Matrix Games Forums

Forums  Register  Login  Photo Gallery  Member List  Search  Calendars  FAQ 

My Profile  Inbox  Address Book  My Subscription  My Forums  Log Out

RE: AI craziness

 
View related threads: (in this forum | in all forums)

Logged in as: Guest
Users viewing this topic: none
  Printable Version
All Forums >> [Current Games From Matrix.] >> [World War II] >> War In The Pacific - Struggle Against Japan 1941 - 1945 >> RE: AI craziness Page: <<   < prev  1 [2] 3 4   next >   >>
Login
Message << Older Topic   Newer Topic >>
RE: AI craziness - 7/13/2004 8:47:38 PM   
The Gnome


Posts: 1233
Joined: 5/17/2002
From: Philadelphia, PA
Status: offline
quote:

It was toned down because it was making the AI completely passive. Can't have it both ways. The AI either needs a kick to get going or a kick to play safe. You can't have it both ways at the same time. My little adventure against the AI in Burma caused the AI to get kicked to be more aggressive. Now you see it is a little too agressive at times.

It is a very tough balancing act. Only large amounts of play paint an accurate picture of where it sits and with all the learning curve issues at the moment, now it probably not a good time to adjust this. Perhaps in a month from now once people are up to speed.


The AI can be built to be aggressive AND not go into death spirals that render the game unplayable. That is just not a valid assesment. Sending ships continually into a harbor that should be closed is not matter of aggressive or passive, it's a flaw that needs to be addressed.

You obviously feel that this is some uncrackable nut like the speed of light or something but it's just not that difficult.

< Message edited by The Gnome -- 7/13/2004 1:48:37 PM >

(in reply to Mr.Frag)
Post #: 31
RE: AI craziness - 7/13/2004 8:53:14 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: The Gnome
You obviously feel that this is some uncrackable nut like the speed of light or something but it's just not that difficult.


Not according to Gary Grigsby.

_____________________________


(in reply to The Gnome)
Post #: 32
RE: AI craziness - 7/13/2004 8:57:10 PM   
Damien Thorn

 

Posts: 1107
Joined: 7/24/2003
Status: offline
quote:

ORIGINAL: Mr.Frag

It was toned down because it was making the AI completely passive.


If the decision would have been for the AI to not try to supply the base then, instead, it should try to supply the base with a heavy escort. If that is not possible, it should try to evacuate the base or suppress the enemy base that is making the supply dangerous. I want an aggressive AI, but not a suicidal one.

(in reply to Mr.Frag)
Post #: 33
RE: AI craziness - 7/13/2004 9:00:25 PM   
The Gnome


Posts: 1233
Joined: 5/17/2002
From: Philadelphia, PA
Status: offline
quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: The Gnome
You obviously feel that this is some uncrackable nut like the speed of light or something but it's just not that difficult.


Not according to Gary Grigsby.


He shouldn't have made a single player version if that's the case - or taken a step back and hired someone capable of doing the job.

(in reply to Nikademus)
Post #: 34
RE: AI craziness - 7/13/2004 9:02:46 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: The Gnome


He shouldn't have made a single player version if that's the case - or taken a step back and hired someone capable of doing the job.


No single player option: not good business decision

Someone more capable: Good luck finding one.

< Message edited by Nikademus -- 7/13/2004 7:03:47 PM >


_____________________________


(in reply to The Gnome)
Post #: 35
RE: AI craziness - 7/13/2004 9:29:26 PM   
The Gnome


Posts: 1233
Joined: 5/17/2002
From: Philadelphia, PA
Status: offline
quote:

No single player option: not good business decision

So do a poor job and say you have it.... that wins friends.

quote:

Someone more capable: Good luck finding one.

GG is a fine game programmer but there's always somebody better. His assets from what I've seen is not on the technical end but building the game system and marrying to a very historical OOB. He combines that with very good technical skills.

AI does not appear to be one of his strengths so yeah, maybe he should find someone better at it.

(in reply to Nikademus)
Post #: 36
RE: AI craziness - 7/13/2004 9:43:14 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: The Gnome
You obviously feel that this is some uncrackable nut like the speed of light or something but it's just not that difficult.


Not according to Gary Grigsby.


Well there you go.... the God of Formula obviously hasn't got a clue when it comes AI programming....

(in reply to Nikademus)
Post #: 37
RE: AI craziness - 7/13/2004 9:44:40 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: The Gnome

So do a poor job and say you have it.... that wins friends.



