Are quick battles vs the AI worthwhile?

[Paper Tiger]

The Quick Battle maps that ship with the game are crafted with AI plans that have to cover every possible eventuality, from force size to force composition. They have to be able to handle any type of force that the AI side purchases. Then there are around 200+ of them, usually crafted by one guy so it's unlikely that they're going to produce spectacular QBs every time. To get a good, and I do mean good QB experience you should customize a map for a particular type of battle and use lots of AI groups and have five plans. I have crafted a few for my own personal QB needs that are designed for Infantry v Infantry Probe/Attack/Assault battles only, (No MEs) and they all work just fine, just like regular scenarios. But they'd suck big time if I were to play them with combined arms groups etc.

[altipueri]

I agree with Streety.

More than ten years on and The AI is not significantly better than CMx1 and the QBs are miles worse. CMx1 had a pretty good random selector and terrain generator. Once you know that the CMx1 AI tends to do a rush at the end you make your own judgement on whether you really won.

So I'm playing more CMx1, but that's also because my great new PC is suffering from the invisible soldier bug and I'm forced to wait for some BF bug fix.

[Streety]

Hi again guys. And thanks for your support altipueri. Just to respond/pick-up on a few comments after my last post:

To Eltorrente - I wish I could write code! And I'm not against H2H on the computer but merely pointing out that H2H was possible with old-fashioned table-top wargames. Yes, the computer can also do all those calculations for you so you don't need your old rulebook, and you don't have to find space and lug all your miniatures around, you don't need an umpire or a third table if you want to include the fog of war, and so on. But you lose something too - the game no longer feels tactile, and whilst you can "mod" images, the enjoyment and skill (and most of the purpose) of military modelling is gone. But my point was that you could do all those things before computers and so the main advantage of the computer for wargaming (the main thing that pre-computer age wargames could not do) is provide an enemy AI.

To Womble, Erwin et al - Yes, a wargame enemy AI is a lot more complicated than a chess game, which is why I said there are a lot more variables and permutations. But my point is, as proved by people having quickly started listing them in this thread (dead ground, morale, etc) they can all be accounted for and every good wargamer (and rulebook) considers these points and their impact. So, given that computing power has come on leaps and bounds since the 1980s, we should nowadays be at the point of having a great enemy AI that can compute all those factors (and more) in the same fashion, only with the benefit of being a lot faster (and thus playable) of course. All it takes is the time to build all those variables and permutations into the AI and test them. And no, I'm not confusing tac-AI with strategic-AI (or AI-plan), but talking about both overall.

I guess another way of putting it is this: early computer wargames were mostly for the solo-gamer, and because the graphics weren't great and few people had internet, a huge proportion of the developers' efforts went into making the computer AI your opponent. But over the years, with improved graphics and internet, an increasing portion of the developers' efforts goes into making games look good, and making good multiplayer games, but at the cost of the proportion of effort spent on developing the enemy AI (and other replayability aspects) for the solo-gamer. So graphics have come on leaps and bounds but AI has not.

CM is a prime case in point: I had hoped that CM2 would be more like taking CM1 and putting most of all their new efforts put into improving some of the AI weaknesses, then some effort into improving random maps (to include rivers & bridges, more building types etc). And then maybe adding more of a campaign-mode if possible. And maybe even an option for large-scale battles (with platoons the smallest unit rather than squads). And then just a little time spent on making the graphics and animations less blocky. Instead, what do we get? A great new graphics engine on which most of the time has spent (and going down to the 1:1 scale - great for modern warfare, but a step back for CM1 WW2 battles), but also the abandonment of truly random maps, and a little AI tweaking as a relatively low priority (partly dealt with by removing the truly random maps). Yippe. It "looks" great guys.

[slysniper]

I agree with you. If anything, the AI in the game is actually weaker. Because now it requires a good AI plan and has been pointed out.

Without someone crafting a good plan, like in a scenario, the AI basically sucks.

So in a QB, in general, you are not going to get a good battle unless you have the skills to go into the game and craft the AI to work correctly.

For many, this is not what is wanted or do they have the time or skills for wanting to do it.

So in some ways CMX1 actually was the better product, in that the machine at least gave you some type of fight that was a little realistic. Now it does not function at all, unless you know how to correct the problems and make it work at times. (but I would not waste my time with it unless you are willing to learn to get in the editor and create good quality Battles.

But you are correct in pointing out how in general, computer games are more fluff and less substance. If you want substance, it is up to you to provide it. And they wonder why many complain and do not want to continue to buy the latest product.

