It is currently 18 Apr 2024, 23:14
   
Text Size

Good/bad AI behavior

A version of WOTC's game by telengard

Moderators: telengard, CCGHQ Admins

Good/bad AI behavior

Postby telengard » 05 Dec 2009, 17:40

I'm curious to hear about player's experiences with the AI, both good and bad. It will help me with tweaking the evaluation code.

Here's a few good things I've seen that actually surprised me a bunch:

Had a creature in C4 (after Vortex'ing) and the AI spawned a Fleshless Reaper in C5 and activated his Voodoo Manipulator's (in B3 if I remember right) Scare to put the creature into the Reaper's Ambush to kill it.

Second one is use of Lady of the Flies's Transpose Enemy/Ally to trigger Failed Experiment's Reactive ability after moving it out of her cell. It sure fits its name as it blew up all alone. :)

Some bad things I see with the AI and haven't fully figured out some of the reasons for doing so:

Use of Appease. The AI will spawn a creature with Appease even if it has no other creature to sacrifice which unless you are running a graveyard manip warband is pretty useless.

I also see the AI sometimes shift backwards (very rare) but it really shouldn't. My guess is that it determines there is a very good chance it will be destroyed so it backs off.

The AI is better now about NOT over-extending itself for conquest points to win the turn, which I'm happy about. I always hated seeing it aggressively go for the 3/4/5 scoring cells when it didn't need to.

What are some things you've seen while playing against the AI?

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby melkor75 » 10 Dec 2009, 21:20

Gimme some time (still busy at work) and i hope i can give u some feedback about it,

cheers!
mm75
melkor75
 
Posts: 68
Joined: 09 Oct 2009, 08:21
Has thanked: 0 time
Been thanked: 0 time

Re: Good/bad AI behavior

Postby telengard » 23 Jan 2010, 23:19

So I've done some investigating into the Appease "issue". This is when the AI spawns a creature w/ Appease and only has itself to sacrifice when other piecess would be better to spawn. Not exactly a smart move by the AI. :)

The issue has to do with the default tree depth, something called the "horizon" effect when doing minmax AI, and the not 1 for 1 ("you go i go") nature of Dreamblade (and ones like it, i.e. MtG).

In a nutshell when the AI chooses to spawn a piece, such as the Knight of Autumn gate it takes up many actions to complete the ability reaching the end of the tree depth so it won't "look" further. Other choices get do further and allow possibly the other player to make moves which will show lower scores as a result.

I know it's a little confusing but it has to do with not every single choice being a single action. Something like spawning a creature w/ Appease involves doing the spawn, resolving Appease, choosing a piece to sacrifice, and doing the sacrifice. Spawning some other piece might be one action.

In the next release I will add support for increasing the tree depth. A way to avoid this altogether (albeit it will take longer to think) is to have a very high tree depth and a phase depth of 2. This would help somewhat. When I cranked up the tree depth to past 6 it would not make that choice.

Sorry for all the technical details... but that's what's going on. :)

Hmmm, this might be a good idea for a topic in the Developer's corner...

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby melkor75 » 27 Jan 2010, 09:36

I'm more than busy at work and cannot help right now mate with testing, sorry.

I hope i can give u some feedbacks soon anyway. It's always nice to know u're working on this wonderful piece of art :wink:

mm75
melkor75
 
Posts: 68
Joined: 09 Oct 2009, 08:21
Has thanked: 0 time
Been thanked: 0 time

Re: Good/bad AI behavior

Postby telengard » 10 Mar 2010, 02:44

Anyone seeing better behavior of the AI during the spawn phase w/ the latest version? i.e. not skipping spawning creatures?

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby telengard » 27 Dec 2010, 04:41

Still hoping to get some feedback on the AI. Is the latest version faster for warbands you are playing? I'm still working hard on the Hybrid and Monte Carlo algorithms. Even at extremely high number of plays they make moves that really surprise me (seemingly bad ones). They do it consistently, so maybe long term they are decent moves. They also make excellent moves that the minmax just wouldn't do. The hybrid was supposed to give the best of both worlds but I need to figure out why the AI isn't "sensing" beneficial things like spawning columns, etc.

On the good AI side, I played a game tonight and the AI really surprised me. One one turn it spawned Carrion Hulk and put Tattooed Squashbug in the graveyard to pay for Corpsebuilt. Next turn the Squashbug came out w/ Death Chanter's Animate Dead 3, and then its Shapeshifter ability was used next turn to put War Chief in D3 which it was building up a stronghold at.

