The CPU opponent- is it limiting CMx2 possibilities?

[Lt Bull]

Hello,

Good to see lots of discussion about suggestions for new features/imporvements to include in CMx2.

I haven't seen any suggestions to improve the CPU opponent. In a way, I am not surprised as I think most people are more interested in seeing enhancements in the actual game features in favour of improvements in the CPU opponent. In other words, they want a game which gives the player more things to think about and control, more options and decisions to make and choices to take, more variety in the types and levels of challenges of a battle, ultimately more detail and complexity (I am not talking about "eye candy" stuff). A game that is richer in scope for the player of overall "command experience", than what is currently in CMx1.

eg. Currently in CMx1, the player only really makes decisions on what units to pick before a battle (if a QB), where to place them (if the scenario allows you), the orders to give those units during a battle and thats about it. All these decisions are made within the context of trying to achieve the scenario objectives. Regardless of what kind of battle the scenario designer is trying to create/simulate, these scenario objectives have been limited in definition and measured by the same set of simple/basic artitrary and abstract parameters for the last 6yrs (points based on capturing "flags", killing/capturing enemy units, and much less commonly, exiting units off the map).

This certainly raises an intersting issue however relating to the CPU opponent and what kind of game CMx2 can/will be. It gets back to a fundamental issue in computer game development. The more complex the "rules", "features" and "objectives" of a game, the more complex (and difficult/time consuming) the coding required to program a CPU opponent to competently play that game will be. In a game like CM in it's current form, the coded CPU opponent already has enough of a hard time interpreting what the game is about and how to control and orchestrate a virtual digital army to achieve the scenario objectives (or to prevent you from doing it).

Now think of the kind of suggested new features you want to see in CMx2, basically the expanded set of features that define the whole "command experince" of playing the game. Things that would make the player feel more involved and have more control over the outcome. eg. pre-battle recon, more detailed/specific/realistic scenario objectives, improved treatment of FOW, etc.

Now think that many of these potentially great ideas may be/will be/have been whacked on the head by BF just because the implications of including them in the game would be too complex/difficult/time consuming when considering the coding/re-coding requied for the CPU opponent to understand them.

I am sure that most of you would agree that the most challenging, rewarding and engaging gaming experience in CM is when you play human vs human. I know BF would never ever release CM without a CPU opponent coded. However, for those of us who are only really interested in playing other humans, it is worth considering what features are being omitted from CMx2 because of the implications they would havev on the coding of the CPU opponent . I don't know what proportion of time/resources BF expect to spend just on coding/re-coding a CPU opponent during the whole CMx2 development process, but I would bet it be a very large proportion, perhaps even 30-60%.

Im just wondering if BF would ever consider expanding the possibilities of the game features, free of the associated burden of having to also code the CPU opponent to competently follow/understand them, just so that the aspect of human vs human gameplay can be specifically taken beyond any of the current limitations. I would think that if a game feature was to be included in CMx2 that was only intended for human to human gameplay, the time/resources to implement that into the game would be significantly reduced as compared to what it would be if a CPU opponent also had to be coded bfor it.

Practically, would BF consider adding specific gameplay features to the CMx2 that are only made "active"/available in human to human gameplay?

eg. allow scenario designers the tools in CM to design "HUMAN vs HUMAN PLAY ONLY SCENARIOS" with a range of new victory conditons (perhaps other features like pre-battle recon) that are EXCLUSIVELY inteneded for human vs human scernarios. Given this freedeom, perhaps BF could develop a new line of "human vs human specific scenario design tools" for scenario designers and game features that could open up the posibilities up creating more "realistic" scenario designs/objectives and gameplay like never before. (I have listed a few ideas on alternate battle victory conditions in this post ).

If you consider the growing number of poeple who exclusively play computer games vs human opponents, perhaps this is not such a ridiculous concept. Just because a CPU can't/hasn't been competently programmed to play the game, doesn't make it a bad game or a game no one would want to buy and play soley against humans. Contrary, it's most likely to be a very good game because the game developers haven't been bound/limited by the need to code/teach a CPU opponent to play it!

I don't know of any CPU opponent who has ever bought a CM game because they liked the experience of playing the game. And I don't think there would be any CPU opponents complaining if they didn't get to play CMx2 using the same features/rules as what is available in human to human gameplay.