[Michael Emrys]

This will have to be brief. Streety, you are making some good points and I wish I had more time to respond to them in greater depth. But one thing you need to keep in mind is that programming AI is really, really hard. At all levels of cybernetic application it is really, really hard, no matter how many Ph.Ds you hire to do it. The reason is that computers do not "think" the way that the human brain does, and trying to get them to generally runs afoul of some limitation or other very quickly and often.

Comparing chess to a game like CM is really not on. Chess is a game that plays to the peculiar strengths of the computer: quickly analyzing a finite set of moves and possible responses. And those moves are precisely and rigidly defined. So, hard for a human, easy for a computer.

CM is chaotic. The possible outcomes of any chosen act within the context of a vast number of simultaneous action multiply almost beyond imagining. Playing a game like CM requires seeing the forest rather than counting the trees. Easy for a human, hard for a computer, at least to this date.

AI design is still awaiting its Einstein, someone who has a totally original insight on how to approach the problem that will allow very complex problems to be simplified to the point where your average game designer has a fair shot at producing an AI that will always respond in a human-like way.

Michael

[mjkerner]

This will have to be brief. Streety, you are making some good points and I wish I had more time to respond to them in greater depth. But one thing you need to keep in mind is that programming AI is really, really hard. At all levels of cybernetic application it is really, really hard, no matter how many Ph.Ds you hire to do it. The reason is that computers do not "think" the way that the human brain does, and trying to get them to generally runs afoul of some limitation or other very quickly and often.

Comparing chess to a game like CM is really not on. Chess is a game that plays to the peculiar strengths of the computer: quickly analyzing a finite set of moves and possible responses. And those moves are precisely and rigidly defined. So, hard for a human, easy for a computer.

CM is chaotic. The possible outcomes of any chosen act within the context of a vast number of simultaneous action multiply almost beyond imagining. Playing a game like CM requires seeing the forest rather than counting the trees. Easy for a human, hard for a computer, at least to this date.

AI design is still awaiting its Einstein, someone who has a totally original insight on how to approach the problem that will allow very complex problems to be simplified to the point where your average game designer has a fair shot at producing an AI that will always respond in a human-like way.

Michael

Well put, Michael, and I agree 100%. But you know, a couple of "simple" triggers, along with the increased number of AI groups provided in version 2.0, would be a huge help to scenario designers. I realize the problems that can occur when developing triggers (discussed ad infinitum on a thread here last year), but certainly some basic ones could be achievable.

[A Canadian Cat]

To Eltorrente - I wish I could write code! And I'm not against H2H on the computer but merely pointing out that H2H was possible with old-fashioned table-top wargames. Yes, the computer can also do all those calculations for you so you don't need your old rulebook, and you don't have to find space and lug all your miniatures around, you don't need an umpire or a third table if you want to include the fog of war, and so on. But you lose something too - the game no longer feels tactile, and whilst you can "mod" images, the enjoyment and skill (and most of the purpose) of military modelling is gone. But my point was that you could do all those things before computers and so the main advantage of the computer for wargaming (the main thing that pre-computer age wargames could not do) is provide an enemy AI.

Interesting point of view. I was just talking with a friend last week about what the greatest advantage to CM was. I would argue that it is the support for fog of war. It is true that you *can* get that feeling with table top or counter based games but only with lots of extra effort and resources. I never had the pleasure to play a table top game with redundant tables and an umpire. I had a hard enough time just getting an opponent to play against. So my previous experience was that you could always tell what the other player was up to at least to some extent.

The beauty of this game is you have no idea what the other guys is doing or what forces you are facing. It is an amazing feeling. And I have grown to believe that it is the most important factor that makes this game so good. Don't get me wrong I love the 1:1 modeling, excellent graphics, WEGO and the tac AI. They all contribute to my enjoyment of the game. But it is the fog of war that sets this apart from any previous war gaming experience I have played.

CM is a prime case in point: I had hoped that CM2 would be more like taking CM1 and putting most of all their new efforts put into improving some of the AI weaknesses, then some effort into improving random maps (to include rivers & bridges, more building types etc). And then maybe adding more of a campaign-mode if possible. And maybe even an option for large-scale battles (with platoons the smallest unit rather than squads). And then just a little time spent on making the graphics and animations less blocky.

Again I am not sure I agree. I played CMBB and CMAK while CMBN was in the final stages of development and my impression is that the prebuilt QB maps are better than the generated ones (I have seen some pretty odd environments in the old game). I also think the AI is better - or at least can be better. Because it is not just a generic AI but rather has human input that is specific to the map. The AI can be more challenging when it is programmed well.

[A Canadian Cat]

I agree with you. If anything, the AI in the game is actually weaker. Because now it requires a good AI plan and has been pointed out.

