Matrix Games Forums

Berkut,

It wasn't really aimed at you my friend....some people around feel the AI should be scrapped altogether...I have a problem with that...I get the AI isn't as good as a human but I'm really not bothered by that...I've been beaten by the AI enough times in games in the past and beaten even more...the thing is was it FUN...and I have to say games with apparent poor AI like the CLose Combat series I had loads of fun playing so thats whats important...I have a great time palying CMSF and at the moment BftB both against the AI both games I have a great time playing...some of the best game playing experiences I've had was against an AI in CC and Combat MissionBB and SF.

So I know its not going to be as good as a human opponent but as long as I enjoy it I really don't care...when I do win it's because I'm such a great tactician not becasue the AI was dumb[;)][:D]

ORIGINAL: wodin


So I know its not going to be as good as a human opponent but as long as I enjoy it I really don't care...when I do win it's because I'm such a great tactician not becasue the AI was dumb[;)][:D]

EXACTLY!!!!!!

Well, I would not say anything like that the AI ought to be scrapped - that would be fine with me, but I realize that would be a foolish business decision.

I am just saying that people should not get their hopes up that the AI is going to be fundamentally any better than AIs in complex wargames have always been - not very good. And the more complex the game is (both mechanically and in the strategic complexity of the operational situation), the less competent the AI is going to be.

This looks like a rather highly complex game mechanically about a highly complex situation. I have no great hopes for the AI.

ORIGINAL: Berkut

Well, I would not say anything like that the AI ought to be scrapped - that would be fine with me, but I realize that would be a foolish business decision.

I am just saying that people should not get their hopes up that the AI is going to be fundamentally any better than AIs in complex wargames have always been - not very good. And the more complex the game is (both mechanically and in the strategic complexity of the operational situation), the less competent the AI is going to be.

This looks like a rather highly complex game mechanically about a highly complex situation. I have no great hopes for the AI.

I don't think anyone is expecting a DEEP BLUE AI nor do I think anyone is expecting anything near human competence either. But after reading the testers' comments about the AI I think Gary is well on the way to achieving something remarkable in terms of the AI.

What I would like to know is what Gary believes to be acceptable AI performance.

Well in my AAR, the AI is filling gaps, springing reserves, and doing counters now. Were not done yet and it is a headache already. We have testers testing every game we have, and thank god for Pavel (Helpless), he has picked up so much of the bug work to allow Gary to do alot of AI work...This is one hell of a test and program team, and everything is being looked at...Theres alot of smart guys here, and I'm not close to being the brightest...

You are one of the best testers of the team, unlike me... I'm only a witch who must improve a lot to reach the level of you, Jose and Andy.

We believe that the majority of players will only play against the AI, so as with all of Gary's games we will work hard to make an AI that provides an enjoyable game. So far so good.

The AI (as with all AI's) will never be as good as a good human player. However, if you're willing to give it various advantages, it can give you a good challenge and a fun game.

We have had 5 programmers working on this game at one time or another. For the past 6 months Gary has spent at least 1/2 his time on the AI. Thanks to the help of Pavel, Karl and Jim, Gary has been able to spend more time on the AI than he otherwise would have, and we have more features than we otherwise would have had. The AI is still not the critical path item in development. We expect Gary will continue to work on it and improve it up until release. Gary'and I have been doing this together for 28 years, and it's an art not a science when balancing trying to get a good AI done and trying to get a game published in a reasonable time.

Hi!
The AI is in its development and of course it should be. So is the AI for chess.Not so many years ago it was said that
the AI for a chess program never could beat a human master. We know now about that. Of course thats an another thing with more limited rules and space.
Its also a bit easier to compare by using the chess rating system- The ELO-points.
But what if the development had stopped?

The AI must go on being better and better. We all benefit from that.
So it had been earlier. Remember steps like the war games Subhunter( u-boat) and Firebrigade (arond Kiew-43) in the 80's.
( to mention the start)

This game ,War in the East, is a part of the ongoing work on the AI and a big effort is made.
Another thing is that it has a long way to go to beat an experienced human,
who could either play "too" sound or using errors of the AI.
An unique thing with the wargaming AI is that its coupled to the map,
deployment,time elements and the historical flow. Its not a technical thing alone.
That interference makes possibilities.
Even AI-AI play has its value but of course a human to human play exceeds in many aspects.
I guess that is hard to beat but we know that to play the AI often is the only possibility at hand.