Will be interesting to hear everyones comments on this.

Lt Bull

[Cull]

Ok, I'll bite.

"If you consider the growing number of poeple who exclusively play computer games vs human opponents, perhaps this is not such a ridiculous concept."

I won't, and it is.

Seriously though, I play 95% against the AI and therefore I stand utterly opposed to what I think is your proposition, which is to neglect the CPU AI in favor of human v human play.

I entirely favor improvements to the AI, even if means limiting game features, rather than expanded human vs. human features. After all, most of what I think you're wanting---which, by the way, I'm not very clear on---can already be done to some degree with a good scenario designer and/or gentlemanly play between human opponents.

What human v human features do you think are being handicapped (or quashed for CMx2) due to the presence of CPU AI coding? You mention "pre-battle recon, more detailed/specific/realistic scenario objectives, improved treatment of FOW", but could you expand on that, or on what you imagine those could become?

This is thought-provoking, and I'd be interested in a "philosophical" reply from The Guys, vis-a-vis all the different aspects between human v human and human v CPU and how important each is to the game. How much of the workload does the CPU AI consume? How much would be gained, feature-wise, by cutting it back? Etcetera.

Hmmm...maybe time for another "How do you play CM?" poll.

[ August 29, 2005, 11:51 AM: Message edited by: Cull ]

[jim crowley]

I enjoy a good PBEM every now and then and, yes, a good human player can present a great challenge.

However not every human player is uniformly good all of the time; I'm sure I've made more than my fair share of howlers. And some players can be fairly lame.

I also find that I feel much more under pressure to get turns out in a timely fashion, when playing someone else and while I like the competional side of the game I don't get as much enjoyment as I do from playing the AI in "simulation" mode.

In short, its good to have both modes of play available to cater for all tastes.

[simovitch]

I for one would not buy it if there was no vs-AI play.

I would like to see built-in "programmable AI" settings allowed by the scenario designer, like: "hold at all costs", "fighting withdrawal" "probe" "all out attack" etc. I think the Talonsoft series did this, to some varying degree of success.

[Sequoia]

Here's an idea I had that I posted in an old thread that perhaps is worth repeating

I've been kicking around the idea of allowing scenario designers to create AI files for single player games. At the minimum, these files would allow the scenario designer to plot movement etc for units begining with the first turn through whatever turn the ploted movement is completed. Better would be additional conditional movement orders (move until contact is a conditional movement order that already exists). Best would be allowing an aggressiveness setting and allowing the AI to follow a basic plan. Users would be allowed to submit new "plans" for a side which could be set to always follow Plan A or Plan B or to randomly select a plan if more than one exists for a given scenario.

[Salkin]

I'm with Cull on this one.

When I play CM , I want to sit back relax and do it on my own time. Not squeese it in in the hour my human oponent has to spare.

I want to take my time and think and sit in front of the computer with a game of CM for hours just enjoying the experience.

I don't want the sense of being there get ruined by some guy typing :

"Nice move, buddy !"

or

"That was cool ."

or maybe

"Hey, did you watch the game the other night ?!"

Single player is where it's at for me. I don't like the feeling of prestige that comes into it when I play another player, I just like the single player experience.

And I would like it even more with a good campaign map .

//Salkin

[aka_tom_w]

Steve has addressed this issue.

Somewhere sometime ago he admited they had an idea of how much % of the time and which % of the total number of players play against the AI (Solo play)

At that time Steve told us that he figured MOST of the Players play against the AI (solo play) MOST of the time. (or something like that.)

AND as such they would be trying to appeal and sell to that market demographic. Maybe someone else can find his post to that effect, it might even be in the Steve's recent CMx2 bones thread, but I don't think so...

I would say they know how important a good AI opponent is and that they will strive to make the game work so the AI provides the player with a half decent challenge. With the addition of the things they have added to the scenario designer (hopefully ways to better program or "hint" the AI) I would suggest they are working HARD on makeing the AI opponent a real CHALLENGE for most players.

(But I am only guessing)

For me I would say of all the game I have played on CMBO CMBB and CMAK easlily %80 -%90 of those games are against the AI.

(Sometime I wins and sometimes it beats me, but I like to think I win most of the time )

-tom w

[Salkin]

If anyone can find it Tom , it's you .

I also recall reading this post , and my answer in that post is the same as the one I did above.