Without someone crafting a good plan, like in a scenario, the AI basically sucks.

So in a QB, in general, you are not going to get a good battle unless you have the skills to go into the game and craft the AI to work correctly.

<snip>

My previous comments about a well crafted AI in CM2x being better than in CM1x was based on playing scenarios where the AI can be crafted to the units in the order of battle.

I have played very few QBs against the AI - I have played lots against humans.

Just to clarify.

[A Canadian Cat]

AI design is still awaiting its Einstein, someone who has a totally original insight on how to approach the problem that will allow very complex problems to be simplified to the point where your average game designer has a fair shot at producing an AI that will always respond in a human-like way.

Well put. I only have a small amount of experience working on expert systems which are a very simple form of AI and the effort involved is very large. Modern AI experts often don't even consider expert systems to be AI now a days:)

If such a break through is made it will be a very interesting 10 years that follow that.

[Streety]

This will have to be brief. Streety, you are making some good points and I wish I had more time to respond to them in greater depth. But one thing you need to keep in mind is that programming AI is really, really hard...... Comparing chess to a game like CM is really not on.....

Michael

Argh! This will have to be brief too Michael. For the THIRD TIME, I am not equating a wargame AI TODAY with a chess AI. And I am fully aware (and previously stated each time) that a battlefield AI is a far more complex creature. But what I AM saying is, if we could create challenging chess computers 30 YEARS AGO, then surely today, with all the advances in tech and programming in 30 years, we can surely cope with creating a more complex AI to deal with a CM battle. And this was PROVED by the AI in CM1, which despite some flaws still presented a challenging opponent for the solo-gamer, and its a great pity that most of the effort into CM2 was put into graphics - when putting it into further improving the weakest areas of CM1's AI, even if (after much effort) it was only a modest gain - would (along with not getting rid of truly random maps) have been of much better benefit to solo-gamers and replayability than CM2 provided.

[womble]

Argh! This will have to be brief too Michael. For the THIRD TIME, I am not equating a wargame AI TODAY with a chess AI TODAY. And I am fully aware (and previously stated each time) that a battlefield AI is a far more complex creature. But what I AM saying is, if we could create challenging chess computers 30 YEARS AGO, then surely today, with all the advances in tech and programming in 30 years, we can surely cope with creating a more complex AI to deal with a CM battle. And this was PROVED by the AI in CM1, which despite some flaws was overall not too bad for the solo gamer, and its a great pity that most of the effort into CM2 was put into graphics - when putting it into further improving the weakest areas of CM1's AI, even if (after much effort) it was only a modest gain. That (along with not getting rid of truly random maps) would have been of much better benefit to solo-gamers and replayability than CM2 provided.

You're neglecting the changes between the two games in the demands on the AI. Not only has AI programming improved since the days of CMBO, the challenges set are more difficult. The terrain system alone is probably an order of magnitude (or two if there's a square law involved) more difficult for a machine to analyse than the CMx1 system. Hell, it's not even practical for the cursor to show what terrain you're mousing over. If that sort of data can't be extracted to present to a human in a coherent way, what chance of turning it into a hard (or even fuzzy) value for a computational engine to operate on? This factor alone is the reason there can't be random maps, and part of the beauty and draw of the game: its level of detail is difficult for people to get to grips with, let alone an AI.

I don't think a CMAK with improved AI would have been quite the thing needed to keep Steve and Charles in fine scotch and inspiration to develop this franchise as they are.

However, having said that, the modular nature of the new code might allow at some stage when all the fire, flares and funnies have been plugged in (or even at an earlier stage) for AI to be radically overhauled. I'm optimistic that the new triggers we've been promised will make writing AI plans more rewarding and the plans themselves more challenging. I'd also like to think that at some stage, QB maps could be allowed to have specified OpFors for given levels and modes of engagement. Ideally, a separate OoB file would allow a variety of different forces to be added to any given map. I don't imagine I'm the only one who enjoys putting together OOBs, so if the OOB could be separated out, you'd see plenty of 'em submitted to the Repository for combination with maps.

[Streety]

No, I'm not neglecting all the changes - many here are mistaking improvements in tac-AI to work with the more complex terrain and in AI-plans for editors, rather than what I'm talking about which is improvements to the overall benefit of the solo-gamer in having an independent computer opponent which can generate truly random maps to fight you over them forever more without needing any further input from you. Which is what CM1 gave us and what I hoped CM2 would still give us - only more of and better.

