The CPU opponent- is it limiting CMx2 possibilities?

[Cull]

Originally posted by Battlefront.com:

...The TacAI is the part of the game that needs to know in what context is breakdancing the correct choice...

Steve

Duh....always.

[Zalgiris 1410]

This has been interesting but I'm a still a bit of a ludite so generally when I talk about the AI I'm referring to the CPU opponent StratAI not the TacAI. (Too complicated for me, but great so far IME in CMx1 thanks for that.)

As someone who has never actually played against a human opponent (I don't count myself as one) my main interest is in the area of StratAI immprovement for CMx2 to increase the enjoyment of solo play.

Essentially I would like to have the CPU opponent able to use artillery and on board mortars properly, make co-ordinated attacks, know when and how to tow guns forward on the attack and to set up and fight on the defence effectively etc. All StratAI resonsibilities to be sure.

Also I would like to be able to choose or radomise between different StratAI opponent setting options to provide veriety and to increase re-playability of battles.

Senario designers should suggest which option they think should be chosen for solo play, they could even nominate a few options that might be unwarrented and be unavailliable if they are realy concerned.

This kind of arrangement could even be used in multi-player games if one of the players could no longer continue or there weren't enough players to begin with and thereby provide a specifiable CPU/StratAI opponent fill in, hopefully more adequitely according to the situation.

An improved CPU/StratAI could also be use by a player in either solo or human vs human as an optional subordernant commander of certain elements or to deal with unplotted or forgotten about units. This would also allow the player to concentraite on what's important to them or even provide them with different recommendations from which they could follow or interpret their own solution from.

IME some techniques to use for getting around some of the CPU/StratAI poor tactics have been to give it planes which it doesn't control instead of Artillery spottered assets since it fires them all at one central place though rockets fire too wide for it to totally foul up with. It does also seem to do much better with re-inforcement arriving or else as first turn surpressed Artillery spotters in designed senarios. Bearing in mind above give it light mortars or off board 81mm mortars instead of othem on board.

It doesn't employ Infantry guns properly either and doesn't move guns forward effectively when on the attack, even just to get them in range of targets. Appart from attatching them to vehicles and making the CPU opponent unable to adjust them the only over option is to exchange a few guns for spottered assets as above or for SPA which it tends to use without due care.

I am seeing posts including improved AI in the main poll thread and I have been reading them as I imply I would at the start of this post- as calls for a much better CPU/StratAI opponent for solo play. (Sorry to be winging, doesn't mean that I haven't been greatful for what was manged in CMx1.)

Edit: What I would really like as a player is to be albe to employ as a hotkey is an option where the StratAI provides order according to a nominated tactical precept which is also adjustiable though that can be set by me at a default if their are options during a game. This would help me to handle large scale battles that I tend to play and I think it would immprove playability especially for others to ensure that they can keep up while against human opponents on line. (I like to imagine it as depicting an orders group so having a reset button like with artillery plotting would be important for this too.)

Am I asking too much?

[ August 30, 2005, 08:49 PM: Message edited by: Zalgiris 1410 ]

[Battlefront.com]

The StratAI will hopefully be better in CMx2 than in CMx1. But there is only so much we can do with it. Unfortunately, programming AI is akin to filling a leaky bucket with water using a teaspoon. It seems that no matter how fast you go, and how smart you are about it, the level never really gets much higher. And while you are expending so much energy and concentration on filling the bucket, you aren't paying attention to the chicken and steaks on the BBQ. Then you have to remind yourself, people did not come to the party for a drink of water

Steve

[Michael Dorosh]

Originally posted by Battlefront.com:

The StratAI will hopefully be better in CMx2 than in CMx1. But there is only so much we can do with it. Unfortunately, programming AI is akin to filling a leaky bucket with water using a teaspoon. It seems that no matter how fast you go, and how smart you are about it, the level never really gets much higher. And while you are expending so much energy and concentration on filling the bucket, you aren't paying attention to the chicken and steaks on the BBQ. Then you have to remind yourself, people did not come to the party for a drink of water

Steve

I never thought you'd find an analogy worse than your sports car analogy....but you just did.