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby melkor75 » 07 Jan 2011, 11:25

Ciao Telengard,

a couple of days to play your great game. Please find below a game log. I had an issue at very first turn of a game where a blind envoy didn't move on a turn, weird indeed.

** Dreamblade (wxWidgets version "1.0.38") started **

Seed: 1294398997
Number of threads: 0
Silver player -> AI
Silver AI style: 4
Minimax/Monte Carlo hybrid algorithm being used:
Phase depth: 2
Tree depth: 6
Num iterations: 200
Num turns: 4
Gold player -> Human

Silver warband:
Hawk-Eyed Instigator
Zungar Blademaster
Clawclan Scout
Blind Envoy
Seraph
Zungar Winddancer
Lunar Handmaiden
Skydancer
Zungar Citadel
Knight of Tomorrow
Shimmersword Merc
Shimmersword Sergeant
Bonecarved Idol
Pearlthorn Gargoyle
Elite Gladiator
Malefic Steamroller

Gold warband:
Lord Slobber
Cannibal Pariah
Acolyte of Pain
Walking Worms
Hellshrieker
Pale Horse
Unwishing Well
Night Queen
Lady of the Fang
Medusa
Infernal Preacher
Boogeyman
Beetleback Mass
Carrion Hulk
The Sweeper
Dreadmorph Ogre

Player Gold is executing "Start game"
Player Silver is executing "Roll for initiative (Silver)"
Player Gold is executing "Roll for initiative (Gold)"
Player Silver is executing "Rerolling initiative : First turn 1 rolled"
Player Silver is executing "Roll for initiative (Silver)"
Player Silver is executing "Proceed to next phase"
AI Silver monte carlo simulations for action Spawn Blind Envoy (C) in [A5] had 93 win(s)
AI Silver monte carlo simulations for action Spawn Clawclan Scout (B) in [A5] had 84 win(s)
AI Silver monte carlo simulations for action Spawn Knight of Tomorrow (F) in [A5] had 83 win(s)
AI Silver monte carlo simulations for action Spawn Seraph (D) in [A5] had 104 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [A5] had 87 win(s)
AI Silver monte carlo simulations for action Spawn Zungar Blademaster (A) in [A5] had 101 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 31 win(s)
Multiple matches, or none, doing minimax: orig # 7 updated # 2
Player Silver is executing "Spawn Seraph (D) in [A5]"
AI Silver monte carlo simulations for action Spawn Clawclan Scout (B) in [A5] had 99 win(s)
AI Silver monte carlo simulations for action Spawn Zungar Blademaster (A) in [A5] had 94 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 78 win(s)
Multiple matches, or none, doing minimax: orig # 3 updated # 2
Player Silver is executing "Spawn Clawclan Scout (B) in [A5]"
Player Silver is executing "Proceed to next phase"
** Dreamblade (wxWidgets version "1.0.38") started **

Seed: 1294399070
Number of threads: 0
Silver player -> AI
Silver AI style: 4
Minimax/Monte Carlo hybrid algorithm being used:
Phase depth: 2
Tree depth: 6
Num iterations: 200
Num turns: 4
Gold player -> Human

Silver warband:
Malefic Steamroller
Clawclan Scout
Shimmersword Sergeant
Pearlthorn Gargoyle
Elite Gladiator
Bonecarved Idol
Blind Envoy
Skydancer
Seraph
Knight of Tomorrow
Zungar Citadel
Lunar Handmaiden
Shimmersword Merc
Hawk-Eyed Instigator
Zungar Winddancer
Zungar Blademaster

Gold warband:
Lord Slobber
Cannibal Pariah
Unwishing Well
Night Queen
Lady of the Fang
Medusa
Pale Horse
Carrion Hulk
Infernal Preacher
Boogeyman
Beetleback Mass
Dreadmorph Ogre
The Sweeper
Acolyte of Pain
Hellshrieker
Walking Worms

