Balancing out commanders and the commanded

[Tarkus]

Originally posted by Battlefront.com:

teams of players that don't work together will likely fail together.

Eheh, being the company and the game designers that you are, I kind of figured that one out. I was refering to the nuts and bolts of the system but it is obviously too early to discuss it. With all this recrudescence of bones all over the place, I tend to get over enthusiastic about where all this is going and hope for still more details all the time.

Thanks for letting us in a bit. It's really nice to get a glimpse of where you guys are going.

Best

[Tarkus]

As for AI controlled friendly units, IIRC in the what we want thread there was this idea of using this feature as a potential counter to the table-top effect of map edges, like the map would be bigger with assets a player doesn't have any control on but that are set by the scenario designer to simulate neighbor forces, ready to defend. I guess that would be a feature to be cautious with...

[Battlefront.com]

AI will never be even remotely as good as a Human player. Well, not that we can realistically put into a commercial product anyway! So when people dream up great ways to have the AI come in and directly affect their ability to control or at least win battles... keep in mind that these AI players will be marginal at best, complete and utter fools at worst. Yeah, Human players span that spectrum too, but generally you don't care because you're playing AGAINST them instead of with them

If AI were easy to program there are tons of great things we could come up with to keep the CPU cycles from being wasted. But the reality is that we would be fools to make a game design, at this low tactical level, be dependent upon AI for core game features (other than the opposing player, of course). We'd get more curses than cheers from players so why bother?

Co-op play does involve a lot of things that I simply can't think about right now, so I can't answer any questions and won't retain anything that people suggest at this point. Having said that, since we are designing the game to be as realistic as possible, many of the potential features needed for co-op play will already be there "by accident".

Steve

[Jaws]

As I see it.

The AI can be divided in levels, for CM it will be squad, platoon, company and battalion. Basic for these levels are attack (go forward), defend (halt) and delay (retreat).

The most important basic must be ATTACK for the AI as I will try to explain in some examples.

I try to tell this so simple because it is the most easiest to learn in real life to soldiers and it is easiest to program (IMHO). So what must be programmed is some triggers to tell the AI to do one of the above mentioned disciplines.

I.e. if a squad meets no enemy… he will attack (go forward). If he makes contact with the enemy he will defend (halt) until the odds are good enough to attack (go forward) or if the odds are bad he will delay (retreat)(not panic as CM squad do now).

From company level and higher we can talk about combined arms and that’s an other AI to program. Basically you can divide combined arms in tanks, infantry, arty and engineers, all these disciplines still have their basics attack, defend and delay. Only certain terrain types are more likely to enter for one of them because of their nature.

We are still talking about AI here.

So now combining levels. A infantry platoon conduct one of his basics the attack, but is halted in open terrain and start delaying to some trees in the rear. Here comes the combined arms AI who will throw in some tanks and arty (because open terrain is more preferable for them) and because ATTACK is the most important basic for the AI.

Now there is only one thing left… the objective… well in general… take a history book and we will find:

Hills (like hill 112 or 123 Normandy)

Crossroads (Bastogne)

Bridges (Pegasus)

Towns (Caen)

Dominant terrain pieces

Of course you can have specific object like a gun battery but than we must scripted scenario’s and here we are just talking about a general AI.

Ok start shelling

[A.E.B]

Hi all

I realise that my limited orders solution was just a band aid to what is in reality an intractable problem of realism vs playability and fun.

When I was designing my game (a pen and paper RPG with added computer resources) I ran immediately into a problem: I wanted a high degree of realism in how things worked (magic had to have its own set of consistent rules that tied into the melee/missile and skill system in a meaningful way (ie the player should never come to question how the rules were designed).

Sounds easy?

My first effort was a complete failure, I designed from the base up and generated a cumbersome nightmare that failed to free the players from the game mechanics so they could immerse in the game.

For my second effort I flipped the whole thing around, and started with the results I wanted and then created a rule system that used the same system of resolution for every action, but required only one 2d10 die roll to determine level of success or failure combined with a card system to through in major variables that complete alter the game balance from action to action.

The playtesters loved it.

But this game was under the hood nothing but arbitary statistics, a couple of tables, cards and die rolls. The way the rules worked were not realistic at all, but the end result appeared not only realistic but was simple and thankfully fun.

This I believe is the challenge facing the next CM - how to achieve a final game result that feels realistic while remaining both fun and simple to utilise (ie the interface, not battle tactics).

I agree that the AI is a issue: I played most CMBB senarios from the CD but only played one from the CMAK CD. The AI was simply unable to provide a realistic challenge as it lacks human unpredictability. Play a human and you will often be surprised but new inovative tactics, or a reaction that renders your own plans mute.

The AI is easily baited to its death, and is often predictable.