[WineCape]

Bingo! Excellent analogy. MD, the reason Steve has (fresh) wit is because he's not fatigued by being in front of his monitor for 24 hours.

[Salkin]

Steve rocks the socks !

//Salkin

[Battlefront.com]

I always like coming up with new ways of 'splain things to the masses in ways they will understand. Buckets (aka things that hold ice and beer) and BBQ are the tools of the common man

I had another analogy that involving being stuck in a car and having to go to the bathroom. But I figured people don't get stuck in cars very often (except for in the Great White North in winter), so I passed on it due to the lack of connection with the masses.

Steve

[Battlefront.com]

Taking MD's criticism to heart, I have made some improvements to my analogy:

The StratAI will hopefully be better in CMx2 than in CMx1. But there is only so much we can do with it. Unfortunately, programming AI is akin to filling a 16oz red pastic cup with beer from a keg that is just about to kick. No matter how much you want that cup of beer to be filled, after a while you realize that all you are getting for your effort is just a bunch of frothy head and very little actual beer. And while you are alternating cursing at and pleading with the inanimate object, you realize you aren't paying attention to the chicken and steaks on the BBQ. Then you have to remind yourself, people did not come to the party for a single cup of beer that is 60% head.

Better?

Steve

[Cameroon]

Originally posted by Battlefront.com:

I had another analogy that involving being stuck in a car and having to go to the bathroom. But I figured people don't get stuck in cars very often (except for in the Great White North in winter), so I passed on it due to the lack of connection with the masses.

Steve

Drive around Washington DC during rush hour. THOUSANDS of people stuck in their cars Too bad I'm one of them. I'm sure LA and others are similarly bad, but DC recently earned the distinction of 3rd worst traffic in the US. Go us......

Now, if you can figure out a way for me to play CM while I'm stuck in rush hour...

[Battlefront.com]

Laptop and a power adaptor

Steve

[Cameroon]

Originally posted by Battlefront.com:

Laptop and a power adaptor

Steve

I'm sorry, I should have qualified that with a "Without getting a ticket or arrested."

Hmm.... my aged iBook would work for CMBO.... heheh

[easy-v]

Correct me if I am wrong, but LT BULL seems to be asking, essentially

Why not code CMx2 to include features of the game that the AI is not/cannot going to use, that would only be used in H2H or PBEM play?

I don't think he is asking for a game without AI (that would be dumb, on the face of it as well as pointed out by BFC). He's just saying don't let the AI stop CMx2 from being the wargame of the century.

And I agree. I think it would be a tremendous boon to the CM communities.

[Battlefront.com]

easy-v... see above comments. The main stumbling blocks to features are TacAI and hardware limitations, not strategic level AI. So the suggestion is really not of any practical use.

As it is we already follow the "we're coding the best game system, not the best AI" philosophy, so there really isn't much to change anyway.

Steve

[Treeburst155]

I'm happy to learn the computer opponent is NOT a big stumbling block to game features. As for the TacAI, if it's as good as in CM1, then it will be good enough for me.

Treeburst155 out.

[Zalgiris 1410]

Originally posted by Battlefront.com:

The main stumbling blocks to features are TacAI and hardware limitations, not strategic level AI.

As it is we already follow the "we're coding the best game system, not the best AI" philosophy, so there really isn't much to change anyway.

The Strat AI is my main concern as a solo player, and I think it would be good for human vs human play if it could be improved to the point where it functions effectively enough that players would be comfortable to use it as a subordinate with advice or for controling units in non essential areas or even as a stand in for multi-player games. This could make plotting turns faster and it would make it more likely that players would both return turns and finish games, IMHO.

I've been reading a lot of the posts in the poll and improving the CPU opponent / StratAI is numbering highly amoung the top fives, so far BFC I haven't heard much about this topic in reply. Sorry to prod, but for me its any improvements in the CPU opponent that will determine mostly how much more I will enjoy CMx2. I really hope it can be use by players as I suggest and it would be great if it has differrent tactical philosophies or personalities to choose from.