Player Gold is executing "Start game"
Player Silver is executing "Roll for initiative (Silver)"
Player Gold is executing "Roll for initiative (Gold)"
Player Silver is executing "Proceed to next phase"
AI Silver monte carlo simulations for action Spawn Blind Envoy (C) in [A5] had 101 win(s)
AI Silver monte carlo simulations for action Spawn Clawclan Scout (B) in [A5] had 85 win(s)
AI Silver monte carlo simulations for action Spawn Knight of Tomorrow (F) in [A5] had 62 win(s)
AI Silver monte carlo simulations for action Spawn Seraph (D) in [A5] had 94 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [A5] had 92 win(s)
AI Silver monte carlo simulations for action Spawn Zungar Blademaster (A) in [A5] had 91 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 35 win(s)
Multiple matches, or none, doing minimax: orig # 7 updated # 4
Player Silver is executing "Spawn Blind Envoy (C) in [A5]"
AI Silver monte carlo simulations for action Spawn Zungar Blademaster (A) in [A5] had 97 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 77 win(s)
Player Silver is executing "Spawn Zungar Blademaster (A) in [A5]"
Player Silver is executing "Proceed to next phase"
Player Gold is executing "Spawn Cannibal Pariah (b) in [E1]"
Player Gold is executing "Spawn Walking Worms (d) in [E1]"
Player Gold is executing "Proceed to next phase"
Player Silver is executing "Choose to shift for this action phase"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B5] had 83 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A4] had 88 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [B5] had 86 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A4] had 85 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 78 win(s)
Multiple matches, or none, doing minimax: orig # 5 updated # 5
Player Silver is executing "Shift Blind Envoy (C) to [B5]"
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [B5] had 83 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A4] had 92 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 77 win(s)
Multiple matches, or none, doing minimax: orig # 3 updated # 2
Player Silver is executing "Shift Zungar Blademaster (A) to [B5]"
Player Silver is executing "Proceed to next phase"
Player Silver is executing "Choose to shift for this action phase"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B4] had 89 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [C5] had 72 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A5] had 67 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [B4] had 75 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [C5] had 75 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A5] had 62 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 73 win(s)
Player Silver is executing "Shift Blind Envoy (C) to [B4]"
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [B4] had 85 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [C5] had 90 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A5] had 91 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 73 win(s)
Multiple matches, or none, doing minimax: orig # 4 updated # 3
Player Silver is executing "Shift Zungar Blademaster (A) to [B4]"
Player Silver is executing "Proceed to next phase"
Player Gold is executing "Choose to shift for this action phase"
Player Gold is executing "Shift Cannibal Pariah (b) to [E2]"
Player Gold is executing "Shift Walking Worms (d) to [E2]"
Player Gold is executing "Proceed to next phase"
Player Gold is executing "Choose to shift for this action phase"
Player Gold is executing "Shift Cannibal Pariah (b) to [D2]"
Player Gold is executing "Shift Walking Worms (d) to [D2]"
Player Gold is executing "Proceed to next phase"
Player Silver is executing "Proceed to next phase"
Player Silver is executing "Roll for initiative (Silver)"
Player Gold is executing "Roll for initiative (Gold)"
Player Gold is executing "Proceed to next phase"
Player Gold is executing "Proceed to next phase"
AI Silver monte carlo simulations for action Spawn Clawclan Scout (B) in [B5] had 128 win(s)
AI Silver monte carlo simulations for action Spawn Clawclan Scout (B) in [A5] had 118 win(s)
AI Silver monte carlo simulations for action Spawn Hawk-Eyed Instigator (J) in [B5] had 103 win(s)
AI Silver monte carlo simulations for action Spawn Hawk-Eyed Instigator (J) in [A5] had 111 win(s)
AI Silver monte carlo simulations for action Spawn Knight of Tomorrow (F) in [B5] had 115 win(s)
AI Silver monte carlo simulations for action Spawn Knight of Tomorrow (F) in [A5] had 105 win(s)
AI Silver monte carlo simulations for action Spawn Lunar Handmaiden (H) in [B5] had 113 win(s)
AI Silver monte carlo simulations for action Spawn Lunar Handmaiden (H) in [A5] had 113 win(s)
AI Silver monte carlo simulations for action Spawn Seraph (D) in [B5] had 117 win(s)
AI Silver monte carlo simulations for action Spawn Seraph (D) in [A5] had 120 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [B5] had 118 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [A5] had 121 win(s)
AI Silver monte carlo simulations for action Spawn Zungar Winddancer (I) in [B5] had 121 win(s)
AI Silver monte carlo simulations for action Spawn Zungar Winddancer (I) in [A5] had 104 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 104 win(s)
Multiple matches, or none, doing minimax: orig # 15 updated # 6
Player Silver is executing "Spawn Clawclan Scout (B) in [B5]"
AI Silver monte carlo simulations for action Spawn Seraph (D) in [B5] had 127 win(s)
AI Silver monte carlo simulations for action Spawn Seraph (D) in [A5] had 116 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [B5] had 128 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [A5] had 125 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 110 win(s)
Multiple matches, or none, doing minimax: orig # 5 updated # 3
Player Silver is executing "Spawn Skydancer (E) in [B5]"
Player Silver is executing "Proceed to next phase"
Player Gold is executing "Choose to shift for this action phase"
** Dreamblade (wxWidgets version "1.0.38") started **