This AI issue confronts all games where a human faces AI. In steel panthers the fastest units of an attacking AI would arrive first, so you had a wave of transports, then tanks, then infantry. Even games like Half-Life use clever scripting to cover the predictibity of much of its AI, but given the same situation the AI in even that game would produce the same reactions.

In my effort at computer game design the AI was a disaster. Armies with melee and missile weapons have to operate as a whole, not split into randomly reacting units. This meant that each AI unit had to react to what every other AI unit in sight was doing. Hence the units of pikemen had to advance as a solid line, or attackers would attack the flanks. With if/else tables of reaction for each AI unit, a AI army functioned as a mob of uncordinated individual units. Programmed to follow the leader, often nothing happen as the lines froze as each reacted to the surrounding units reacting to them.

To use a CM example, the AI does not coordinate AFVs and infantry. Tanks advancing into terrain that allows for infantry close assault should be escorted by infantry. If the escorting infantry is engaged, both the tank and its infantry should fight as a whole.

So what the CM engine should aim at is, given a tank and infantry want to advance into a town say, that the tank and infantry should try to cooridinate their actions: how that result is achieved is of secondary importance.

This is only constructive critisism. I have been playing CM now for years, longer than any other game. I have played several hundred PBEMS. And I keep playing for the challenge and the fun.

CMX2 needs to keep the spirit of its predessessors while updating the engine that delivers that spirit. DOOM3 is an example of a game that failed to keep its original spirit with newer software. I believe that CM will not fail that test.

A.E.B

[Tarkus]

It is notorious that coding a realistic, intelligent AI is one of the biggest challenge in the world of computer, for many reasons that are stated here. If creating a chess player is so complex, and chess, however marvelous the game is, still is a game of strict contingency, it's really not that hard to understand how complicated an opposing AI, with way more options on his hand, not the least of which is appraising the relative efficiency of the tool at his disposal, must be. To keep the comparison with chess, a Queen is a Queen (you do not, AFAIK, have a Queen A1, a Queen A3E8 HVSS, etc) and yield the same results every time: it moves in a certain way, and kill all the time. A Sherman, on the other hand, is very useful in some instances, and pretty useless in some others, and many times, it depends on how you deploy it in accordances to terrain, friendly forces, intelligence, ennemy equipment and deployment, etc.

That is why I thought a way to import human imput in the tacAI might be an interesting addition. Supposing this could be toggled on and off, one could play against the intended plan prepared by a human designer, perhaps download and play against other plans for the same scenario, or against the AI we know and love , or against a human player. I suppose it could simply boils down to allowing the designer to plan ahead, and I already forsee some big problems with that approache, but still, I'd be curious to hear what the people that knows think about that. Sort of a scripted, optionnal, modular AI.

EDIT: Of course, this idea does nothing to help a tank commander to decide on his own, during a turn, to smartly react to a spotted threat, which is yet another side of the whole AI equation, but still...

Cheers

[ January 18, 2005, 05:48 AM: Message edited by: Tarkus ]

[John Kettler]

Fascinating thread!

As I see it, our gaming salvation may (note use of conditional) lie in several different elements artfully blended together. They are:

Graphic control measures, to include map grids, unit boundaries, phase lines, artillery/mortar/rocket concentrations (NOT our present TRPs, in addition to), LDs, FPFs/FPLs, assembly areas, objectives, no fire zones, etc. If we had these, we could get the AI to behave more realistically and greatly simplify battle planning and setup, not to mention creating all sorts of juicy coordination problems, vulnerabilities and so forth. There's a reason attacks are aimed at unit junctions!

Formations for all levels depicted, be they the infinitely lamented convoy/follow order, the classic two up and one back attack, AFV formations (line, wedge, diamond, left/right oblique, column, coil, herringbone, etc.) If we had these as pulldown menus or similar, we could save a great deal of time and frustration.

SOPs, a much requested feature which would also simplify troop handling, improve artillery use and much more. Simply being able to call in fire support at harassment, suppression, sustained fire and max ROF would be a huge improvement, let alone if we had historically available fire patterns, fuze options and shell types. Imagine the possibilities if units had SOPs which automatically precluded much of the present martial insanity. What if instead of having to constantly micromanage every squad, we could issue simple orders most of the time and save micromanagement for the situations which really need it? Why should we have to kill ourselves to move a company up the road? "Company A, route march to (insert grid or place name here)."

Order limits. Believe it was GDW's Assault series which used these effectively to illustrate the very real differences in flexibility between NATO and Warsaw Pact forces. The WP had fewer orders per turn than NATO, but could do a lot with standard battle drills. The downside was that it was hard to

respond effectively to a fast developing situation.

