see the Rage win. But because they started in the upper side, expected that the Drakol would have the opportunity to offer a decent challenge. I was right only in part: the Rage won but the Drakol doesn't existed for them. The Rage crushed the Drakol before the turn 150 (I can bet that it happened at the turn 141 or 142). Here the link:Rage_vs_Drakol_B2.zip Then, the standings for the group B2 are: Pyrochette 3 points Namovans 2 points Rage 1 point Drakol 0 point |
Re: AI DEATH MATCH 2
Azorani vs EEE
The EEE had not problems to defeat the Azorani, before the turn 220. Thanks God, the Azorani decided to surrender before lose all of their planets. Here the link: Azorani_vs_EEE_B3.zip Fazrah vs Earth Alliance The Fazrah played without problems starting in the Up side, and always had the initiative against the EA. Early on the game, was pretty obvious who will reach the victory... The Fazrah won at the turn 240 Here the link: Fazrah_vs_EA_B3.zip The standings in the Group B3 are: Fazrah 3 points. EEE 2 points. Earth Alliance 1 point. Praetorians 0 point. Then, the Fazrah, EEE and EA go to the next round... [ March 19, 2003, 14:10: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Hi MB
Did some testing with the Gron regarding the design switching we discussed some days ago. It has confirmed my guess that the design switching does not happen, when all of their kamikaze attack ships come with: .... Majority Weapon Family Pick 1 := 34 Majority Weapon Family Pick 2 := 2 ... However, this approach means that immediately after the TKP I is discovered the AI will use it, rendering the designs weaker in comparison to the CSM V they also could have. So, your idea of having two DN designs with different weaponry is much better. No design switching and the smaller ships still can use the best available weapon (or the weapon I would like them to use). Using your approach makes it possible that the first TKP-ships which appear are coming with a TKP V and the talisman. Great, just great. Can you or anyone else confirm that this design switching for ships only occurs several turns after the DN class is discovered? There is no switching before, e.g. between a battleship and a cruiser? I just want to be sure because regarding units it seems that the design switching happens much earlier and much more frequently. When I equip a small WP with CSMs and my medium WPs with TKPs the AI begins immediately and quite often (every second or third turn) switching back to the small WP, it does not wait until large WPs are researched. |
Re: AI DEATH MATCH 2
Quote:
Quote:
My personal guess is that related with the Weapons. My theory is that if you reach the max develop for a weapon, and also does exist other design with a different weapon, then, the AI will start to rotate the designes. [ March 19, 2003, 14:13: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Praetorian vs United Flora.
The UF had not problems to defeat the Praetorians, at the turn 153 Here the link: Praetorian_vs_UFlora_B4.zip Orks vs Aquilaeian The Aquilaeian defeated the Orks before the turn 193. Here the link: Orks_vs_Aquilaeian_B4.zip Then the standings in the group B4 are: United Flora 3 Aquilaeian 2 Orks 1 Praetorian 0 |
Re: AI DEATH MATCH 2
Piundon vs Cue Cappa.