In that post Steve also admited that there's noway they can do an AI as challenging as a human but they would at least try to make the AI decent.

//Salkin

[Znarf]

I play exclusively against the AI, and I enjoy making scenarios for Human v. AI. I don't have time to play Human v. Human (kids, work, etc.) Battlefront has the best tactical game on the market, but if it were to skimp the AI to make more micromanagement available for Human v. Human players, sales would suffer greatly! Many times I won't even consider a game if the AI is lacking, regardles of the human v. human possibilities.

[Treeburst155]

Yes, it's a shame design decisions, development time and effort must be spent on a computer opponent. People played games long before the computer came along. They played other people; and in the case of wargamers, they often played themselves.

I have no need of a computer opponent. Too bad I'm in the minority. Have you ever played CM against yourself? I promise you it won't be long before you actually surprise yourself with an ambush. It is not difficult to put out of mind the orders you just gave to the other side.

Treeburst155 out.

[Battlefront.com]

If we released CMx2's first game without an AI for single players I can almost guarantee you that there would be no second CMx2 game. Why? Because we'd be flipping burgers (or baking Struedel in Moon's case, making Gezpacho in Fernando's, and spreading Vegimate in KwazyDog's case). So its a non-starter idea.

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. However, if we could cut it out without hitting the unemployment line a few months after release, we sure as heck would. The time that goes into it is bad enough, but the torrent of abuse that is always heaped on the AI (in any game, no matter what) makes us think "why bother". But we know that whatever abuse we take for a "piece of crap" AI the customer has still bought the game and plays against the AI anyway.

Steve

[ozi_digger]

Originally posted by Battlefront.com:

The time that goes into it is bad enough, but the torrent of abuse that is always heaped on the AI (in any game, no matter what) makes us think "why bother". But we know that whatever abuse we take for a "piece of crap" AI the customer has still bought the game and plays against the AI anyway.

Steve

BFdotcom, I own three-and-a-half versions of CM. I've always come back to it after putting it down for a while. Why? Probably the immersion factor, which in a large part is due to the superior AI. So good work guys, I hope the next title carries on the same high standard.

*hug*

I'm gonna take my pills and have a lie down now...

[Lt Bull]

I have already mentioned that I am not suggesting BF kill the CPU opponent off or "downgrade" or "cut it back" it in any way and I know they wouldn't per se, so please don't turn this into a discussion about that. Just the CPU opponent . The quality and kind of challenges the CPU opponent currently provides will NEVER go backwards. There are however alternatives in how BF could approach things, big picture wise.

Im just pointing out that there may be otherwise great feature ideas that could be HUGELY appreciated and enjoyed by a HUMAN player but would present BF with a programing nightmare just trying to code/recode/teach the CPU to understand it/play along with it competently. If BF found out they COULD economically implement special human vs human features into CMx2 WITHOUT comprmising the existing CPU opponent, would they?

Just bare in mind that CMx1 scenario designers are already doing what I am suggesting, at the "expense" of the CPU opponent. Just look at how many CMx1 scenarios have been created with the note "RECCOMENDED FOR HUMAN VS HUMAN PLAY ONLY" using the limited CMx1 scenario design toools currently available! Was anyone complaining then about the concept of separate human vs human scenarios? That this was "denigrating"/"cutting back" the CPU opponent? I am sure many of these scenario designers would have even felt somehow frustrated/limited in their ability to more accurately express their design intent on the scenarios by only being to limited to using a very narrow range of "CPU opponent friendly" scenario design tools. ie. "flags".

Originally posted by Cull:

What human v human features do you think are being handicapped (or quashed for CMx2) due to the presence of CPU AI coding? You mention "pre-battle recon, more detailed/specific/realistic scenario objectives, improved treatment of FOW", but could you expand on that, or on what you imagine those could become?

Take "scenario specific objectives/conditions" for example. Consider scenario designers beign able to include a variety of "victory triggers" or sets of them into scenarios that are very specific to the scneario they are trying to create (either to make the scenario more intersting to play or becasue it more accurate describes the objectives of a historical battle they are trying to simulate). For example:

a) Clear area X of all enemy units.