Reaction checks. Long a feature of gaming with miniatures, these could be done inside the game engine when a variety of events occur and could go a long way toward taming the Borg beast. For a range of examples, please see MODERN WAR IN MINIATURE, by Michael Korns. Goes into spotting, ability to react in a given time period, environmental and lighting impact, fear effects, etc. I believe that such checks, cross indexed with

troop quality/morale/casualty level/health/fatigue and a random factor, would add much to the game--without making more work for the players. Were such measures in place, then modeling things like delayed tracer ignition on German MGs and smokeless/reduced flash powder from German tank and antitank guns would automatically act to enhance the utility and survivability of both. These were real advantages which are historically documented yet aren't modeled in the current engine, whereas things likethe Allied traverse rate advantage is. Where in real life a single Pak or SP could bring an advance to a screeching halt while a frantic search was made to locate and deal with the shooter, in the present game it's more like such a weapon gets off one or two shots before being spotted and pounded. Likewise, we need to be able to represent the effects of things like camouflage netting and measures to suppress the firing signature (tarps, watering or oiling the ground) of DF guns. And rangefinders!

C3 modeling, to include command delays, the short range limits for standard intervehicle sets (making radio cars valuable in certain scenarios), at least simplified net depiction (w/ breakdowns as random and event driven events), clear differentiation between vehicles with transceivers and receivers only. For example, during the battle of Kursk, the captured T-34s (all w/ transceivers) of Das Reich sat on a hill and systematically killed the company command tanks in the oncoming waves of Soviet T-34s, the only ones in the unit with transceivers. Similarly, during Goodwood the Germans systematically targeted command tanks wherever found (even KOed the ALO tank), creating a great deal of disruption. ISTR that in TANKS FOR THE MEMORIES (www.tankbooks.com)only two tanks, the platoon leader and his deputy, in each U.S. Sherman platoon were fitted with transceivers.

Redistribute/share ammo order. Shows up in quite a few accounts, would add realism and considerably enhance value of reinforcements, especially if some sort of simple logistics model were implemented (unit of fire for ammo, for example). If ammo and fuel could be transported and stored during a game, then trucks and other transport would be far more accurately depicted, less cavalierly used, and would pose a real tactical concern to commanders at all levels. Supply issues would get important in a big hurry, as would rear area security, air defense and more.

Immersion. Enhance via radio/field phone reports tied to battlefield events/higher command decisions, etc. A C64 game years ago had a message crawl from units which spotted something, took lumps, were surrounded, etc. How difficult would this to do as VO?

Briefings and map integration. Love the way this was handled in Panzer Elite.

What do the rest of you think about these ideas when implemented in combination? I see lots of potential, but I'm not a programmer.

Regards,

John Kettler

[aka_tom_w]

Originally posted by Tarkus:

That is why I thought a way to import human imput in the tacAI might be an interesting addition. Supposing this could be toggled on and off, one could play against the intended plan prepared by a human designer, perhaps download and play against other plans for the same scenario, or against the AI we know and love , or against a human player. I suppose it could simply boils down to allowing the designer to plan ahead, and I already forsee some big problems with that approache, but still, I'd be curious to hear what the people that knows think about that. Sort of a scripted, optionnal, modular AI.

Cheers

"Sort of a scripted, optionnal, modular AI."

Scripted or "hinted" AI for scenario design

Especially on defence could provide some REAL challenges for the human player.

I think we are asking for some better ways to "hint", script, or program the AI, specifically in scenario design.

The idea of invisible flags has come up, but perhaps there might also be (also invisible) things like primary or secondary objectives or objectives along the way (path of attack) that "should" be held and taken at certian time intervals (+/- 3-5 minutes) within the scenario.

Perhaps with some of these new tools (programable scenario design aids), a scenario could be designed that would give the AI (especially on defense, or on the attack in very specific conditions) a fighting chance or even provide a real challenge for some newer, greener players.

-tom w

[ January 18, 2005, 07:10 AM: Message edited by: aka_tom_w ]

[Tarkus]

There are some real nice ideas up there John! It kind of sums up the recent what we want threads. I particularly like your "reaction check" idea. Perhaps that would justify another thread to expand on it ? From here it looks as if it might add some random results within a realistic frame...

Any idea about how a comm net might be implemented ? I'm interested in these questions as well, but I know no games that tried this, let alone succeeded.

All of this is quite interesting. I guess we're not yet to the point of flooding these poor guys with our zillion suggestions (yeah, like we did not already ) but I feel we can brace ourselves for that moment. Maybe it's time to resurrect that old CMx2 suggestion thread. Or prepare a summary to be launched the very first day the CMx2 forum will be on. EDIT: I volunteer to set it up.

