.com.unity Forums - AI DEATH MATCH 2

.com.unity Forums (http://forum.shrapnelgames.com/index.php)

- Space Empires: IV & V (http://forum.shrapnelgames.com/forumdisplay.php?f=20)

- - AI DEATH MATCH 2 (http://forum.shrapnelgames.com/showthread.php?t=8617)

Re: AI DEATH MATCH 2

Quote:

Originally posted by Dralasite:
<blockquote>quote:<hr />... and looks like the predictions about who would win this final, are accurated.

I'm not counting any chickens until they hatch http://forum.shrapnelgames.com/images/icons/icon12.gif

MB, thanks again for doing this death match for us! If I recall correctly, this thread identified at least one bug (the colony bug), and brought up at least 3 or 4 Posts about (distinct) methods for (re)designing AIs. I'd call that a great success.<hr /></blockquote>http://forum.shrapnelgames.com/images/icons/icon7.gif Thanks.

Also, we discovered here other nasty bug: the AI doesn't consider the Bonus to know if need to ignore the AI_Planet_Types.txt or not

Really hope that at least this bug could be fixed... but I'm starting to believe that will not exist more SE4 patches.

Re: AI DEATH MATCH 2

Quote:

Originally posted by cybersol:
Thanks for the encouragement. I have finally finished the new map I have been working on. This map is based on the system layout of Master Belisarius's Death Match 2 map, but each system map has been completely reworked.

Map Use: To use put a observer as player one in the far away Zitrod system with no warp points, put AI Oppenent 1 as player two in the upper central Aldan system, and AI Opponent 2 as player three in the lower central Sutran system. Change the planet type and atmosphere to match the each players starting colony and atmosphere preferences. If you fail to do this, then some players will get additional planets that ruin the movement point and planet type balance of the map.

Goals/Changes:

A) Balance of planet type and atmosphere changed to be movement point based instead of number of systems away based.
Example: If there is a nice mineral rich rock oxygen planet 8 movement points away from the nearest home planet, then there are also nice mineral rich ANY ANY planets 8 movement points away.

B) Make the overall map symmetric to 180 degree rotation. If you picture the galaxy map and each system map inside and rotate 180 degrees about the central system between the AI opponents, then the resulting galaxy map and each system map looks identical. The AI opponents starting positions have switched sides and the human player now observes from far away on the other side, but the position of objects within the systems looks the same.
Example: If AI opponent one enters a system and turns left to find a nice juicy mineral rich ice methane planet, then AI opponent 2 can enter the equivalent system on his side of the map and turn left and find a mineral rich ice methane planet of the same juicyness.

C) Closely balance total number of facilities available for each planet type and atmosphere so the AI's are on a similar footing. There are 5 medium rock and ice planets, one of each atmosphere in every system. There are also 4 large gas planets, one of each valid atmosphere. Finally there is a tiny rock none and ice none moon like planet in each system. This gives 32 facilities per system (fps) for gas giant players and rock/ice none players, but only 28 fps for rock/ice other atmospheres. There are seven total systems on each players side so that is a 28 facility disadvantage for rock/ice other atmospheres types. So for more facility balance, 4 additional medium rock and 4 additional medium ice planets were added among the seven systems (all the average distance away with the average quality). Now rock/ice other atmospheres have 31.4 fps, gas other have 32.0, and rock/ice none have 33.7 fps. Gas player have to colonize the fewest planets for the same number of facilities while rock/ice none have to colonize the most planets. I think the overall result is closely balance, with slight biases in the same direction as they would be on random maps.

D) Add some variety in planet qualities while maintaining balance. Among the 7 systems near each player are seven types of balanced planet sets:
Planet Type, Mineral Value, Organic Value, Radioactive Value
1, 115, 115, 115
2, 30, 95, 140
3, 95, 140, 30
4, 140, 30, 95
5, 30, 30, 70
6, 30, 70, 30
7, 95, 30, 30
Type 1 planets are all in the single system away from the opponent, while type 2-4 planets are distributed among the three closest systems toward the opponent, and type 5-7 systems are distributed among the three second closet systems toward the opponent. The distribution occurs like this: rock type 2, ice type 3, and gas type 4 are in one system; then ice type 2, gas type 3, and rock type 4 are in another; and finally gas type 2, rock type 3, and ice type 4 are in the Last of the set.

E) Changed the system names so player 2(3)'s home system starts with A(S), those one away start with B(T), and two away start with D(X). The middle asteroid systems start with M.

AI Dueling Map

A very interesting map, thanks Cybersol!
I'll do some test later, and will let you know how it worked to me! http://forum.shrapnelgames.com/images/icons/icon7.gif

Re: AI DEATH MATCH 2

THE UNITED FLORA IS THE NEW CHAMPION!
=====================================

Congrats Rollo!!!! ( http://forum.shrapnelgames.com/images/icons/icon8.gif Again you defeated me! http://forum.shrapnelgames.com/images/icons/icon8.gif )

The UF defeated the Tessellate 2 more times (turn 260 starting in the down side and turn 360 starting at the upper side), and then, they won the Final 5 vs 1.

Here the link to both games:
Tessellate_vs_UFlora_FinalC.zip
UFlora_vs_Tessellate_FinalC.zip

Is evident that the UF victory on this second Death Match was more than deserved (they have played 34 games and only lost 2!).