Destroy X number of enemy armor units.

with conditions

c) Units X, Y, Z can not move until after Turn 6.

d) Without losing more than 4 tanks (or more than 30% inf casualties)

e) Bridge can not be detonated before turn 10. A 200 point penalty will apply if detonated thereafter.

f)Reinforcements X,Y,Z may be released at a 300 victory penalty cost.

You get the idea? Very easy for us humans to rationaliose/reconcile and work with. All that would be needed would be for the game to do the "tracking"/"book keeping"/"score keeping"/play "God" and "referee" for us. Apart from that, it does eveything else it currently does in a CMx1 game.

The kinds of scenario specific/layered/composite victory conditions I am talking about are really no different to the kind and variety you might find in a game like Squad Leader or many other board wargames. I find the kinds of victroy conditions in these kinds of games to be much more varied, interesting, challenging, realistic and sceanrio specific than the current limitations of the CMx1 system. (If anyone does have a copy of SL handy, perhaps they could provide an example of one of the more intersting victory conditions in one of the scenarios).

eg. capturing victory location (flag) X within 10 turns will result in double "victory points".

or

control of victory location X generates 100 "victory points" per turn.

or

double victory points for destruction of enemy Panthers.

Now while all these kinds of conditions sound easy enough for us humans to interpret and to concieve/plot/develop/monitor/alter a plan of attack/defence to achieve these specific objections using our digital army while meeting a set of precice conditions all based on calculated risks and trade-offs and reasoning, for a CPU opponent (or for the people who are trying to code it) it would be a MONUMENTAL task.

Unlike us humans, CPU opponents just don't see and interpret the CM battlefiled and the intentions of the scenario designer like we can. We have personality, we remember things, we can take calculated risks/make assumptions and know how to alter our plans if things start looking bad, we know how to access complex objectives that have specific conditions etc. A CPU only sees "1"s and "0"s and a set number of "algorithms" programmed in it. It doesn't read the scenario brief notes to try and find clues on what to expect. It doesn't even "remember" what happened in a movie even one turn ago.

eg. We watch a movie, and see a KT briefly move into our LOS then out of our LOS behind some woods. We remember that and react accordingly throughout the rest of the battle. It might mean a lot to us. But from a CPU opponent point of view (unless BF can correct me here) it just does not have the ability/capacity to remember/react to this information in any way. If you save the game at that point, shut your PC down and reload the saved game a week later, do you think the CPU opponent "remembers" that it saw your King Tiger move behind those woods last turn a week ago and adjusts its plans accordingly for the rest of the game? Just a very simple example.

Originally posted by Cull:

This is thought-provoking, and I'd be interested in a "philosophical" reply from The Guys, vis-a-vis all the different aspects between human v human and human v CPU and how important each is to the game. How much of the workload does the CPU AI consume?

Yes, that's the intention.

Lots of people are seeing this proposition as a kind of "threat" to the CPU opponent. I really don't undertand why. YOu could still play against the CPU opponent while others play head to head with other players in specialy created "human vs human" scenarios. Not really that differnet to what is happening already.

Originally posted by Treeburst155:

Yes, it's a shame design decisions, development time and effort must be spent on a computer opponent. People played games long before the computer came along. They played other people; and in the case of wargamers, they often played themselves.

I have no need of a computer opponent. Too bad I'm in the minority.

An extreme point of view, but illustrates a good point. You once NEVER were in the minority. Wargames of the past (pre-computers) were not designed with the same limitations as PC wargames are desgned today. That limitation being that a CPU opponent must also be able to play it. Perhaps that is why so many people are so fond of them. The rules/depth/complexity of the wargqme was only as complex and imaginative as what the HUMAN players could deal with. The games were designed exclusively for humans and the rules were designed exclusively to be played by humans. The rules/features were NOT dumbed down to a level that ensured a CPU opponent was also capable of understanding.

Why not KEEP it this way, implementing "CPU opponent unfriendly" features (at little expense) that are available/recommended for human vs human gameplay, while still keeping in the current set of CMx1 "CPU opponent friendly" features for human vs CPU. No harm done, everyones happy. BF saves themselves a HUGE ammount of CPU opponent develpoment time while still making the game richer at least in the area of human vs human gameplay without the burden they normally impose on themselves of considering the CPU opponent.