Best

[ January 18, 2005, 05:51 AM: Message edited by: Tarkus ]

[aka_tom_w]

Originally posted by John Kettler:

Fascinating thread!

snip

C3 modeling, to include command delays, the short range limits for standard intervehicle sets (making radio cars valuable in certain scenarios), at least simplified net depiction (w/ breakdowns as random and event driven events), clear differentiation between vehicles with transceivers and receivers only. For example, during the battle of Kursk, the captured T-34s (all w/ transceivers) of Das Reich sat on a hill and systematically killed the company command tanks in the oncoming waves of Soviet T-34s, the only ones in the unit with transceivers. Similarly, during Goodwood the Germans systematically targeted command tanks wherever found (even KOed the ALO tank), creating a great deal of disruption. ISTR that in TANKS FOR THE MEMORIES (www.tankbooks.com)only two tanks, the platoon leader and his deputy, in each U.S. Sherman platoon were fitted with transceivers.

snip

John Kettler

Fascinating post!! (where's the smiley for EXCITED)

Command delays are a BIG issue, I suspect the guys at BFC have been struggling to get them just right...

John, your post is very interesting but I am not sure that Steve is all that interested in a detailed model of ALL possible radio communication within a scenario or within the game. (I hope I am wrong about this) BUT in the past I think he has said that simply modeling ALL radio commuication (the com net) is NOT the be all and end all of solving all problems with realism or command delays in the game.

Command delays seem to be REAL dodgy grey area that they can make as "fuzzy" as they want (given there are likely very little historical data on what is the EXACT command delay for one player commanding a Battalion in a war game sim ). This is not a bad thing as it is a variable they can tweak ANYWAY they want and no one can really call them on it on the grounds being ahistorical or just plain NOT right.

BUT what I am still hoping is that if they are still using command delays in the new engine there can also be some form of spotting delays so info going up AND down the chain of command is somehow slowed down or made uncertian or degraded BOTH ways. (I am not real sure how this might actually WORK in the game and what good or positive impact it might have on game play BUT I hope they will try it out )

What about units getting the WRONG orders (frustrating for the player ? you bet!) the added uncertianty would be interesting, but how to do it JUST right would be the tricky part. Perhaps 1 order in 1000 was just plain WRONG. In the past orders issued to units have NEVER been misinterpretted or just plain wrong, NOW how do you really model that in a game?? :confused: I have NO idea.

BUT way back in the Early CMBO days Steve mentioned (after a ride/goose chase (of sorts) in the infamous Weasle), that it would be fun and/or interesting to model units getting LOST in the game. I have often wondered how my units NEVER ever got lost in the woods at night when I asked them to sneak or attack through deep woods either in the thick fog or in darkness.

Now, ot be fair they did try to model friendly fire at night, and if I may say so the response was (at least on my part) never to play night time battles.... (BUT it was a GOOD idea on their part.)

That was a VERY informative post John!

Thanks

-tom w

[ January 18, 2005, 07:23 AM: Message edited by: aka_tom_w ]

[aka_tom_w]

Some things must be stimulating their creative juices...

The Skunkworks (secret) forum now has 289 post (maybe about CMx2 who knows?) and the last post was last evening:

January 17, 2005 10:31 PM

They are FOR sure having internal discussions and just some of them may be reading what we are posting here in this public forum.

he he

Keep a close eye on that Skunkworks forum and you can see the last post and how many posts have been made. (but the damn door is still locked! grr)

-tom w

[ January 18, 2005, 09:24 AM: Message edited by: aka_tom_w ]

[Michael Dorosh]

Originally posted by John Kettler:

Fascinating thread!

As I see it, our gaming salvation may (note use of conditional) lie in several different elements artfully blended together. They are:

What do the rest of you think about these ideas when implemented in combination? I see lots of potential, but I'm not a programmer.

Regards,

John Kettler

You're slipping well into simulation territory, and away from a game. Radio reports would be cool, for example, but there are many players who don't want to have to learn radio jargon, know what a "phase line" is, or need to clear their own line of departure, set up Forming Up Points, wounded nests, or ammo points...they just wanna blow stuff up...I'm not one of them, but you do need to take into account your market.

Being a programmer isn't necessary to see these impacts; having a feel for consumer base is...the more technical detail you put into the game, the less successful it will be commercially. Period. The question is where to draw the line.

Think of it as a flight simulator; the more technical options the player must be proficient with (propeller pitch? fuel mixture?) the fewer people will buy it.

[ January 18, 2005, 09:56 AM: Message edited by: Michael Dorosh ]

[Battlefront.com]

Keep in mind that in order to make some of these suggestions work we'd have to do a heck of a lot of "active" AI modeling. Active AI is that which does not follow scritps and does more than simply react to stimuli. It is, for lack of a better term, capable of independent and on the fly "thinking". This is the hardest form of AI to program and it is therefore not surprising to see so few examples of it in the real world (or Universities ). To expect us humble game makers to come upw with something that thousands of highly eductated and specialized AI programmers can not is rather, well, silly. It simply is not going to happen.