BTW BFC could you explain more clearly what you mean exactly by saying that you are following the "we're coding the best game system, not the best AI" philosopy. Are you referring to the TacAI or the StratAI or just total AI in general? :confused:

[ September 01, 2005, 07:43 PM: Message edited by: Zalgiris 1410 ]

[Battlefront.com]

BTW BFC could you explain more clearly what you mean exactly by saying that you are following the "we're coding the best game system, not the best AI" philosopy. Are you referring to the TacAI or the StratAI or just total AI in general?

StratAI. A tactical game like this without good TacAI will be crap. No amount of mult-player options will salvage it.

Don't ge me wrong... we would LOVE to make a much better StratAI. I think we can make an improvement over CMx1, but don't expect a Human like challenge from CMx2's AI. We simply do not have the time nor the resources to get our PhD's in AI and do R&D on a static game system for a few years. We've got games to make

Steve

[Pvt. Ryan]

Originally posted by Battlefront.com:

We simply do not have the time nor the resources to get our PhD's in AI and do R&D on a static game system for a few years. We've got games to make

Steve

Aren't there some proteins you can sprinkle into Charles's jar that would take care of that?

[WineCape]

Aren't there some proteins you can sprinkle into Charles's jar that would take care of that?

Ahemm,

Whadda you think I'm trying to do the last 5+ years? ---> FEED THE JAR <----

You think I'm a lover of MD and his reasonable wit and JasonC's typing skills to return everyday to this forum? Pffft.

In fact, I can confirm that Charles is not a jar with just jelly. He has a voice. Crackling over a trans-atlantic phone line, but nevertheless, a human audible sound. So squash any rumours and Steve's innuendo that he communicates with Charles through a medium of the female gender.

Charl Theron

--------------------

“I am simultaneously embarrassed and proud to state for the record that only two bottles of your fine South African wines remain in my wine rack. The rest gave their lives honorably in the service of the construction of CMAK over the past several months.” -- Charles Moylan, 28 Nov03

[Lt Bull]

easy-v...yep, that pretty much sums it up.

Originally posted by Battlefront.com:

As it is we already follow the "we're coding the best game system, not the best AI" philosophy, so there really isn't much to change anyway.

This is an encouraging statement but a little unclear.

BF have not really indicated if they would consider the option to add features to the game that would only be used in H2H or PBEM play.

This is especially relevant in the area of defining scenario victory conditions.

As I have mentioned before, the more complex/varied/layered the victory objectives of a scenario, the more interesting and realistic the possibilities for human players, but the more "confused" and incompetent a CPU opponent will become/appear.

This aspect of the game DIRECTLY tests the CPU opponent/StratAI capability to co-ordinate all unit's under it's control to achieve the global scenario objectives. (see this thread for a range of alternate ways of defining/measuring victory conditons (ultimately to make it MORE interesting/realistic for the HUMAN player).

I would really like to know if BF are planning to revise/expand this area of the game (given its sensitvity to the CPU opponent)?

If not, what is it about the current system (ie. "flags" that are only of value to whoever controls it at the end of the last turn, etc) that is worth keeping? It really limits the possibilities for scenario designers/players. It's not as if "it can't be done any better".

Lt Bull

[Battlefront.com]

BF have not really indicated if they would consider the option to add features to the game that would only be used in H2H or PBEM play.

Actually, I thought I couldn't have made it any clearer... NO. The reason is there is very little time gained from this. The feature itself + TacAI, regardless of the StratAI involvement, is what takes all the resources and runs into the limitations. So cutting the StratAI out of the loop, as you suggest, does nothing for anybody.

That being said, if there is a good game feature that we want to put in, have the time for, and it is practical in every way BUT the StratAI, then we would see about putting it in and not having it available in single player mode. But so far, after 9 years of development, we haven't run into anything like that.

Steve

[Kellysheroes]

Ahhh I'm so glad CMX2 will have a competent AI, I'm another 95% to 98% cpu opponent player as well for all my games. It plays when I'm ready to play and it goes home when I'm ready for it to go home. PBEM is too slow and setting up a game online these days seems to take too long.