Seed: 1294399302
Number of threads: 0
Silver player -> AI
Silver AI style: 4
Minimax/Monte Carlo hybrid algorithm being used:
Phase depth: 2
Tree depth: 6
Num iterations: 200
Num turns: 4
Gold player -> Human

Silver warband:
Clawclan Scout
Knight of Tomorrow
Zungar Citadel
Lunar Handmaiden
Blind Envoy
Skydancer
Seraph
Zungar Blademaster
Zungar Winddancer
Hawk-Eyed Instigator
Malefic Steamroller
Bonecarved Idol
Shimmersword Merc
Pearlthorn Gargoyle
Elite Gladiator
Shimmersword Sergeant

Gold warband:
Pale Horse
Walking Worms
Lord Slobber
Cannibal Pariah
Acolyte of Pain
Dreadmorph Ogre
Hellshrieker
Carrion Hulk
Infernal Preacher
Boogeyman
Unwishing Well
Night Queen
Lady of the Fang
Medusa
The Sweeper
Beetleback Mass

Player Gold is executing "Start game"
Player Silver is executing "Roll for initiative (Silver)"
Player Gold is executing "Roll for initiative (Gold)"
Player Silver is executing "Rerolling initiative : First turn tie"
Player Silver is executing "Roll for initiative (Silver)"
Player Gold is executing "Roll for initiative (Gold)"
Player Silver is executing "Proceed to next phase"
AI Silver monte carlo simulations for action Spawn Blind Envoy (C) in [A5] had 90 win(s)
AI Silver monte carlo simulations for action Spawn Clawclan Scout (B) in [A5] had 78 win(s)
AI Silver monte carlo simulations for action Spawn Knight of Tomorrow (F) in [A5] had 66 win(s)
AI Silver monte carlo simulations for action Spawn Seraph (D) in [A5] had 99 win(s)
AI Silver monte carlo simulations for action Spawn Skydancer (E) in [A5] had 98 win(s)
AI Silver monte carlo simulations for action Spawn Zungar Blademaster (A) in [A5] had 85 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 41 win(s)
Multiple matches, or none, doing minimax: orig # 7 updated # 3
Player Silver is executing "Spawn Blind Envoy (C) in [A5]"
AI Silver monte carlo simulations for action Spawn Zungar Blademaster (A) in [A5] had 86 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 63 win(s)
Player Silver is executing "Spawn Zungar Blademaster (A) in [A5]"
Player Silver is executing "Proceed to next phase"
Player Gold is executing "Spawn Cannibal Pariah (b) in [E1]"
Player Gold is executing "Spawn Walking Worms (d) in [E1]"
Player Gold is executing "Proceed to next phase"
Player Silver is executing "Choose to shift for this action phase"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B5] had 72 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A4] had 75 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [B5] had 85 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A4] had 98 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 80 win(s)
Player Silver is executing "Shift Zungar Blademaster (A) to [A4]"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B5] had 88 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A4] had 88 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 105 win(s)
Player Silver is executing "Proceed to next phase"
Player Silver is executing "Choose to shift for this action phase"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B5] had 78 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A4] had 78 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [B4] had 92 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A3] had 97 win(s)
AI Silver monte carlo simulations for action Shift Zungar Blademaster (A) to [A5] had 71 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 75 win(s)
Multiple matches, or none, doing minimax: orig # 6 updated # 2
Player Silver is executing "Shift Zungar Blademaster (A) to [B4]"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B5] had 75 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A4] had 82 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 89 win(s)
Multiple matches, or none, doing minimax: orig # 3 updated # 2
Player Silver is executing "Shift Blind Envoy (C) to [A4]"
Player Silver is executing "Proceed to next phase"
Player Gold is executing "Choose to shift for this action phase"


I'll be back with other issue i may find ... (is there a way to attach a file here to avoid long posts?),

mm75
melkor75
 
Posts: 68
Joined: 09 Oct 2009, 08:21
Has thanked: 0 time
Been thanked: 0 time

Re: Good/bad AI behavior

Postby telengard » 07 Jan 2011, 18:21

Hi again melkor,

Thanks for the report.

It looks like you were using the hybrid AI, that would explain it sometimes not moving a piece when it probably should. I still can't recommend using that flavor of AI yet. It has issues unless a *VERY* large number of interations are done. I still recommend using straight up MinMax, which in the last release got a lot faster and the next release even better (I'll be doing a release probably this weekend).