I dont believe Gary did a poor job. I am saying that programming an AI in a wargame is extremely difficult and few games pull it off. Combat Mission is one of my most recent aquisitions....a truely outstanding piece of work. AI play rather sucks though. Did they do a poor job too?

quote:


GG is a fine game programmer but there's always somebody better. His assets from what I've seen is not on the technical end but building the game system and marrying to a very historical OOB. He combines that with very good technical skills.


Didn't say there wasn't. I said "good luck finding one" especially for a game as large and complex as this one. If/when that person steps forward and puts his money where his mouth is, i'll be the first to shake his hand. Until them, comments such as "This or that should be easy....etc etc" are nothing but talk.

quote:



AI does not appear to be one of his strengths so yeah, maybe he should find someone better at it.



Based on results one could make such a assertation towards most of the programming community vis-a-vis wargames. I'm not a programmer, but as a tester i'm familiar with how things often dont go the way they might be expected, especially the more complex the coding gets. I also know Gary spent a long time on the AI routines. Therefore while i as much as anyone else would love to see a better AI, i wont condemn him or make disparaging comments.

As i said originally.....its a thorny problem. The AI is improved from UV but challenges remain.

_____________________________


(in reply to The Gnome)
Post #: 38
RE: AI craziness - 7/13/2004 9:44:45 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: Mr.Frag

quote:

If it's there it definately seems not to be working... or doesn't have a strong enough say in the AI decision tree. I think if this got tuned the ship->port->sunk death spiral could be avoided.


It was toned down because it was making the AI completely passive. Can't have it both ways. The AI either needs a kick to get going or a kick to play safe. You can't have it both ways at the same time. My little adventure against the AI in Burma caused the AI to get kicked to be more aggressive. Now you see it is a little too agressive at times.

It is a very tough balancing act. Only large amounts of play paint an accurate picture of where it sits and with all the learning curve issues at the moment, now it probably not a good time to adjust this. Perhaps in a month from now once people are up to speed.


I still fail to see the relation between the two. An AI can be made to play aggresively without going into death spiral strategies. If it can't, well then you have a fundamentally fatal design flaw in the entire AI engine...

(in reply to Mr.Frag)
Post #: 39
RE: AI craziness - 7/13/2004 9:47:12 PM   
The Gnome


Posts: 1233
Joined: 5/17/2002
From: Philadelphia, PA
Status: offline
quote:

ORIGINAL: ZOOMIE1980

quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: The Gnome
You obviously feel that this is some uncrackable nut like the speed of light or something but it's just not that difficult.


Not according to Gary Grigsby.


Well there you go.... the God of Formula obviously hasn't got a clue when it comes AI programming....


I think he has a good clue actually... the AI is painfully, excrutiatingly close to being good. There are just a couple of death spirals to remove and it will be challenging through the entire game.

I just get frustrated when I get the sense that people think that last step is as impossible as solving the Grand Unification theory of physics.... the AI is 90% there.

(in reply to ZOOMIE1980)
Post #: 40
RE: AI craziness - 7/13/2004 9:49:58 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: The Gnome

quote:

No single player option: not good business decision

So do a poor job and say you have it.... that wins friends.

quote:

Someone more capable: Good luck finding one.

GG is a fine game programmer but there's always somebody better. His assets from what I've seen is not on the technical end but building the game system and marrying to a very historical OOB. He combines that with very good technical skills.

AI does not appear to be one of his strengths so yeah, maybe he should find someone better at it.


I remember dealing with this man from back in the 1980's.Brilliant designer, sixth sense when it comes to balanced game formulas. Just KNOWS what works and what doesn't. But he's a walking brick as a programmer. Apparently he and his minions are still tightly locked into their completely obsolete hard-coded fixed array based programming models. That was necessary back in the Apple II days when all you had to code in was Applesoft Basic, but not even desirable now. As for AI's, well, I don't think he ever got into that very much. I completely rewrote the AI's of many of his early Apple II vintage games once I got the source code to them,from the ground up, focusing on fairly simplistic techiniques like weighted script branching, intelligent pathing and such, and yes, it STILL ran on an Apple IIC!!!