GoranW

So it's probably safe to say that however hard the creators work on the AI it will not be a match for a very good human player. It is also safe to say that without a credible AI the game cannot succeed economically in the long run given the market demand from players not interested in PBEM.

What I find rather amusing in some respects is the high place some put PBEM. While I prefer the competition of playing vs human opponents there are major issues with PBEM which while not the same problems, often rise to the same level as the problems we associate with playing against AI. Very good (experienced) PBEM opponents will know the order of appearance and initial placement of their enemies forces, they know what units are fixed and when or how they are released, they know when and where reinforcements arrive. And in my experience, the better PBEM players use that information to their distinct advantage. In some respects that smacks of "gamey" because to know what one should not know and then use it to win is not entirely right.

I'm guessing some here will insist that they dont examine in some detail their opponents status before they play a scenario but I've played enough PBEM to know most good players know all they need to know to gain advantages they shouldnt have. This may not be a blatant attempt at cheating. Rather it may simply be the result of the amount of experience a player has with the game and scenarios. If both sides have it well then maybe it all balances out in the end. If one side is more experienced and knowledgable about the scenario played then the advantages to one side are in some ways the same as the advantages held by human opponents over AI.

The game needs both a "good" AI and a well implemented PBEM component. If that is achived the game will be a smashing success.

From what I've seen my hopes are high that's exactly what we will get.

[quote]ORIGINAL: The SNAFU

"So it's probably safe to say that however hard the creators work on the AI it will not be a match for a very good human player. It is also safe to say that without a credible AI the game cannot succeed economically in the long run given the market demand from players not interested in PBEM. "

That depends on what you mean by " be a match" and a "very good human player".
I think that the AI in this game is credible and gives a match, that is a challenging opposition, even to an experienced wargame player.

"What I find rather amusing in some respects is the high place some put PBEM. While I prefer the competition of playing vs human opponents there are major issues with PBEM which while not the same problems, often rise to the same level as the problems we associate with playing against AI. Very good (experienced) PBEM opponents will know the order of appearance and initial placement of their enemies forces, they know what units are fixed and when or how they are released, they know when and where reinforcements arrive. And in my experience, the better PBEM players use that information to their distinct advantage. In some respects that smacks of "gamey" because to know what one should not know and then use it to win is not entirely right. "

Its an interesting aspect of PBEM you mention. Although the problem is lessend by the fact that the deployment and time elements in this game
is very " historically true" and in that way depending on common facts open to everybody. ( Like special technics in chess ex rook -endings and so on.)
This is possible in a game which is specialized in a certain war/operation, however in this case gigantic in itself.
The game is also so complex that details like " time and deployment of a certain reinforcement" have less distinct importance.


/quote]
GoranW

>Very good (experienced) PBEM opponents will know the order of appearance and initial placement of their enemies forces, they know what units are >fixed and when or how they are released, they know when and where reinforcements arrive.
 
Correct. Which is why the developers should spend enough time to "solve" these problems by introducing variables and triggers to introduce credible uncertainty.
 
1. The initial placement does not really have to be fixed; the players can have latitude in this regard; i.e. so many units need to be in the Baltic, Western, Southern fronts etc.  but where is not scripted.
2. Fixed units can have release triggers.
3. Reinforcements can have percentages chances of arriving or not (early or late) and even tactical/strategic options as to where.
 
IMHO the above suggestions are very important issues.

ORIGINAL: Marquo

>Very good (experienced) PBEM opponents will know the order of appearance and initial placement of their enemies forces, they know what units are >fixed and when or how they are released, they know when and where reinforcements arrive.

Correct. Which is why the developers should spend enough time to "solve" these problems by introducing variables and triggers to introduce credible uncertainty.

1. The initial placement does not really have to be fixed; the players can have latitude in this regard; i.e. so many units need to be in the Baltic, Western, Southern fronts etc.  but where is not scripted.
2. Fixed units can have release triggers.
3. Reinforcements can have percentages chances of arriving or not (early or late) and even tactical/strategic options as to where.

IMHO the above suggestions are very important issues.

I agree. Should be an option in the AI Game too

ORIGINAL: critter

ORIGINAL: Marquo

>Very good (experienced) PBEM opponents will know the order of appearance and initial placement of their enemies forces, they know what units are >fixed and when or how they are released, they know when and where reinforcements arrive.

Correct. Which is why the developers should spend enough time to "solve" these problems by introducing variables and triggers to introduce credible uncertainty.