I've had people email me wanting to play either PBEM or online and then never show up or even start the PBEM game. So, I've pretty much given up on trying to play PBEM anymore or online.

But, mano mano if I want to play a game of Age of Empires or some silly RTS game, there's hordes of players for those type games. If only turn based games had such a population and fan base.

And Spartan by Slitherine still has the best AI I've ever played against with difficulties that challenge instead of easy as hell like RTW's VH/VH vanilla version.

So, hurray for good AI's with difficulties that will always provide a challenge for most of us. CM AI is still good for defending with proper settings for me, so, I'm having fun. It suks for attacking, but, I never liked playing defense anyways. heh

[mobear]

I occassionally lurk this forum and have rarely posted, but it seems whenever I do lurk, I see at least one thread in which the following familiar pattern emerges:

1) Gamer raises important legitimate question or issue.

2) Cheerleaders unable to read plain English mostly (or completely) misinterpret said question or issue.

3) BF representative "answers" cheerleader's "concerns" while mostly (or completely) ignoring original gamer's question or issue.

4) Original gamer has to repeatedly explain and detail his original question (along with giving many examples and what-if situations), even though he had already adequately explained it in his initial post, just to get a modicum of an answer from BF representative - i.e. "pulling teeth").

This is such a thread.

I find the entire issue of "Strategic AI" as opposed to "Tactical AI" to be somewhat disingenious on the part of BF. I've been playing computer wargames sine 1985, starting with Gary Grigsby's old titles (Carrier Force, War in the South Pacific, etc.), and this is the only game, the only forum and the only developer that has made this distinction.

Now, it may very well be that these two AI's are indeed two separate entities, completely [edit: or largely] independent of each other and coding and implementing them in the game might indeed be two separate issues from a programmer's point of view, but from a gamer's point of view it really doesn't matter what you call it, or whether you make a distiction between the two.

I find it hard to believe that "in 9 years of programming", no one at BF has ever come across an idea that was scrapped because the AI (either one) would not be able to adequately handle it. Specifically, for example, I find it hard to believe that BF could not have come up with a more elaborate victory point system for CM, other than the current "capture the flag" system. If indeed they didn't, then I think they do in fact need to start thinking a little outside of the box, specifially, instead of automatically just thinking "computer game", how about thinking "head-to-head human game" and then tacking on the AI in whatever way works best?

I don't mean to belittle the important aspect of having a good AI, but the point is that the overall game design should not be compromised just to be able to code an adequate AI - which is the point being made in this thread by Lt. Bull.

In other words - and I think this is a big part of what Lt. Bull has already eloquently written - design it primarily from the point of view of a head-to-head game first and only then add the AI to it, as adequately as possible.

The isssue raised by LT. Bull in this thread is one of the most relevant I've seen in any gaming forum in my 20 years of gaming and I've had a mind of raising it myself a few times over the years, but realized in every case that it would be akin to pissing into a hurricane. There have been numerous times in the past, with various games, when the developer was subsequently asked "Why didn't you include this in the game", or "Why didn't you design it this way instead of that", where the answer given was that "The AI would not be able to handle it", or some similar response.

As for the people who don't like head-to-head play, let me say this: TCP/IP indeed requires both players to be online at the same time and for this reason is not for everyone. I, for one, also don't particularly care for it.

But in my 20 years of gaming, including over 10 years of various PBEM'ing, I have yet to hear a legitimate reason for not liking PBEM'ing.

The reason most often given is that you can only play one turn at a time. Not a valid argument, simply because you can play more than one game at a time. In fact, most people who PBEM, do in fact play more than one game at a time (not all, but most). In other words, PBEM lets you play as many or as few games as you wish - and let me tell you, when you're playing another person you will find yourself taking considerable more time making your moves than otherwise. From personal experience, playing 3-5 games at a time seems perfect, as there are always at least two or three turns waiting for me when I get home from work.

So all this babble about PBEM being too slow, or not being able to play enough is just that - babble, mostly from people who have never tried it. And, may I add, you're also missing out on *most* of the fun in the game by not playing another person.