< Message edited by ZOOMIE1980 -- 7/13/2004 7:51:21 PM >

(in reply to The Gnome)
Post #: 41
RE: AI craziness - 7/13/2004 10:00:35 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: The Gnome

quote:

ORIGINAL: ZOOMIE1980

quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: The Gnome
You obviously feel that this is some uncrackable nut like the speed of light or something but it's just not that difficult.


Not according to Gary Grigsby.


Well there you go.... the God of Formula obviously hasn't got a clue when it comes AI programming....


I think he has a good clue actually... the AI is painfully, excrutiatingly close to being good. There are just a couple of death spirals to remove and it will be challenging through the entire game.

I just get frustrated when I get the sense that people think that last step is as impossible as solving the Grand Unification theory of physics.... the AI is 90% there.


Yes, this is the IRRITATING take you get from the ENTIRE Matrix group! You have some fairly simple AI complaints, and because it is the AI and all these people are 100% PBEM players, they dump it all in the refuse bin as "undoable". Yes, this AI is pretty close, and are not talking a Deep Blue effort here one would think.

However, I think what we are seeing is these guys have locked themselves into a pretty rigid, inflexible basic design here, to the point that what seems to professional programmers from outside that appear to be relatively minor, almost trivial changes, are for some reason, almost undoable!

I remember several weeks back a poster thought it would be a very good feature to allow the user naming of TF's. Now to me, that means adding a very simple edit field to the TF dialog and maybe a new string data member to the TF class or struct, and then a line of code to display the TF name, rather than number wherever the TF info is displayed. Trivial, really. However, the response was that to add such a highly unstable element to the game at such a late stage was impossible!!! Say what???? The discussion went downhill from there......basically EVERYTHING seems to be a mission to Mars to get it added....

(in reply to The Gnome)
Post #: 42
RE: AI craziness - 7/13/2004 10:05:35 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: ZOOMIE1980
Yes, this is the IRRITATING take you get from the ENTIRE Matrix group! You have some fairly simple AI complaints, and because it is the AI and all these people are 100% PBEM players, they dump it all in the refuse bin as "undoable". Yes, this AI is pretty close, and are not talking a Deep Blue effort here one would think.

However, I think what we are seeing is these guys have locked themselves into a pretty rigid, inflexible basic design here, to the point that what seems to professional programmers from outside that appear to be relatively minor, almost trivial changes, are for some reason, almost undoable!


I look forward to your wargame design and AI code if/when you ever produce it.

_____________________________


(in reply to ZOOMIE1980)
Post #: 43
RE: AI craziness - 7/13/2004 10:07:23 PM   
Mr.Frag


Posts: 13410
Joined: 12/18/2002
From: Purgatory
Status: offline
quote:

I look forward to your wargame design and AI code if/when you ever produce it.


As soon as he figures out his Tivoli problems

(in reply to Nikademus)
Post #: 44
RE: AI craziness - 7/13/2004 10:09:08 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: The Gnome

/quote]

I think he has a good clue actually... the AI is painfully, excrutiatingly close to being good. There are just a couple of death spirals to remove and it will be challenging through the entire game.

I just get frustrated when I get the sense that people think that last step is as impossible as solving the Grand Unification theory of physics.... the AI is 90% there.


I understand. Believe me i do...as i've mostly been an AI player with little experience (or desire at times) to do PBEM type play. My experiences inside the dev forum though just have made me very much more sensitive to how difficult this aspect is. That doesn't mean that i just put on a monk's robe and sing hymns and mantras to Gary and company but i do curb my expectations in regards to the AI.

That and i've somewhat reluctantly embraced PBEM play. That was my whole motivation behind exploring 2 and 3-day turns. I have no wish to spend 3 years playing someone 1 day/turn at a time.

So far it's been mostly rewarding. I can appreciate though the poster who commented about wanting to 'role play' a bit.

_____________________________


(in reply to The Gnome)
Post #: 45
RE: AI craziness - 7/13/2004 10:09:20 PM   
Grotius


Posts: 5798
Joined: 10/18/2002
From: The Imperial Palace.
Status: offline
I've seen the exact same problem around Malaya -- the AI blindly sending transports to supply the first base it captured on the Indian Ocean. But in my game, the AI woke up after two turns of pounding and sent KB two hexes away from Singapore -- AND started flying Long Range CAP over all the transports. I lost 35 planes in my most recent turn as a result, and I scored only one hit on an AK to show for it. I can't tell whether the CAP is coming from KB or Indochina, but most of the transports I attacked in my most recent turn were protected by Zeroes.