Here is an axiom that has yet to be proven wrong... the more a game design relies upon active AI to make a game worth playing, the less likely the game will wind up being worth playing. Some games, like most of the RTS titles out there, have practically no AI at all of any sort, yet they can be a lot of fun to play. FPS games have similarly have little ot no AI programming, instead relying on trigger points and reactive AI scripts. Vehicle sims (flight, tank, ship, etc.) also have very limited amounts of AI programming, and again... great stuff. So on and so on.

What this axiom means for us is that the more we make the core game functions dependent on AI in general, and active AI in particular, the more likely we will wind up with a game that will not be worth playing. Therefore, do not expect us to come up with designs to work around Borg or other issues which use extensive AI work. It just isn't the way to go.

Steve

[Battlefront.com]

Radio net simulation doesn't solve anything relating to Borg or God issues in and of itself. It is nothing more than a fundamental building block to be used by other designs to simulate C&C. Think about it... picture CMAK with a full on radio net simulated. Will that suddenly stop you from ordering a beat up squad from running up and over a hill to see what is on the other side and getting that information to you? No. Will it magically simulate the disruption that the loss of leadership brings to subordinate units, especially ones with inherently less initiative? Nope. So on and so on.

Likewise with smokeless powder or getting lost at night. These things can only be realistically simulated to the degree the Borg and God stuff are curbed. It does no good to add an extra hiding modifier to an AT gun when the enemy already has far too many chances of spotting it in the first place (note that AT guns are harder to spot, inherently and artificially, within the game as it is right now. If they weren't, they would be even easier to pick off). Getting lost at night is not possible when the player has complete knowledge of the map and where his units are at any given time. So on and so forth.

These are examples that fit right in with this thread's origins. A simulation must be seen as having parts that are individually fairly meaningless, but when combined together create something much bigger. That is why we are attacking this thing from the top down as we did with CMx1 and as AEB did with his game. CMx1 is great because it is result orientated. We gave the simulation a vastly detailed basis and then figured out what kind of things we wanted the player to experience with all that otherwise meaningless data.

CMx2 is being designed the same way, but with the added benefit of almost 8 years of CM development and 12 years of game development experience. We have a very good idea what we need to do and how we need to do it. Now we just need to actually do it

Oh... I wouldn't pay too much attention to the Skunkworks message count. It really doesn't mean much. Most of the work is happening on Charles and my computers and not in any sort of forum at the moment. Too much input too early is a bad thing

Steve

[willbell]

Steve,

I remember from one of your older posts that CM is not a simulation, it's a war game. Strictly speaking, a simulation can only be from and individual POV. So CM is an abstraction, a way out way of playing army men. So the most important question is not whether it is accurate in all it's details, but accurate enough that we feel like we are fighting a real battle, but get to do all the fun stuff too (tanks, canons, machineguns, flame thrower etc.

So a multiplayer option can do one of two things. It can let multiple players each play a group of battles that are somehow all inter-related in goal, or it can let more than two players play the same game where there is for instance a two battalions on each side and each of the four players run one of them.

The first option is in addition to, and doesn't effect the regular two person game we know as CM. It is a program that is almost a game in and of itself, it manges a kind of metacampaign with several independent CM battles going on at once. And it watches a big map, and lets the individual games know how their battle effected the big map, and where they will be on the next turn. One of the group of players is the commander and gives orders as to where next on the big map everyone is to go. And on the next turn everyone plays a battle again, each one being unique and based on what happened on the big map. So you are always playing CM like we do now, but in a wider context. But of course there are all kinds of abstractions to deal with here, but work them out in the spirit of the design of CM to begin with.

The second type is obvious, it requires more memory, but the some units are controlled by commander A and others by commander B, and they move them about either cooperatively or not, and the results are worked out by the game engine.

[aka_tom_w]

Originally posted by Battlefront.com:

Oh... I wouldn't pay too much attention to the Skunkworks message count. It really doesn't mean much. Most of the work is happening on Charles and my computers and not in any sort of forum at the moment. Too much input too early is a bad thing

Steve

RATS!!!

Good thing there is plenty of POSITIVE commuication out here in the public forum then!!!!

Many Many thanks to Steve for all the insight into CMx2 development progress.

As you can tell some of us are WILDLY excited about the possibility to have some imput into the design and development, even if all you are actually doing is reading our words and idea's and laughing out loud.

Thanks for telling all the things the game cannot and will not do.

(As well as the features you hope it will include). It certianly helps frame our expectations. And for that we all thank you!

-tom w

[aka_tom_w]

Originally posted by Battlefront.com:

Here is an axiom that has yet to be proven wrong... the more a game design relies upon active AI to make a game worth playing, the less likely the game will wind up being worth playing.