It's funny. (War)games were designed to be fun/challenging/complex just for humans to play against other humans. That's what they were judged on. PCs come along. CPU opponents come along. (War)games are now designed ONLY if a CPU opponent can be trained to play it. Now people are prepared to give up any rules/depth/complexity/mechanics/concepts of a (war)game they once enjoyed just so that a CPU opponent is capable of playing it against them in some downgraded/simplified version of what they would ordinarily expect to play against humans. Not only that, if a PC (war)game is released with a poor/no CPU opponent, then it considered a BAD game, regardless of what the actual game is. The whole concept/limit/focus of "the game" has shifted from a game to be played by people, to a game that PRIMARILY can be played by a CPU opponent. This NEVER used to be the case. It doesnt have to be the case.

The PC can be used as an interface for human players to interact within a concept of a very complex game (with the CPU just acting as a "referee", "God", "score keeper") free of limitations. However, if you expect to have the CPU REPLACE a human player, it is a COMPLETELY differnt and more complex and difficult proposition.

Doesn't this just illustrate what I am trying to say we are missing out on as human players? The concept of playing against another human (like in every pre-PC wargame of old) takes a back seat in more ways than one to the CPU opponent.

Lt Bull

[ August 30, 2005, 12:08 AM: Message edited by: Lt Bull ]

[Lt Bull]

Originally posted by Battlefront.com:

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."

Is this because the CPU opponent (or StratAI) has not needed to be modified/adapted since it was first developed in CMBO as there have been NO changes at all to the definitions which define the objectives for the CPU opponent in a scenario? If so then that comment may be a bit misleading.

Originally posted by Battlefront.com:

The time that goes into it is bad enough, but the torrent of abuse that is always heaped on the AI (in any game, no matter what) makes us think "why bother"."

Are you refering to the TacAI, StratAI (CPU opponent) or both when you say AI?

You will never get any abuse from me for having programed a "poor" CPU opponent. Perhaps those that do are the same people who just don't quite appreciate how complex it is to try and "humanise" a machine using "0"s and "1"s to take the place of a human opponent in a virtual battlefield who would otherwise plan, concieve, predict, anticipate, gamble, adapt, modify, evaluate, reasses, remember etc using a digital army at their disposal over a series of finite "turns" to achieve a set of "objectives". For what it's worth, I consider the CPU opponent in CMx1 to be more like a kid brother (or a punching bag). Sure it might catch you off balance a few times but it's just not a fair fight right from the start.

The TacAI is vastly more important than the StratAI/CPU opponent for the reasons you have mentioned. But virtually all the suggestions I put forward do not affect the TacAI in any way.

Originally posted by Salkin:

Single player is where it's at for me. I don't like the feeling of prestige that comes into it when I play another player, I just like the single player experience.

What are your expectations of how "clever" a CPU opponent can be for you when you play single player? If BF could train well behaved pet house monkeys to play CM better than the CPU opponent, would that interest you? Just a joke, but conceptually interesting.

Originally posted by Znarf:

Battlefront has the best tactical game on the market, but if it were to skimp the AI to make more micromanagement available for Human v. Human players...

Who said ANYTHING about increasing micromanagement (ie. affecting the TacAI)? Contrary, the suggestions all relate to making the context of the battles more realistic/interesting, making more sense both tactically and strategically without any changes to micromanagement in human vs human battles because you don't have to program humans to understand a battle of strategy and tactics. The more a game like CM is about strategy and tactics the more we will like it. The more CM is about strategy and tactics the LESS capable the CPU opponent will be. It's a fact.

Consider the hypothetical scenario below which tries to illustrate what I am talking about:

There is a set ammount of resources/$/time to develop CMx2. At the very minimum, CMx2 will be a flashier version on CMx1. All the stuff you like and can do now in CMx1 wll still be there in some form. BF ask us what NEW features ("big picture stuff") we want to see ADDED to CMx2. Disregard things like fancier graphics/visuals/sounds or more unit types as these are going to be improved regardless and can not be considered "new" to the game.

Lets say there are 10 new popular innovative suggestions that actually would have implications on the CPU opponenet. (Honestly, I think very few actual gameplay features requested are suggested considering the CPU opponent perspective and all the associated implications. In a way, so they should).

Design Philosophy A says it must be CPU opponent friendly first, and human player friendly second. Perhaps only 4 of those features go in the game and they discard the other 6 (for reasons I have described).

Human vs CPU gameplay gets 4 new features

Human vs human gameplay get the same 4 features.