This seems to be the move (or lack of one as the case may be) you saw, AI saw that going to the next phase did better:

Code: Select all
Player Silver is executing "Shift Zungar Blademaster (A) to [A4]"
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [B5] had 88 win(s)
AI Silver monte carlo simulations for action Shift Blind Envoy (C) to [A4] had 88 win(s)
AI Silver monte carlo simulations for action Proceed to next phase had 105 win(s)
Player Silver is executing "Proceed to next phase"
With only 200 iterations, it's really not enough given all of the potential combinations that come up. Hybrid will play better if that number is jacked up a lot higher, but AI moves could take quite a bit longer.
Feel free to try it out.

Did you set the AI to hybrid, or did it default to that? As far as I can see in the code, the default should still be MinMax.

Not sure if you can attach files or not. I thought you could attach, but certain extensions (like .txt) may be blocked. Maybe Huggybaby will see this post and clarify.

Pretty cool warbands you are using, did you make those yourself?

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby Huggybaby » 08 Jan 2011, 03:35

re: attaching files
We do block lots of extensions, but we enable them as needed and there are probably some that should be enabled now that aren't. So, if there's an extension you want to use but can't, just let me know.
User avatar
Huggybaby
Administrator
 
Posts: 3205
Joined: 15 Jan 2006, 19:44
Location: Finally out of Atlanta
Has thanked: 696 times
Been thanked: 594 times

Re: Good/bad AI behavior

Postby telengard » 08 Jan 2011, 04:13

Huggybaby wrote:re: attaching files
We do block lots of extensions, but we enable them as needed and there are probably some that should be enabled now that aren't. So, if there's an extension you want to use but can't, just let me know.
The files that would be posted for debugging issues w/ the game would be called "game.log". I just tried to attach a test one and it said "The extension log is not allowed.". I can always change the extension to something that is supported if that is easier/more secure.

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby Huggybaby » 08 Jan 2011, 05:19

You should be able to attach log files now.
User avatar
Huggybaby
Administrator
 
Posts: 3205
Joined: 15 Jan 2006, 19:44
Location: Finally out of Atlanta
Has thanked: 696 times
Been thanked: 594 times

Re: Good/bad AI behavior

Postby telengard » 08 Jan 2011, 15:48

Huggybaby wrote:You should be able to attach log files now.
Thanks Huggybaby!

~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Re: Good/bad AI behavior

Postby melkor75 » 10 Jan 2011, 11:38

Great,

so i hope to test again with "MinMax" AI and i get back here with log file attached directly.

I tested "Human Ai for those 2 parties" and they are pretty tied, as i won 5 to 4 and quite all the matches was quite close ...

I'm curious to see against "your" AI, i'll get back here with the results :wink:

Keep it up mate!

mm75
melkor75
 
Posts: 68
Joined: 09 Oct 2009, 08:21
Has thanked: 0 time
Been thanked: 0 time

Re: Good/bad AI behavior

Postby melkor75 » 10 Jan 2011, 14:34

Here a first issue ...

AI could ONLY spawn the Zungar Citadel Location as it is the only mini it has, but it didn't spawn it.

mm75

(damned...it seems i can't attch the log file :( )
melkor75
 
Posts: 68
Joined: 09 Oct 2009, 08:21
Has thanked: 0 time
Been thanked: 0 time

Re: Good/bad AI behavior

Postby telengard » 10 Jan 2011, 16:55

melkor75 wrote:Great,

so i hope to test again with "MinMax" AI and i get back here with log file attached directly.

I tested "Human Ai for those 2 parties" and they are pretty tied, as i won 5 to 4 and quite all the matches was quite close ...

I'm curious to see against "your" AI, i'll get back here with the results :wink:

Keep it up mate!

mm75
Hmm, sounds like the AI is doing a half decent job. :)

thanks,
~telengard
Author of Dreamblade:
viewtopic.php?f=51&t=1215
User avatar
telengard
DEVELOPER
 
Posts: 379
Joined: 23 May 2009, 23:04
Has thanked: 2 times
Been thanked: 27 times

Next

Return to Dreamblade

Who is online

Users browsing this forum: No registered users and 20 guests


Who is online

In total there are 20 users online :: 0 registered, 0 hidden and 20 guests (based on users active over the past 10 minutes)
Most users ever online was 4143 on 23 Jan 2024, 08:21

Users browsing this forum: No registered users and 20 guests

Login Form