And no I'm not talking about instead "simply" having CMAK with an improved AI (though for the solo-gamer that would have been better than what CM2 provides). But rather I'm saying that the balance of priorities for CM2 would have been better served by taking CM1 and putting AI improvements first, randon map improvements second, campaigns third, and graphics/animations improvements fourth. Instead we got a massive graphics engine overhaul, and although there was some AI work, much of it was obviated by doing away with truly random maps and a fully independent AI mind. So for the solo-gamer, despite more complex terrain-matching and tac-AI modelling etc, the downsides more than outweigh any benefits with the new system. But I hope you are right that one day the CM2 AI will be radically overhauled. In the meantime, I'm stuck with CM1.

[womble]

No, I'm not neglecting all the changes...

This sure looked like you were:

...what I AM saying is, if we could create challenging chess computers 30 YEARS AGO, then surely today, with all the advances in tech and programming in 30 years, we can surely cope with creating a more complex AI to deal with a CM battle...

...many here are mistaking improvements in tac-AI to work with the more complex terrain and in AI-plans for editors, rather than what I'm talking about which is improvements to the overall benefit of the solo-gamer in having an independent computer opponent which can generate truly random maps to fight you over them forever more without needing any further input from you. Which is what CM1 gave us and what I hoped CM2 would still give us - only more of and better.

The problem that you're not seeing is that for the latter improvement, everything needs to get better. Which means the AI becomes impracticable to code without input. Which requires AI plans for the editor. These things can't be isolated. You disagree with the balance that BFC have drawn, fine. Just don't go insisting that just because computers are better AI has to be better too. The rules of Chess haven't changed recently. The rules of CM changed for a much more complex set on the advent of CM2.

[slysniper]

Streety, you are not stuck with playing CMX1 if you would just dedicate yourself to learning how to work the editor and create yourself some battles. Is it easy , no, is it fast to do , no.

But once you create one that works, you will see that The AI is actually a little better in that with human aid in giving it commands, Like holding a trench line or fox holes (which in CMX1 the AI troops get out of all the time if there is a flag anywhere on the map) That you can have more challenging battles.

I do feel your pain in that the game cannot provide any real challenge on its own at the moment. But if you understand why they removed that aspect of the AI and basically has AI battle plans designed by humans, it is presently the way they found to allow the AI to be a better challenge.

As for pointing out how they have not invested in making a AI that also creates a decent battle plan and can give you some challenge. Many have given their opinions as to why it is not being done or the challenges of doing it.

I am afraid that in the future you will be seeing more tools for a human to program what they want the AI to do, but I doubt you will see AI programming its own plans anymore any time soon.

[WynnterGreen]

Each to his own I suppose, but I can't help thinking that you're not even playing the game if you're not playing verse humans.

AI that can compete with a human with even rudimentary strategic aptitude in a free form environment like Combat Mission is a LONG way off.

The nuances of human behaviour in planing, taking risks, being patient, observing patterns in an opponents tendencies, adjusting on the fly or even esoteric notions, like intimidation are well beyond the scope and resources of Battlefront.

That's not a disparagement of Battlefront, that's true of AI specialists and research groups with resources that dwarf those available to this game.

In my opinion playing vs the AI should constitute a few scraps to familiarize yourself the 'mechanics' of the game in preparation to play humans.

You can stack the teams in favour of an AI opponent, add volume or wind up the difficulty, but everything that makes playing verse a human complex and rewarding is missing.

It's just 'different' playing with a person.

Matching wits with a human being is a primal thing, even via the conduit of a PC.

[altipueri]

Not playing a human being is the primal thing for me. I can't stand humans who play wargames. I gave up playing wargames forty years ago because all I saw was old men arguing about what was or was not fair or realistic. Of course to a teenager the old men were probably all of thirty.

So PC wars gaming when I discovered it in the time of Sid Meier's Gettysburg was great. Good enough AI for an hour or two of history games and no bitching because my favourite tank hit a mine. Reload, replay have fun.

[Erwin]

"I gave up playing wargames forty years ago because all I saw was old men arguing about what was or was not fair or realistic."

Not at all like what it's like now on these forums natch hehe.

[slysniper]

Not playing a human being is the primal thing for me. I can't stand humans who play wargames. I gave up playing wargames forty years ago because all I saw was old men arguing about what was or was not fair or realistic. Of course to a teenager the old men were probably all of thirty.

So PC wars gaming when I discovered it in the time of Sid Meier's Gettysburg was great. Good enough AI for an hour or two of history games and no bitching because my favourite tank hit a mine. Reload, replay have fun.

Well, very true, but playing someone now on a computer game, takes much of that away. There is no rules to fight over, (except some old men want to make them here). There is no should or should not happens - you get what the game gives you- except some old men will whine on this forum how game results are incorrect, but they are doing it here, not to who they played. So sounds like to me your ready to start playing Humans again, since the game has removed most of your biggest issue:)

[altipueri]

Actually of course it has been heaven for wargames the last fifteen years or so.