So it *does* seem to me that the AI knows how to use CAP and LRCAP if it assesses a threat properly. Or maybe it was scripted to send KB to Malaya anyway? Either way, the fact that it's now protecting those convoys suggests that it's not an insurmountable programming task.

(in reply to ZOOMIE1980)
Post #: 46
RE: AI craziness - 7/13/2004 10:12:31 PM   
Tenzan


Posts: 101
Joined: 6/5/2004
Status: offline
quote:

I suspect these kind of AI problems are probably common to at least the first releases of nearly all games. It perhaps virtually impossible to programme into an AI the flexibility to changing situations of a human player. It will tend to stick rigidly to its pre-programmed "plan" (not that some human commanders haven't been guilty of the same...)


Yep-The AI in any current wargame will at best be a 'sparring partner', especially if it followes the 'rules'..Some games get by with a smarter looking AI by having it cheat, essentially. All I ask for is an AI that helps me keep 'willing suspension of disbelief' that it's a credible opponent-In other words, no -glaring- screw-ups...So far WITP has been good looking..hardly any complaints..I saw alot more kneeslappers (and saw them alot more frequently) in UV.

(in reply to Nikademus)
Post #: 47
RE: AI craziness - 7/13/2004 10:24:45 PM   
Mr.Frag


Posts: 13410
Joined: 12/18/2002
From: Purgatory
Status: offline
If you want the AI to be allowed to Cheat (which is how 95% of all games handle this), kick up the skill level 2 notches from Historical.

Anyone complaining about the AI on Historical level is pretty much wasting forum space. You want to actively play a game against the AI, Hard is the *minimum* level to play at. Below that is for training.

(in reply to Tenzan)
Post #: 48
RE: AI craziness - 7/13/2004 10:25:50 PM   
Moquia


Posts: 174
Joined: 7/12/2004
Status: offline
I don't know how the AI is programmed, so I don't know if it is possible. But if we could acces the values the AI uses, maybe through the database editor, we players could tweak the numbers and maybe come up with a better compromise between an agressive or dormant AI.

_____________________________


(in reply to Nikademus)
Post #: 49
RE: AI craziness - 7/13/2004 10:27:35 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: Mr.Frag

quote:

I look forward to your wargame design and AI code if/when you ever produce it.


As soon as he figures out his Tivoli problems


Ugh!!! Don't get me started!!!

(in reply to Mr.Frag)
Post #: 50
RE: AI craziness - 7/13/2004 10:28:32 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: ZOOMIE1980
I remember dealing with this man from back in the 1980's.Brilliant designer, sixth sense when it comes to balanced game formulas. Just KNOWS what works and what doesn't. But he's a walking brick as a programmer. Apparently he and his minions are still tightly locked into their completely obsolete hard-coded fixed array based programming models.


I suggest submitting your resume to Matrix. I'm sure they'd be happy to take someone on who can educate them all on the proper way to do programming

_____________________________


(in reply to ZOOMIE1980)
Post #: 51
RE: AI craziness - 7/13/2004 10:28:53 PM   
The Gnome


Posts: 1233
Joined: 5/17/2002
From: Philadelphia, PA
Status: offline
quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: The Gnome

/quote]

I think he has a good clue actually... the AI is painfully, excrutiatingly close to being good. There are just a couple of death spirals to remove and it will be challenging through the entire game.

I just get frustrated when I get the sense that people think that last step is as impossible as solving the Grand Unification theory of physics.... the AI is 90% there.


I understand. Believe me i do...as i've mostly been an AI player with little experience (or desire at times) to do PBEM type play. My experiences inside the dev forum though just have made me very much more sensitive to how difficult this aspect is. That doesn't mean that i just put on a monk's robe and sing hymns and mantras to Gary and company but i do curb my expectations in regards to the AI.

That and i've somewhat reluctantly embraced PBEM play. That was my whole motivation behind exploring 2 and 3-day turns. I have no wish to spend 3 years playing someone 1 day/turn at a time.

So far it's been mostly rewarding. I can appreciate though the poster who commented about wanting to 'role play' a bit.


I don't expect the second coming of Alexander for the AI, trust me. And that's what's the most frustrating part about the AI. It gives you an entertainingly challenging job then goes ape and self-destructs all of a sudden.