Therefore, do not expect us to come up with designs to work around Borg or other issues which use extensive AI work. It just isn't the way to go.

Steve

If I may.....

May I suggest this list of resources:

Book list of AI resources and sources

I am still hoping to locate the ulimate "How to" guide free on the web.....

don't laugh I am betting its out there somewhere...

So says this guy:

"Useful Publications on AI

I don't have many books on game AI, actually. Most of what I know is self taught through hard experience and experimentation. The web is, frankly, the single best source of information on this topic--no group of dead trees is ever going to be able to match it for sheer volume of information. I have picked up a few tomes that I've found useful, however, and they're listed below. I've found each of these books to be of interest and use. I recommend them all.

I've also come across some other book lists that may be useful; they're listed below. "

Book list of AI resources and sources

online AI resources

Open source freeware for AI

Quote: (this one AI Geek with his own web page says this:)

"CMU Artificial Intelligence Repository -- If you don't already know and use the CMU Artificial Intelligence Repository, you should. This university-sponsored site is chock-full of information on every facet of AI research, and has an entry form for submitting other sites of interest. Highly recommended."

Open source repository of freeware for AI

or this: (?)

Discussion forum for AI in game development...

now where is that online "how to make a video game AI guide"

Posted in the Very Best of humour in the spirit of helpful co-operation...

-tom w

[ January 18, 2005, 12:05 PM: Message edited by: aka_tom_w ]

[Battlefront.com]

A simulation is an environment created to recreate some sort of process (generally real world, but not necessarily so) independent of that process. The intention of the simulation is to show the observers, whomever they might be, what is possible and not possible in that given environment. Simulating a factory assembly line, growth of mold spores, the characteristics of an engine in different climates, etc. all fit that definition.

Games are in and of themselves not simulations. They are merely abstract concepts that are put together to entertain, teach, or both.

Wargames are generally a mix of both game and simulation. The more abstract the system, the more it is a game and not a simulation. The more the system behaves according to real world variables, the less of a game and the more of a simulation it is.

We consider Combat Mission, in all its flavors, to be a simulation first and a game second. Meaning that we have attempted to simulate real world characteristics of combat before there is any thought as to how those factors "play" as a game. It is only after we have the sim elements in place (on paper at least) that we figure out how it affects the game. We then generally shape the game to fit the sim, not the sim to fit the game. This last point is the key difference between CM and the others that came before it or since.

It just so happens that in making a great sim we made a great game at the same time. That was, afterall, our intention. Many said it couldn't be done, including many of our now devoated fans, but we proved them wrong. Yea for us

Steve

[Battlefront.com]

As an example for the above philosophical point...

Chess is what? A game? I think nobody would disagree with this. A wargame? Few would argue this to be the case, though some do. Is it a simulation? Nobody can argue that it is.

Combat mission is certainly a game, definitely a wargame, and arguably a simulation. Now, how good it is at each is a matter of opinion, but particularly for the first two. The latter, the sim claim, is the one thing that is more emperical than personal opinion. The stronger the arguments are that something is a sim, the more likely it is. Few sims are perfect so perfection is not necessary. Just surpassing a certain threshold of realism is enough.

Steve

[Sirocco]

A fundamental problem with more realistic gameplay is that it must lead to a more confused battlefield.

As an example, a platoon is advancing with two squads. The first squad reaches it's objective and as it does so spots a HMG which is in a position to fire on the second squad which is moving to it's own objective. The second squad cannot yet spot the HMG itself, and is therefore unaware of it.

If we assume the second squad is out of earshot of the first let's suggest, for the benefit of simplicity, that the first squad sends a runner back to the platoon leader, trailing the first two squads.

Now the platoon leader is aware of the HMG. He sends a runner to the second squad. When the runner reaches it the second squad becomes aware of the HMG.

Again, for simplicity's sake, we'll assume CM1 behaviour, and the second squad goes to ground in the nearest available cover.

Assuming there is no radio link between the platoon leader to the company leader, and from the company leader to the battalion leader, the platoon leader sends a runner back to the company HQ with details of the contact.

As battalion leader you're unaware of this. You won't see the changes to the situation until the runner reaches the company commander, and the company commander sends a runner to the battalion HQ.

But what happens to the platoon in the meantime?

As a designer there are two options. You can have the AI make a decision, with all the attendant problems related to that, or you can allow the player to set certain SOP guidelines.

When the battalion commander is made aware of the situation through reports, the map updates, and orders issued have a clearer appreciation of the situation.

The end result is that the player might feel that he's grasping at clouds as the battalion commander, in perhaps the same way that people wrestled with Russians attacking over open ground in the CMBB demo.