1. The initial placement does not really have to be fixed; the players can have latitude in this regard; i.e. so many units need to be in the Baltic, Western, Southern fronts etc.  but where is not scripted.
2. Fixed units can have release triggers.
3. Reinforcements can have percentages chances of arriving or not (early or late) and even tactical/strategic options as to where.

IMHO the above suggestions are very important issues.

I agree. Should be an option in the AI Game too

This gets my vote too. Devs should reconsider their design choices I think.

[/quote]

1. The initial placement does not really have to be fixed; the players can have latitude in this regard; i.e. so many units need to be in the Baltic, Western, Southern fronts etc. but where is not scripted.
2. Fixed units can have release triggers.
3. Reinforcements can have percentages chances of arriving or not (early or late) and even tactical/strategic options as to where.

Agree with 2 & 3 but not 1, notably for Russian placement. The inept, even suicidal, configuration of the Soviet forces conceived out of Stalin's eagerness not to provoke the Germans made the rapid advances of 1941 possible. The campaign would have progressed in an entirely different way given sensible Russian dispositions, for example by concentrating their armoured formations and establishing a defence in depth. The Soviet player starts off with a bad hand, so be it. German initial placements are another matter.

I think it would be nice to have the option to start the 41 campaign earlier to simulate a "what if" Germany did not invade Yugoslavia and Greece in 41.

ORIGINAL: zbig

I think it would be nice to have the option to start the 41 campaign earlier to simulate a "what if" Germany did not invade Yugoslavia and Greece in 41.

The Spring rains of 41 would have prevented a much earlier start for the tanks, but they could have crawled along in the mud on foot.

ORIGINAL: zbig

I think it would be nice to have the option to start the 41 campaign earlier to simulate a "what if" Germany did not invade Yugoslavia and Greece in 41.

Such a what-if gives the Germans an earlier start in Russia.

But you would have to include the threat of British troops driving north from Greece. Without that threat the what-if means nothing.

-

ORIGINAL: Joe 98

ORIGINAL: zbig

I think it would be nice to have the option to start the 41 campaign earlier to simulate a "what if" Germany did not invade Yugoslavia and Greece in 41.

Such a what-if gives the Germans an earlier start in Russia.

But you would have to include the threat of British troops driving north from Greece. Without that threat the what-if means nothing.

Not really much of a threat given the situation.

No German invasion of Yugoslavia and Greece also means no German invasion of Crete. Without that decimaiton it allows for the possible air-assault on Malta by the 7th airborne. If Malta is captured Rommel's supply situation improves enough to take Egypt AND the British would be kicked out of the MED. Which may have convinced Turkey to join the Axis providing another passage into Russia.

While I don't think this would allow Germany to beat Russia, it would have given Russia the Middle East after the war.

I agree, an early start would necessarily involve also changing or interfering with the ops in Yugoslavia, Greece and Crete. On the one hand, it would of course be unimaginably awesome if the game map of this engine would one day (after patch, release of modding tools for map making, or with a nice "Europe 1939-1945" addon) stretch from northwest africa all the way to Urals, and allow the player to play a whole campaign including actions leading to Barbarossa. But that appears unlikely to happen in one game with one engine by these present developers. The focus of the engine is purely Russia, unlike the WITP-AE team that decided with their counterpart to simulate the whole war in the entire theater.

Anyway, the idea of allowing some flexibility in force setups would be a nice feature, yes. Optional, besides a historical setup. Logically, allowing the Germans the flexibility of maybe one "placement and reorganization" turn preceding the start of the battle would be legitimate. Clearly, the surprise moment would suggest that the Germans had more or less freedom of movement before, while the Russian were sitting idle. However, adding a second option might be particularly tempting for PBEM, allowing the Russians similar freedom: Kind of a "What if Stalin had believed his intelligence info?" scenario.

ORIGINAL: Marquo

>Very good (experienced) PBEM opponents will know the order of appearance and initial placement of their enemies forces, they know what units are >fixed and when or how they are released, they know when and where reinforcements arrive.

Correct. Which is why the developers should spend enough time to "solve" these problems by introducing variables and triggers to introduce credible uncertainty.

1. The initial placement does not really have to be fixed; the players can have latitude in this regard; i.e. so many units need to be in the Baltic, Western, Southern fronts etc.  but where is not scripted.
2. Fixed units can have release triggers.
3. Reinforcements can have percentages chances of arriving or not (early or late) and even tactical/strategic options as to where.

IMHO the above suggestions are very important issues.

IMHO you are off topic and your points have nothing to do with the quality of the AI. If you want do discuss these issues please start a different thread.