I think we all need to take a step back and realise that all anyone wants are those issues resolved. All the "well AI is really hard to do so deal with it" and "the AI is worthless" responses are things start to devolve into non-constructive flaming.

The problems in the AI *are* fixable and I don't think anyone wants a major overhaul.

(in reply to Nikademus)
Post #: 52
RE: AI craziness - 7/13/2004 10:32:19 PM   
Grotius


Posts: 5798
Joined: 10/18/2002
From: The Imperial Palace.
Status: offline
I'll say it again: in my game (on the wimpy "historical" setting), the AI seems to have recognized that its suicide transports needed air cover, and for three turns running now it's provided long range CAP. I've lost 55 planes in the past two days.

I'm not saying the suicide-transport thing isn't a problem. I hope the devs will address it. But the AI's response in my game is quite encouraging to me.

< Message edited by Grotius -- 7/13/2004 8:32:35 PM >

(in reply to The Gnome)
Post #: 53
RE: AI craziness - 7/13/2004 10:34:33 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: The Gnome
I don't expect the second coming of Alexander for the AI, trust me. And that's what's the most frustrating part about the AI. It gives you an entertainingly challenging job then goes ape and self-destructs all of a sudden.

I think we all need to take a step back and realise that all anyone wants are those issues resolved. All the "well AI is really hard to do so deal with it" and "the AI is worthless" responses are things start to devolve into non-constructive flaming.

The problems in the AI *are* fixable and I don't think anyone wants a major overhaul.


Possibly, i hope so. To use the current topic of choice, i know that Gary verified that the routine regarding Airbal was "working" yet the results being spit out were not what we all expected. There's just so much code and so many variables....well as i said, i can but sympathise.

I dont have great hopes that the AI will substantially improve but if small issues can be eliminated then people will be happier as a whole (well 'most' people...) The AI air attack routines were recently tweaked to make the AI make better choices vs just blindly attack the same target for heavy losses. I think the ship accumulation in a forward port is addressible (yeah....easy for ME to say) if we can get some good saves on it.

As for the rest......well.....

_____________________________


(in reply to The Gnome)
Post #: 54
RE: AI craziness - 7/13/2004 10:39:27 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: Moquia

I don't know how the AI is programmed, so I don't know if it is possible. But if we could acces the values the AI uses, maybe through the database editor, we players could tweak the numbers and maybe come up with a better compromise between an agressive or dormant AI.


Well this is the core of the issue, finally exposing itself. We have, what I think can be accurately labeled as the "AI Death Spiral" bug (thank you Mr Gnome..). It was a glaring problem in UV. Frag first exposed this bug in his Mandalay test. Regardless of how impossible the AI tried to take Mandalay with Frags extreme buildup, it could not. It just kept impaling itself. His answer was to turn up the "aggressiveness". Well now we have that aggressiveness making the AI send unescorted AK TF's to their slaughter over and over again. Never figuring it out.

After much back and forth, it appears the two are tightly locked in a some sort of catch 22 design. The ONLY apparent parameter that can deal with the "death spiral bug" is the "agressiveness" parameter. Meaning, it is beyond the constraints of the AI design to really address the Death Spiral bug without addressing some basic apparent weaknesses in the underlying design...... A VERY nasty thing to have to do, I can understand.

But to players like me, the Death Spiral bug is a very MAJOR flaw. I simply got around it in my game by just giving myself a house rule to not take advantage of it.

To the poster saying they sent a Carrier TF to answer you, well I think that was nothing more than another script that would have sent that TF in that direction no matter what.... I sat there for over two weeks pounding the hell out of everything they sent my way with two CA's four CL's and four DD's. Their air eventually force a retreat, but no before I blew FOUR AK TF's right out of the water first....all headed to the same place, non were escorted.....i.e. the Death Spiral bug.

(in reply to Moquia)
Post #: 55
RE: AI craziness - 7/13/2004 10:39:51 PM   
Black Cat

 

Posts: 615
Joined: 7/4/2002
Status: offline
I`m far being a Grigsby or Matrix groupie ( well... actually I think Gary IS God ) but I do think some of you ( who I respect BTW ) are taking a very specific issue that seems to happen in those very busy Hexs around Singapore in Feb 1941, with LCU combat, air recon, air combat, ship to ship stuff, subs moving and attacking, Transport TF`s moving, supply and morale calculations and lots of other stuff all happening at the same time and really over blowing the AI faults/problem with it.