[kipanderson]

Hi,

The posts from Steve over the last 24 hours have been hugely reassuring. Those like myself who are at heart quite happy with CM and therefore nervous of some fundamental change in direction…. clearly have nothing at all to worry about.

CMX2 is even being specifically coded with multi-player features in mind for some future version. Even the type of multi-play most likely, simply handing out the platoons and companies, is my preferred model. At the sharp end it would give the best game play, and the most realistic in chaos terms, at the sharp end. In my view. But no matter on the type of multi-play now.

I will not go on for long… have got the information, bones I hoped for from Steve.. and more… but I do think some here are trying to do with CM, hoping to do, things that are best done at another scale. As Steve has also said. Best done in a game where the manoeuvre units are, say, platoons played over a 1:10,000 topographical map.

All the best,

Kip.

[Dandelion]

Originally posted by aka_tom_w:

In the most positive and collegial way possible I would like to wholeheartedly disagree!.

It IS a LARGE part of the problem for many of us here for sure!

With every fiber of my being I bristle at the gamey jeep recon and the bailed out tank crew WAY way out of LOS or C&C (say 500m) that can report back to all other units via some Borg like collective consciousness, ANY and ALL possible spotting information about any and all enemy threats!

This can and should be neutralized or at the very least degraded in some meaningful way!

Absolute spotting means when one unit sees something ALL other units become instantly aware of it. I am NOT sure if we are talking about the same "Absolute Spotting/Borg collective conscious problem?"

If you suggesting that the absolute spotting paradigm as implemented in CMxx is acceptable then perhaps we are NOT playing the same game, and just as surely we do not share the same expectations of the CMx2.

-tom w

Tom,

I delight at your courtesy, and wish to be your equal at this, though I feel the limitations of not speaking my native tounge here. Know at least my desire in this.

The focal problem (quality as I see it) is that you are the Borg.

Not you personally, I mean the Player, any player. Is the Absolute Spotter.

Enemies encountered by your units will have to manifest themselves somehow, if they wish to interfere. Say they kill your Jeep. Your jeep can blow up, be gunned down, just disappear - any type of interference which is not caused by you will be noted by you as a hostile presence. The Borg mind is made aware of a hostile presence. Of some kind. The enemy might show on the map, or might not show, might show up in the wrong place, be just an icon, many interesting options are available. Enemy units/icons/nothing that appear or do not appear might and might not be able to be targeted with direct or indirect fire. I imagine all this to be possible to edit.

Either way you will still have seen those enemies, or at least know they are there, somewhere. You will redirect your will, redeploy, move up reserves, aim barrels, advance with caution - all of it against enemies that only a handful of your men, now lying dead in a jeep, have actually seen/experienced/encountered/fallen victim to, before being able to report anything to anybody.

This is the supernatural awareness of which I am a partisan. Impossible for any IRL commander to have, and unrealistic in a sense (but not in all). I am hoping (and now much reassured) it will remain in future CM titles.

Yours Truly

Dandelion

[Mord]

I am really excited about everything I've been reading in the two various threads. One of the things I hoped for and never thought would happen was the 1:1 man ratio. I don't tend to worry about the bigger issues because there are so many guys on here wanting and arguing points for them more eloquently than I could, that I always feel those are well taken care of without any input needed on my part.Though we are discussing Command and Control here realism is also the sub topic so I don't feel to out of place asking a couple questions concerning it.

Steve, not to get you to mired and distracted from the main point, but is the 1:1 man ratio going to be all encompassing, meaning will gun crews, MG teams, and bailed tank crews be represented as well? And will there be an equal on screen representation of casulties or atleast a toggle? And lastly will the terrain be more apt to appear battle ravaged over the course of a scenario, outside of the buildings that is?

Mord.

[hoolaman]

Originally posted by Dandelion:

[...] the Player, any player. Is the Absolute Spotter.

Enemies encountered by your units will have to manifest themselves somehow, if they wish to interfere. Say they kill your Jeep. Your jeep can blow up, be gunned down, just disappear - any type of interference which is not caused by you will be noted by you as a hostile presence. The Borg mind is made aware of a hostile presence. Of some kind. The enemy might show on the map, or might not show, might show up in the wrong place, be just an icon, many interesting options are available. Enemy units/icons/nothing that appear or do not appear might and might not be able to be targeted with direct or indirect fire. I imagine all this to be possible to edit.

Either way you will still have seen those enemies, or at least know they are there, somewhere. You will redirect your will, redeploy, move up reserves, aim barrels, advance with caution - all of it against enemies that only a handful of your men, now lying dead in a jeep, have actually seen/experienced/encountered/fallen victim to, before being able to report anything to anybody.

This is the supernatural awareness of which I am a partisan. Impossible for any IRL commander to have, and unrealistic in a sense (but not in all). I am hoping (and now much reassured) it will remain in future CM titles.