Design Philosophy B is human player friendly first, CPU opponent friendly where possible. Perhaps 8 features make it in the game, of which 6 are recommended only for human to human gameplay (because they found they were easy to include but just too much work to adequately "train"/adapt the CPU opponent to deal with them all IN GAME to a satisfactory level).

Human vs CPU gameplay get 2 new features.

Human vs human gameplay gets the same 2 new features + 6 other features that are only recommended for human to human gameplay. (eg. optional reinforcements that incur "victory point" penalties)

That's double the total amount of new features actually see the light of day (in some form) with Philosophy A comapred to Philosophy B.

Which philosophy would you prefer BF to take?

Lt Bull

[ August 30, 2005, 03:38 AM: Message edited by: Lt Bull ]

[markl]

The problem is that forums like this do not reflect the actual customer base. Most people play against the computer and probably hardly if ever post in a forum.

And lets be honest it is not a case of 2 features versus 2 + 6 features. Battlefront have prooven they will give us all 12 new features weather we need (want) them or not. Even the current games are feature rich compared to the oposition.

Anyway when all the talk is over I am sure they will provide a very balanced game, as their record shows, and relatively bug free. There will always be something to complain about. But that is BattleFronts problem as they listen to us wingers.

I do not know of any other developer who serves their customer base like BattleFront.

Anyway off to play CMBO. (Still the best of the three, more fun factor)

Cheers MarkL

[Battlefront.com]

Is this because the CPU opponent (or StratAI) has not needed to be modified/adapted since it was first developed in CMBO as there have been NO changes at all to the definitions which define the objectives for the CPU opponent in a scenario?

No. We aren't reusing any code from CMx1 for any feature. What I meant is that if we spent 1 day coding the CMx1's StratAI we probably spent 5 coding the TacAI. It is the TacAI that holds the entire simulation together. Without that, we got bupkis. So chopping down the StratAI or making some features not available to it really doesn't do much for us. Adding features that don't work with the StratAI, therefore, come as additional time spent on the title. And that means all this talk about what to do or not do with the StratAI is pointless. There is nothing to change.

Steve

[Battlefront.com]

Oh, and I also meant to comment on this train of thought:

many of these potentially great ideas may be/will be/have been whacked on the head by BF just because the implications of including them in the game would be too complex/difficult/time consuming when considering the coding/re-coding requied for the CPU opponent to understand them.

Incorrect if you are thinking StratAI. The primary reason for killing features is usually either time or hardware limitations or (far too often) both. AI doesn't even factor into it. When AI does factor into things it is the TacAI not the StratAI.

For example, allowing Soldiers to stand on their heads and breakdance takes a bunch of time to code and do animations for. But it is the TacAI where the real problems lie. The TacAI is the part of the game that needs to know in what context is breakdancing the correct choice, what it needs to do prior and post dancing, how it affects other behavior associated with outside factors (like being shot at while dancing vs. dancing in mud vs. being on top of a vehicle, etc.). The StratAI probably doesn't even need to know about this.

Yeah, a silly example, perhaps, but it is valid.

Steve

[Cameroon]

Originally posted by Treeburst155:

Yes, it's a shame design decisions, development time and effort must be spent on a computer opponent. People played games long before the computer came along. They played other people; and in the case of wargamers, they often played themselves.

I have no need of a computer opponent. Too bad I'm in the minority. Have you ever played CM against yourself? I promise you it won't be long before you actually surprise yourself with an ambush. It is not difficult to put out of mind the orders you just gave to the other side.

Treeburst155 out.

I heartily disagree. I deal with people all day long, and often enjoy computer games specifically because I can do something within an environment that is at least somewhat challenging (and sometimes very challenging) that doesn't include any people.

I like playing real, live opponents, too from time to time, but the vast majority of the time I play vs the AI. In fact, I'm in the (probable) minority that doesn't care if there's multiplayer (or multi-multiplayer). Though if I had to pick one form of mutliplayer, it'd be PBEM rather than TCP/IP. [Edit] And by minorty, I mean those who really don't care about multiplayer vs those who wouldn't use it, but would complain if it was missing.[End edit]

And either I give fewer orders or my memory is better, because I remeber it all way too well to play vs myself with any kind of surprise involved

[ August 30, 2005, 09:12 AM: Message edited by: Cameroon ]

[Cameroon]

Drat

[Battlefront.com]

Treeburst,

People played games long before the computer came along. They played other people; and in the case of wargamers, they often played themselves.