The purpose of this post is not to demean any previous poster or BF, it's just that this is something that I've come across here before and it's an issue that is important. BF wants to know what gamers want - well, here's what at least *some* of us want.

[ September 08, 2005, 05:08 PM: Message edited by: mobear ]

[Salkin]

mobear

That's the thing.

This is what some of us want. Steve already stated earlier that most people playing CM and most of the potential customers out there play the AI.

IMHO this has motivated them to focus on AI.

//Salkin

Clueless foreign chearleader

[dalem]

Originally posted by mobear:

So all this babble about PBEM being too slow, or not being able to play enough is just that - babble, mostly from people who have never tried it. And, may I add, you're also missing out on *most* of the fun in the game by not playing another person.

Pure subjective conjecture on your part. Some people DO find PBEM play too slow and/or too limiting. In your opinion it may be "babble", but to those with those opinions it is quite valid. And you have no idea whether said opinions are "mostly from people who have never tried it." No idea.

-dale

[Battlefront.com]

Woohoo... I haven't had an arrogant and pompous post like this to reply to in a long time.

mobear,

That might be your take on it, but here is my take on it:

1) Gamer raises important legitimate question or issue.

Gamer raises an issue. There might have a point, there might not be. It might be important to a few, it might not be important to anybody else. But it is raised so I respond. Guess that's my fault for responding... other game developers don't bother. Maybe that would be better?

2) Cheerleaders unable to read plain English mostly (or completely) misinterpret said question or issue.

Hehe... this is funny since you apparently can't read plain English either. Either that or the person complaining doesn't like the answer. See next point:

3) BF representative "answers" cheerleader's "concerns" while mostly (or completely) ignoring original gamer's question or issue.

Some people are so desperate to find fault with us they don't read things that are plainly written. Mobear, kindly go back to the first page and reread my very first reply. No wait... you are obviously having problems with this so I will quote:

"As it is most of the AI work is in the TacAI. That is something needed for single and multi-player. So there isn't all that much time to be saved anyway."

This was repeated in every answer I made, including the most recent one. Which stated:

"Actually, I thought I couldn't have made it any clearer... NO. The reason is there is very little time gained from this. The feature itself + TacAI, regardless of the StratAI involvement, is what takes all the resources and runs into the limitations. So cutting the StratAI out of the loop, as you suggest, does nothing for anybody."

So this begs the question... did you not actually read my posts? Me thinks you did but you didn't like the answer and instead decided to try flaming instead. Kinda like a 2 year old not liking the answer "NO" when he requests candy.. Not very productive.

4) Original gamer has to repeatedly explain and detail his original question (along with giving many examples and what-if situations), even though he had already adequately explained it in his initial post, just to get a modicum of an answer from BF representative - i.e. "pulling teeth").

Bullcrap. I posted an answer pretty much right away and it is the same answer that I have had to repeat because some people are either too thick to understand it or are simply not happy with it. Either way, that isn't my fault. I gave a DIRECT and to the point answer.

I find the entire issue of "Strategic AI" as opposed to "Tactical AI" to be somewhat disingenious on the part of BF.

Huh?

I've been playing computer wargames sine 1985, starting with Gary Grigsby's old titles (Carrier Force, War in the South Pacific, etc.), and this is the only game, the only forum and the only developer that has made this distinction.

Some people were sure the world was flat. I am sure the first time they heard it was round they got a big surprise too.

You are apparently not big on facts, but instead are pretty much big on your own detailed, inner knowledge of the game we made. (not). Go back to the earliest postings on this Forum and count how many thousands of times I have made the distinction between StratAI and TacAI. Heck, I think it is even in the manual. There are in fact two entirely different AIs. They are coded independently of each other for one very simple reason: The TacAI is used by all units all the time no matter if there is a Human or an AI controlling that side. It controls low level behavior. The other is the StratAI that is controls higher level behavior when the AI is in control of a side. This is a fact, and it is a fact you can only dispute by looking foolish.