In my AI Game, the AI does very, very smart stuff so far at 1/5/42, actually reacts to your moves in Game and is light years ahead of PW and even UV.

(in reply to Mr.Frag)
Post #: 56
RE: AI craziness - 7/13/2004 10:41:17 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: ZOOMIE1980
I remember dealing with this man from back in the 1980's.Brilliant designer, sixth sense when it comes to balanced game formulas. Just KNOWS what works and what doesn't. But he's a walking brick as a programmer. Apparently he and his minions are still tightly locked into their completely obsolete hard-coded fixed array based programming models.


I suggest submitting your resume to Matrix. I'm sure they'd be happy to take someone on who can educate them all on the proper way to do programming



They can't afford me.

(in reply to Nikademus)
Post #: 57
RE: AI craziness - 7/13/2004 10:42:55 PM   
The Gnome


Posts: 1233
Joined: 5/17/2002
From: Philadelphia, PA
Status: offline
quote:

I'll say it again: in my game (on the wimpy "historical" setting), the AI seems to have recognized that its suicide transports needed air cover, and for three turns running now it's provided long range CAP. I've lost 55 planes in the past two days.

I'm not saying the suicide-transport thing isn't a problem. I hope the devs will address it. But the AI's response in my game is quite encouraging to me.

Good to hear Grotius! I have only had one issue revolving around transports but I didn't push it to see if it would completely devolve.

(in reply to Grotius)
Post #: 58
RE: AI craziness - 7/13/2004 10:44:02 PM   
ZOOMIE1980

 

Posts: 1284
Joined: 4/9/2004
Status: offline
quote:

ORIGINAL: The Gnome

quote:

ORIGINAL: Nikademus

quote:

ORIGINAL: The Gnome

/quote]

I think he has a good clue actually... the AI is painfully, excrutiatingly close to being good. There are just a couple of death spirals to remove and it will be challenging through the entire game.

I just get frustrated when I get the sense that people think that last step is as impossible as solving the Grand Unification theory of physics.... the AI is 90% there.


I understand. Believe me i do...as i've mostly been an AI player with little experience (or desire at times) to do PBEM type play. My experiences inside the dev forum though just have made me very much more sensitive to how difficult this aspect is. That doesn't mean that i just put on a monk's robe and sing hymns and mantras to Gary and company but i do curb my expectations in regards to the AI.

That and i've somewhat reluctantly embraced PBEM play. That was my whole motivation behind exploring 2 and 3-day turns. I have no wish to spend 3 years playing someone 1 day/turn at a time.

So far it's been mostly rewarding. I can appreciate though the poster who commented about wanting to 'role play' a bit.


I don't expect the second coming of Alexander for the AI, trust me. And that's what's the most frustrating part about the AI. It gives you an entertainingly challenging job then goes ape and self-destructs all of a sudden.

I think we all need to take a step back and realise that all anyone wants are those issues resolved. All the "well AI is really hard to do so deal with it" and "the AI is worthless" responses are things start to devolve into non-constructive flaming.

The problems in the AI *are* fixable and I don't think anyone wants a major overhaul.



Agreed, the AI is NOT worthless, far from it. But this is a MAJOR flaw in it. And it seems, it is the ONLY real flaw that stands between it becoming a truely new level in AI development. Unfortunately, it seems a flaw that is too difficult to address because it means altering some underlying design, which, I understand, is a very PAINFUL thing to do.

(in reply to The Gnome)
Post #: 59
RE: AI craziness - 7/13/2004 10:44:13 PM   
Nikademus


Posts: 25684
Joined: 5/27/2000
From: Alien spacecraft
Status: offline
quote:

ORIGINAL: ZOOMIE1980

They can't afford me.


LOL.....so you say.

_____________________________


(in reply to ZOOMIE1980)
Post #: 60
Page:   <<   < prev  1 [2] 3 4   next >   >>
All Forums >> [Current Games From Matrix.] >> [World War II] >> War In The Pacific - Struggle Against Japan 1941 - 1945 >> RE: AI craziness Page: <<   < prev  1 [2] 3 4   next >   >>
Jump to:





New Messages No New Messages
Hot Topic w/ New Messages Hot Topic w/o New Messages
Locked w/ New Messages Locked w/o New Messages
 Post New Thread
 Reply to Message
 Post New Poll
 Submit Vote
 Delete My Own Post
 Delete My Own Thread
 Rate Posts


Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI

0.766