Or they bought the games and let them gather dust because playing solo was about as exciting as watching paint dry and finding other people (or the time) to play against was too difficult. So they didn't play much at all. I think the majority of wargamers have sounded off on that point pretty consistently since computers became a viable tool. So I think it is pretty safe to say that the way games used to be played was simply the way it was, not the way the vast majority of people wanted it to be.

Steve

[dalem]

Originally posted by Battlefront.com:

So I think it is pretty safe to say that the way games used to be played was simply the way it was, not the way the vast majority of people wanted it to be.

Steve

That's the way I see it too. I certainly can flip the board around and play each side in turn - we've all done it one time or another - or play some of the chart&table driven solo games, but having a computer take care of the other side for me is gold. Pure gold.

-dale

[RSColonel_131st]

Of course you realize that we now want a breakdacing command?

[Ron]

I won't argue the importance of TacAI versus StratAI, but to be honest I think the StratAI in CMx1 was the one needing the greatest improvement. I was almost always happy with the TacAI and chalked up the 'bonehead' things it did sometimes to the fortunes of war. If you had a sound plan then everything eventually worked out. I couldn't say the same about the StratAI which never even had any marbles to start with.

Ron

[Cull]

Originally posted by Battlefront.com:

If we released CMx2's first game without an AI for single players I can almost guarantee you that there would be no second CMx2 game. Why? Because we'd be flipping burgers (or baking Struedel in Moon's case, making Gezpacho in Fernando's, and spreading Vegimate in KwazyDog's case). So its a non-starter idea.

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. However, if we could cut it out without hitting the unemployment line a few months after release, we sure as heck would. The time that goes into it is bad enough, but the torrent of abuse that is always heaped on the AI (in any game, no matter what) makes us think "why bother". But we know that whatever abuse we take for a "piece of crap" AI the customer has still bought the game and plays against the AI anyway.

Steve

First, glad to see there is no consideration for reducing or eliminating the single-player aspect. Not that I thought there was.

Second, regarding the "torrent of abuse" :

I don't have to tell you guys that the complaints about the AI are, by and large, simply frustrations that stem from the limitations inherent to AI. That's pretty much game industry-wide.

I'm fascinated with AI. It's amazing how far it's come in gaming and elsewhere, but I lean toward the camp that says there really is no "I" in "AI". Yeah, you can make it seem smarter the more work you want to pour into it, but it's still just "smoke and mirrors". Until, of course, AI becomes learning, creative, and unpredictable....basically more human. But we all know what happens then: your computer eats your dog because it's jealous of your girlfriend, that's what.

But I digress! Mostly I just wanted to say that the AI in CM is, despite it's general and specific shortcomings, very good. I won't say "great" because I have yet to see great AI in any game.

To digress some more: the more realistic games/simulations become the more frustrating the AI (and other) issues. By that I mean: they're so close to reality now that the "problems" stick out that much more. I'm sure somebody knows what I'm saying...

No really, does anyone know what I'm saying? 'Cause I don't.

Lastly, to Lt Bull:

I see your point much more clearly now. I got a little defensive when I thought you were saying "Cut the single-player and put more in for multiplayer!" that I sorta went all googly there for a minute.

Oddly, Railroad Tycoon 3 comes to mind. I won't ramble off on the gory details, but that game has some excellent (built-in) scenario design tools---essentially very simple scripts and triggers---that are surprisingly easy to use. I wonder....

I now I think I'd better go take my medication and lie down.

[Gordon]

Originally posted by Battlefront.com:

For example, allowing Soldiers to stand on their heads and breakdance takes a bunch of time to code and do animations for. But it is the TacAI where the real problems lie. The TacAI is the part of the game that needs to know in what context is breakdancing the correct choice, what it needs to do prior and post dancing, how it affects other behavior associated with outside factors (like being shot at while dancing vs. dancing in mud vs. being on top of a vehicle, etc.). The StratAI probably doesn't even need to know about this.

Yeah, a silly example, perhaps, but it is valid.

Steve

Ok, we can now rule out another possible topic for the first CMx2 release. It won't be "CM: Grand Theft Auto."