Matrix, Paradox, AGEOD, Battlefront. Almost too much to choose from.

I'm keeping a couple of old XP machines just so I don't lose the ability to play some of these games.

[Streety]

Thanks for feeling my pain Slysniper. Yes, the problem with manually manipulating the AI is that it takes away the fun, and much of the point, of having a computer opponent for the solo-gamer.

And to those in this thread who are essentially saying "well, get yourself a human opponent if you don't like the AI", I say that's fine if that's what you want, but I don't want, and I'm talking here about the solo-game and afterall, THIS THREAD IS ABOUT THE SOLO GAME!

And no Womble, I'm not neglecting those other CM2 advances, and you are (as always on this forum) taking me out of context and yourself neglecting the overall thrust of my argument, which is about the impact for the solo-gamer. The thrust of what I'm saying is, that for the solo-gamer playing WW2 battles and coming to CM2 from CM1 and hoping for an improved solo-game, those CM2 advances are more than off-set by a combination of the drawbacks in needing to manipulate the AI and most importantly losing truly random maps.

Furthermore, the smaller 1:1 infantry scale may work well for modern combat or WW2 skirmishes, but many CM1 WW2 gamers would have preferred keeping the same scale and with an added option of going up in scale for very large battles. There are other naff things too about CM2 - like those floating icons when keeping bases would have been better (except making the bases muted, semi-transparent). Yes, I know, once they commited to 1:1 visual representation they'd have to switch to icons, but I was hoping that CM2 would merely calculate on the 1:1 scale in the way that CM1 did not, and not waste the graphics capacity of showing lots of individual guys running through eachother when that graphics juice could have instead gone towards making bigger battles more manageable on our machines.

And then there are the camera controls and lack of better mouse integration, and the detached horizon/sky, which are not much improved on CM1. But whereas that was OK with the more table-top wargame feel of CM1, it's much less OK with the more simulator feel of CM2. And when added to things like icons instead of bases, it overall makes CM2 less tactile for the wargame general than CM1.

So for me, as a solo-gamer playing big WW2 battles, overall, CM2 has so far proved to be a backward step.

[altipueri]

D'accord.

Again.

[Stagler]

What would help, is an an easier match making service or game lobby for online play so you could more easily find a human opponent. Instead of a bugged direct connect system where you often cant get past the latest generation of firewalls or routers.

[Erwin]

What SS and Streety are alluding to is that some of us prefer playing vs the AI.

Ironically, it's been a very long time since I have played CM1 vs the AI and I enjoy playing huge CM1 battles vs a human because it's technically and practically quite easy.

However, due to the increased detail and "finickyness" of CM2 I find playing the AI considerably more challenging (assuming a primary goal is keeping casualties low). It is also a question of time. Plotting a CM2 turn takes a LOT longer due to aforesaid detail and finickyness (especially if you play WEGO).

Finally, it's great to play when I want to play, and not have to wait for an oppo sending me turns.

[sburke]

Thanks for feeling my pain Slysniper. Yes, the problem with manually manipulating the AI is that it takes away the fun, and much of the point, of having a computer opponent for the solo-gamer.

And to those in this thread who are essentially saying "well, get yourself a human opponent if you don't like the AI", I say that's fine if that's what you want, but I don't want, and I'm talking here about the solo-game and afterall, THIS THREAD IS ABOUT THE SOLO GAME!

So for me, as a solo-gamer playing big WW2 battles, overall, CM2 has so far proved to be a backward step.

rotflmao sometimes you do have to state the obvious

I was actually a solo gamer on CMx1 and initially as well on CMx2 (through all of CMSF and initially on CMBN) and I still definitely prefer CMx2 over CMx1 by a large margin. I have to qualify that in that I never cared too much for QBs as I didn't like the maps in CMx1 and I also was not thrilled about the AI. I have seen a number of statements lately about stuff in general that folks preferred from CMx1 and it leaves me scratching my head. I deleted all my CMx1 games a while back and have no interest in them anymore. It's all subjective preference, but I honestly don't get what was so great about CMx1 even after listening to what everyone has to say on it at least in comparison to CMx2. I certainly loved the game at the time and still remember it fondly, but not enough to reload it.

[Erwin]

It's all to do with size... as always.

In this case it's that CM1 was almost operational in scale with a Regiment plus on each side on 4Kx8K maps.

I expect that CM2 will get there eventually. However, the complexity of CM2 mitigates it being as much fun and playable as CM1 is at that scale.