Yours Truly

Dandelion

This is a good point, and must always be kept in mind when we talk about a single player game.

However, we must keep in mind the two separate issues here, which have become known as "God-like view/borg-like swarm" and "Borg spotting".

The god-like view of the battlefield is the issue you are talking about, and I for one also do not wish to see it dimimished in any way from the state it is in in the current CM. I want to be able to see everything my lowliest units can see.

The "borg" problem, as distinct from the "god-like" problem, comes from the game dealing with an enemy spotted by one unit as if it were spotted by all friendly units simulatneously. So if two units are on opposite sides of the map and one unit spots an enemy which both have LOS to, the other unit will react instantly to it. Relative spotting will take care of a lot of issues with realistic spotting and targeting, which in itself alleviates some of the god-like knowledge of the player.

If you are unsure about whether any individual unit has spotted out an enemy that you, the player know about, it will be harder to respond confidently to that threat.

Although I do not wish to restrict the view of the player of his "war-movie", I advocate restricting the troops themselves to following realistic chain of command orders. That way, the player can only play god within the limitations of his simulated commanders.

[John Kettler]

Steve and troops,

This is the best thread we've had in ages. That said, I wasn't trying so much to put together a wish list as I was "thinking aloud" about some potential approaches which might be helpful and therefore proceeded to list some things I've seen done in various places.

If I may hearken back to my military aerospace days at Hughes and Rockwell, we had lots of computer models to run hypothetical air and ground engagements. For example, in developing the AIM-120 Slammer (then AMRAAM), we could model the engagement with the fighter autonomous, under ground controlled intercept control, or under AWACS

control, taking into account onboard sensors (eyeball, assisted eyeball, radar, IR, etc.), tactical data links, command delays, nature and extent of jamming, target type, formation and maneuver state, etc., and would vary these parameters in a systematic, controlled way to determine where the leverage lay in the situation. Measures of effectiveness would be things like time to first intercept opportunity, time to first kill, separation distance at missile away and at impact (important for avoiding counterdetection and counterfire). We could even perform cooperative firing in which the shooter made no emissions at all.

Am describing this in order to show what can be done, and the first place I'd start would be to develop a set of detection criteria and thresholds

for various battlefield events as a function of things like cover, movement status, relative exposure, camouflage, lighting, light and noise discipline (can see the flare of a match from a mile on a clear night), etc. Tractics had a chart which did pretty much this function, in the form of rules controlling spotting. Even a simple version of this would, IMO, really help. Another thing to do would be to fold that back into what the player is allowed to know and when. We already have a kind of spotting progression in place, but suppose we added one in which a cue is provided that something's there/may be there, but only in a pretty broad zone much bigger than our present sound contact (too big to area fire). This would have the twin useful effects of adding to the in-game tension and frustrating the present almost instantaneous spotting. One could also delay posting any kind of contact or detection graphic for a turn or two (maybe more if only crummy communication from cut wires, broken radio, etc.), while perhaps letting the player try to tease out what's going on with what the player's ears can hear only (with deliberately ambiguous directional cues as a function of immediate acoustic environment for the unit potentially able to detect and while working through the background din). The exception to this would be using flares or other means to trigger preplanned events such as defensive artillery concentrations, lift or advance artillery fire, call down fire on previously designated target, etc. Even a one turn delay in posting the contact would do a great deal to enhance the effectiveness of things like antitank guns and SP AT weapons. As it is, SPATs like the Marder, Hornisse, and various Allied cousins can't work as they should because they're generally found the moment they fire and can't take the counterfire, being so lightly armored. The combat reports tell a different tale.

Umpired wargames have been around for centuries, but now we have a computer which can perform the role. Why not have Clausewitzian "friction" as a selectable option and make it variable depending on what's being ordered? If the Army's own tests found that the average FO (arguably a far better map reader than the average 2nd looey) in the preTACFIRE period had a 300m self-localization error, where is it written that our maps have to be GPS accurate? Why not insert chances for things like random delays, misunderstood orders, increasing unit inertia as casualties are taken and fatigue sets in? Again, Army tests found that the GIs in a line rifle platoon were still 80% combat effective after 24 hrs. w/o sleep, but their officer was only 20% effective at decision making and exercising effective command. See where I'm going with this? If we're talking multiplayer, then much of this would seem to be relatively easy to do. CPXs have been played for a long time which relied on written messages and radio to pass vital info up and down the chain, and net modeling would become a real player in the outcome. Imagine trying to fight a battle with 75% of your radios not working or otherwise lost. This happened to one Russian tank army in the counteroffensive after Kursk. Think what could happen should the one transceiver in the tank company suddenly fail.

I may be wrong about this, but I believe there are ways to get there from here without boiling our beloved programmer in his own brain fluid.

Regards,

John Kettler