A very fast game... The Cue Cappa slaughtered the Piundon before the turn 150. Here the link: Piundon_vs_CueCappa_B5.zip Klingon vs Tessellate. The Tessellate defeated the Klingons before the turn 250. During all the game they had the initiative, and the Klingon's with their problems to colonize (due the colony bug http://forum.shrapnelgames.com/images/icons/icon9.gif ) were unable to change their fate. Here the link: Klingon_vs_Tessellate_B6.zip The standings in the Group B5 are: Tessellate 3 Cue Cappa 2 Piundon 1 Klingon 0 [ March 19, 2003, 16:59: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Quote:
Anyway, thanks a lot for sharing your idea about the 2nd DN design, I never would have thought of that one. http://forum.shrapnelgames.com/images/icons/icon7.gif |
Re: AI DEATH MATCH 2
Quote:
What weapons and level, had the medium and small fighters? Quote:
|
Re: AI DEATH MATCH 2
Quote:
Quote:
WPs: Small WPs CSM V, Medium WPs TKP III, switching happened immediately after medium WPs were researched. CSMs V were discovered before medium WPs. That would mean that your presumption is correct. If the highest level for a certain weapon is attained the AI will not "forget" this achievement. It will always switch back to the design using this "achievement" regardless of any other - more recently - discoveries. By the way, switching back still happened from a large fighter with small TKPs III to the small one with small DUCs III. (The AI obviously has quite a good memory http://forum.shrapnelgames.com/images/icons/icon7.gif ) Ships: CSM V was discovered before Cruiser class was researched, at this time TKP II was available. TKP V was researched before battle cruiser tech. Switching occured only after DN class had been discovered (about 20 turns after that) and AI switched back to LCs not cruisers. Ship Designs: Escort to cruiser with CSM, BC to DN with TKP. That would also mean the AI switches 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 question that remains: why does it wait until DN tech is researched... Quote:
|
Re: AI DEATH MATCH 2
It seems to be always the same for the EA when they lose: They start in the bottom right and don't get enough radioactive planets. If I assign them one manually, they will fight back massively and have turned the tight again. I think the problem are the 100%-all-the-same planets for my races. http://forum.shrapnelgames.com/images/icons/icon9.gif
|
Re: AI DEATH MATCH 2
Quote:
Because can't see this problem with the current map and settings. A few points. A) Does exist systems with 140% in radioactives. Then IMHO, your races should select most of these planets as radioactive colony in those systems, unless they're lack of other thing. B) With "low bonus", is not easy to see races with lack of resources. I can't remember a race with lack of resources on this current contest. In fact specially checked the EA against the Fazrah, and didn't see that they were lack of any resource. C) You know that if the AI is lack of something, will assign planets for the resource that need. This is why I did planets with 140%/100%/100%, 100%/140%/100% and 100%/100%/140%. If the AI is lack of radioactives, will have not problems to use one of the planets with 140%/100%/100% or 100%/140%/100% to compensate the problem. [ March 19, 2003, 23:08: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Quote:
Quote:
Quote:
At some moment, the AI realize that can't continue improving the TKP. Then, to design their new fighter, take the best design with other weapon, and then, select the DUC III regardless are placed in smallest designes... Quote:
Usually, the AI upgrade their ships when reach an improvement at some component. But after some time, will not exist more improvements anymore... then, to keep the AI creating new designs, think Aaron did it. Quote:
Think all the AI modders should know this problem... would you mind to write it and our fix in a new post?? Think your english is better than my "pathetic" language skills! [ March 19, 2003, 23:23: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Colonial vs Gron.
The Gron won at the turn 200. The Colonials had not opportunity. Here the link: Colonials_vs_Gron_B6.zip Toron vs Khrel. The Toron defeated the Khrel at the turn 270. Then, does exist 3 races with the same points in the group B6, then, I need to play a Tie Break. Here the file: Khrel_vs_Gron_B6.zip Standings in the group B6. Gron 2 Khrel 2 Toron 2 Colonials 0 |
Re: AI DEATH MATCH 2
Quote:
Quote:
|
Re: AI DEATH MATCH 2
Quote:
|
Re: AI DEATH MATCH 2
Quote:
Never considered this, and then, I'll do my own tests... About the Fazrah vs EA, I have checked how the EA selected the planets: only had 2 Refining Colonies in the systems that I did for it. Why? Because in the planets.txt the EA have this option "Max Per System := 2" Then, agree that this could affect the EA behavior: they were not designed to have many Refining Colonies placed in the same system. But I have assumed that if any AI would be lack of resources, always could use a planet not specially designed for Refining Colony and still get good resources. About the problem with the "Low Bonus", I have my own guess: think the AI doesn't consider the bonus to select the kind of colony. It mean, that they could FORCE a planet to be a Mining Colony, because believe that is lack of mineral without consider the bonus... I have noted that my AIs used the planets with 70/70/70 for Mining Colony, when they still had a lot of Mineral available. |
Re: AI DEATH MATCH 2
Quote:
I think you're right. The AI didn't take the bonus factor into account, the colonizing decisions are based on the "basic" not the "bonus" income. BTW I'm quite sure that this safety device (overriding my colony calls) only takes place if the planet has a value of 50% or more. I never had an AI (no bonus and bonus games) using a planet with a min/org/rad value lesser than 50% as a mining/farming/refining colony. No matter how big its lack of resources was. |
Re: AI DEATH MATCH 2
Quote:
Quote:
|
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
Quote:
But it worked in the previous contest... Don't understand how the low bonus can affect the AI. |
Re: AI DEATH MATCH 2
Quote:
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
Quote:
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
Quote:
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 |
Re: AI DEATH MATCH 2
Quote:
edit: oh and btw.. in reply to your email a while back (can't access that right now): I am honored and of course you have permission to use any of the AI stuff that I did for the UF. glad you like them http://forum.shrapnelgames.com/images/icons/icon7.gif [ March 20, 2003, 22:48: Message edited by: Rollo ] |
Re: AI DEATH MATCH 2
Space Vikings vs Namovans.