This contest doesn't wanted to reach the absolute truth (as was demonstrated before, the game settings never are fair to all the AIs), but regardless this problem and the bugs, after the 211 games of this contest (in the first Death Match I runned 73 games), I think we have some more elements to know the most competitive AIs.

Here a link to the excel file, with the matches and results (hope I have killed all the mistakes!): AI-DEATH-MATCH-2-RESULTS.zip.zip

The AI DEATH MATCH II Final standings here!

======================================
ROUND PLACE POINTS
======================================
Final 1 United Flora 32
Final 2 Tessellate 27
SemiFinal 3 Space Vikings 23
SemiFinal 4 Aquilaeian 15
Round 6 5 CueCappa 16
Round 6 6 Narn Regime 15
Round 5 7 Pyrochette 15
Round 5 8 Toron 13
Round 4 9 Fazrah 9
Round 4 10 EEE 7
Round 4 11 Khrel 7
Round 4 12 Orks 5
Round 3 13 Gron 6
Round 3 14 Rage 4
Round 3 15 Namovans 3
Round 3 15 Earth Alliance 3
Round 3 17 Sallega 2
Round 3 17 Piundon 2
Round 2 19 Klingon 2
Round 2 20 Teslik 1
Round 2 20 Drakol 1
Round 2 20 Azorani 1
Round 2 20 Praetorians 1
Round 2 20 Colonial 1
Round 1 25 Cylon 0
Round 1 25 Corrian 0
Round 1 25 Romulan 0
Round 1 25 Toltayan 0
Round 1 25 Ekhazan 0
Round 1 25 Shadows 0
Round 1 25 Sergetti 0
Round 1 25 XiChung 0

[ May 24, 2003, 01:26: Message edited by: Master Belisarius ]

Re: AI DEATH MATCH 2

United floras 3
Tesselate 1

Re: AI DEATH MATCH 2

Quote:

Originally posted by Master Belisarius:
A very interesting map, thanks Cybersol!
I'll do some test later, and will let you know how it worked to me! http://forum.shrapnelgames.com/images/icons/icon7.gif

Re: AI DEATH MATCH 2

yay! hooray for the Flora!
Thanks a lot for running this extensive contest MB http://forum.shrapnelgames.com/images/icons/icon7.gif .

And also congrats to you and all the other modders.

Rollo

Re: AI DEATH MATCH 2

Quote:

Originally posted by Rollo:
yay! hooray for the Flora!
Thanks a lot for running this extensive contest MB http://forum.shrapnelgames.com/images/icons/icon7.gif .

And also congrats to you and all the other modders.

Rollo

http://forum.shrapnelgames.com/images/icons/icon7.gif Thanks!

Re: AI DEATH MATCH 2

Quote:

Originally posted by cybersol:
<blockquote>quote:<hr />Originally posted by Master Belisarius:
A very interesting map, thanks Cybersol!
I'll do some test later, and will let you know how it worked to me! http://forum.shrapnelgames.com/images/icons/icon7.gif

Please I would love to hear your thoughts on the map.

And congratulations on finishing the running of the second death match. I thought it was an awesome undertaking, with loads of potential value to the modders and the players. Thanks so much!<hr /></blockquote>Thanks Cybersol!
I was running some games with your map. Really liked your job... it's more balanced than the map that I did. Would liked to have this map to run the contest! http://forum.shrapnelgames.com/images/icons/icon7.gif
But thanks anyway, because will use it in my game test...

PS: the "colony bug" still happen to the races that start in the upper side...

[ May 26, 2003, 02:26: Message edited by: Master Belisarius ]

Re: AI DEATH MATCH 2

Quote:

Originally posted by Master Belisarius:
Thanks Cybersol!

I was running some games with your map. Really liked your job... it's more balanced than the map that I did. Would liked to have this map to run the contest! http://forum.shrapnelgames.com/images/icons/icon7.gif
But thanks anyway, because will use it in my game test...

PS: the "colony bug" still happen to the races that start in the upper side...

I'm glad you like it. I was inspired by the Death Matches, SE4Batch, and your Version of the map. So it would not exist without you. You can use it for the next Death Match if your computer ever recovers from this one http://forum.shrapnelgames.com/images/icons/icon12.gif

I uploaded it right when I finished it and I'm running SE4Batch on an OLD machine, so I've only done 4 games on it my self. 2 on each side for the Sallega vs. Gron. The Sallega definitely tend to experience the colony bug when on the upper side. But the Gron do not really.

Only two trends I noticed. One was that the Sallega use LR Colony ships with no cargo storage and these tend to end up filled with all sorts of mines and fighters. The gron use much higher capacity colony ships and although they sometimes get some non-population cargo, they rarely get completely filled. Does the colony bug correspond with the amount of trash taking up space in a races colony ships? The other trend was that races like the Gron and United Flora that only use a very limited number of colony ships tend to have less severe effects from the bug. Has your vast experience with this bug told you anything?

[ May 26, 2003, 13:14: Message edited by: cybersol ]

Re: AI DEATH MATCH 2

Quote:

Also, we discovered here other nasty bug: the AI doesn't consider the Bonus to know if need to ignore the AI_Planet_Types.txt or not

Really hope that at least this bug could be fixed... but I'm starting to believe that will not exist more SE4 patches.

Sadly I think you may be right. Certianly I wouldn't expect any patch until Star Fury is out. After that maybe SEV will be started? Who knows?