Now, it may very well be that these two AI's are indeed two separate entities, completely [edit: or largely] independent of each other and coding and implementing them in the game might indeed be two separate issues from a programmer's point of view, but from a gamer's point of view it really doesn't matter what you call it, or whether you make a distiction between the two.

Well, if that isn't an arrogant statement... I don't know what is. OF COURSE IT MATTERS... because whatever you call it doesn't make a hill of beans difference. What goes on in the code matters. That's all. Plain and simple. TacAI is where the bulk of the AI programming time went into and it is required whether there is a super StratAI or none at all. It is the backbone of the game itself.

I find it hard to believe that "in 9 years of programming", no one at BF has ever come across an idea that was scrapped because the AI (either one) would not be able to adequately handle it.

Well, I can't think of one. But then again... I'm not as well informed as you so I'm probably wrong. However, I can think of several things that the Humans can do that the StratAI can not do, or can only do to the barest extent. What specifically can't the StratAI do that the Human can? Off the top of my head Embark/Disembark towed weapons is certainly one. And if I count things that the StratAI handles very poorly because we didn't have time to code up the perfect AI, yet didn't want to denny the feature to Human players, we could be here all night But ignore that because it defeats your pompous position that you know something that we don't.

Specifically, for example, I find it hard to believe that BF could not have come up with a more elaborate victory point system for CM, other than the current "capture the flag" system.

We came up with that system because it was the system we wanted. Sorry you didn't like it. Oh, and AI didn't have anything to do with it. In fact, we could have probably made a smarter AI if we had made a more elaborate victory condition system. So your argument that we didn't do this because we didn't want to do the AI programming for it is, just like everything else you've said, flat out wrong.

If indeed they didn't, then I think they do in fact need to start thinking a little outside of the box, specifially, instead of automatically just thinking "computer game",

Ah... so that's why we could only conceptualize the first true 3D wargame with a WeGo system and unparalleled realism and all of that. We weren't thinking outside of the box. Good God, think of what we could have achieved if we were as smart as you. The moon and the stars would be ours now! Oh the humility of it all...

I don't mean to belittle the important aspect of having a good AI, but the point is that the overall game design should not be compromised just to be able to code an adequate AI - which is the point being made in this thread by Lt. Bull.

But as I have said a dozen times, the premise doesn't stand up to scrutiny. Look, I don't know what you do for a living but it is my opinion that you suck at it. Now, like you I don't need to have any hard evidence. I have made my statement and I am convinced of my superior knowledge. Therefore, why don't you stop sucking at what you do? Hey... talking out of my ass is easy now that you've shown me how to do it. Thanks!

The isssue raised by LT. Bull in this thread is one of the most relevant I've seen in any gaming forum in my 20 years of gaming and I've had a mind of raising it myself a few times over the years, but realized in every case that it would be akin to pissing into a hurricane. There have been numerous times in the past, with various games, when the developer was subsequently asked "Why didn't you include this in the game", or "Why didn't you design it this way instead of that", where the answer given was that "The AI would not be able to handle it", or some similar response.

Other developers are entitled to answer for their own games. For us, we haven't once not included something simply because the StratAI would be too difficult to handle it. However, we have not coded lots of ideas because we didn't think programming the TacAI was practical. And since the TacAI is required for head to head Human play, that doesn't leave your position with much footing. We have also not programmed thousands of features because either the hardware can't handle it, they conflict with other game design elements, are simply too difficult to code for the gain, or are (plainly put) stupid ideas.

In this case the notion that games should be made with features that are the best, whether the AI can handle them or not, is not a bad suggestion. It also is nothing new. It's a very old concept and it is akin to board wargames. But in this case there really isn't much to be gained from purposefully excluding or including features based on the StratAI programming. We don't do that. At least we haven't so far. But at some point I am sure we'll come up with an idea that can't be created due to StratAI considerations. And then we'll have to make a decision about "should we include it even if the StratAI can't do it?". Don't know when or what feature will do this, however we will make a decision about it based on the overall vision for the game and our limited development time. That means, for those of you who can't understanding English very well, we'll look at it on a case by case basis.

Steve