The Space Vikings got a deserved victory against the Namovans. They won before the turn 340. Here the link: Vikings_vs_Namovans_C1.zip |
Re: AI DEATH MATCH 2
Quote:
And yes, I think you could find some interesting data. For example, I don't know if you already knew the problem described by Rexxx, about the ship designes. Quote:
This is something different: check the game between the Aquilaeian and Gron. During near of 100 turns, the Aquilaeian only built Ice colonies. Although the Aquilaeian sent to die lots of colony ships, always the new colony ship constructed were Ice colony ships... [ March 21, 2003, 00:36: Message edited by: Master Belisarius ] |
Re: AI DEATH MATCH 2
Quote:
Quote:
Always considered the UF one of the best (probably the best AI), not only for the challenge... your ideas were very original. |
Re: AI DEATH MATCH 2
I'm back...
Pyrochette vs EEE. Another long game. I'm starting to consider exclude the Pyrochette from the rest of the contest! http://forum.shrapnelgames.com/images/icons/icon8.gif Yea, near to 24 hours playing... The Pyrochette defeated the EEE by points at the Turn 500. They started with problems to colonize their available planets, and the EEE invaded some Pyrochette's systems, but slowly the Pyrochette changed the course of the game, and finished at the offensive. Here the link: Pyrochette_vs_EEE_C2.zip |
Re: AI DEATH MATCH 2
United Flora vs CueCappa.
The United Flora defeated the CueCappa at the turn 310. They had a few problems in the early game, but after some turns, they invaded the CueCappa's systems and reached a deserved victory. Here the link: UFlora_vs_CueCappa_C4.zip |
Re: AI DEATH MATCH 2
Fazrah vs Aquilaeian.
The Aquilaeian won by points at the turn 500. During the first part of the game, the Fazrah performed very well and had the initiative attacking Rove, and in fact, when I was going to sleep believed they would win. But no, at the end the Aquilaeian resisted and got the AAr system from the Fazrah. Here the link: Fazrah_vs_Aquilaeian_C3.zip |
Re: AI DEATH MATCH 2
Tessellate vs Gron.
2 Combo races (Cristalline-Religious vs Temporal-Religious). The Tessellate had the usual problems (just build Ice Colony ships during many time), but anyway they attacked the external Gron's system early on the game, then, the Gron had problems to expand too. The Tessellate won at the turn 300. Here the link: Tessellate_vs_Gron_C5.zip |
Re: AI DEATH MATCH 2
Toron vs Narn Regime.
Both races signed a Military Alliance, and continued with this treaty until the end of the game... The Narn won by points at the turn 500. Here the link: Toron_vs_Narn_C6.zip |
Re: AI DEATH MATCH 2
Vikings vs Khrel.
The Khrel reached a deserved victory at the turn 250. The Vikings had the usual problems starting in the up-side... but anyway, they were unable to stop the Khrel fleets. Here the link: Vikings_vs_Khrel_C1.zip |
| All times are GMT -4. The time now is 07:55 AM. |
Powered by vBulletin® Version 3.8.1
Copyright ©2000 - 2025, Jelsoft Enterprises Ltd.
Copyright ©1999 - 2025, Shrapnel Games, Inc. - All Rights Reserved.