Jump to content

1:1 Representation Issues Vol. 1 INFANTRY TRAINING


Recommended Posts

Reading the trench thread led to some very interesting issues that were hinted at towards page 3 or 4 of the "Left Turn into the Uncanny Valley" thread.

What exactly do we mean when we talk about "training", as far as the depiction of same in CM:SF, and what does this mean in 1:1 representation? Some people seem to be confusing training, morale and fitness which are all interrelated to be sure, but they are not the same thing, and the subtle differences in the state of all three are a cornerstone of the infantry modelling in Combat Mission. I'd love to see a discussion of how this is going to evolve in the new engine and how the new 1:1 system will portray it.

Training

Training refers not just to the method of education by which civilians are converted into soldiers, but can also refer to the state of knowledge that those soldiers possess once converted. Training is the method by which soldiers are shown how to do things, from brushing their teeth (I remember fondly Warrant Officer Waterhouse instructing us all on this in our armoury basement back in 1988) to eviscerating someone with a bayonet.

Training includes Individual Training (generally Basic Training which includes general knowledge stuff such as rank structure, military law, organization, how to wear the uniform, military courtesy, and basic weapons handling and fieldcraft, and Trades Training, which teaches a soldier how to perform a specific task in a field unit, be it infantryman, artilleryman, driver, etc., and leadership training, which qualifies a soldier to hold command ranks from corporal to colonel) and Collective Training (which is where you perform as a unit to carry out combat-related missions; in the infantry this might be a patrol (recce patrol, fighting patrol, contact patrol), relief in place, deliberate assault, hasty ambush, passage of obstacles, etc.).

Morale

Morale includes several key factors but can basically be described, for want of a better term, as happiness, or satisfaction. How motivated are your soldiers? Do they want to be in the military? Considerations that have an effect would be whether or not they are draftees or volunteers, whether the war is seen as righteous, how long the soldiers have been in the military, how many days in combat they have seen (the longer, the more fatigued they are, but also the more experience - for the ones that live), how good the self-perception of the unit is (a Republican Guard or Airborne unit will generally have higher morale than perhaps a Militia or National Guard unit), and even the state of training will have an effect - a well trained unit will feel self-confident. Some of these are double-edged swords; a unit that has won recent victories may have high morale because it is confident, or it may be resentful at being volunteered too often for hazardous assignments.

Fitness

Fitness in CM refers to physical fitness - what kind of raw material is the army receiving; starving farmers from a Depression, fat bankers not used to hardship? For units in the field a long time, have they been eating well? The troops in Stalingrad or at Bataan were on starvation rations and their ability to exert themselves was severely curtailed. What effect should this have in the game?

1:1 Representation

So how should all this manifest itself with the new 1:1 rep?

One of the first things a soldier learns how to do is fire his nation's service rifle. Firepower factors in CM would naturally reflect cyclic rate of fire, calibre and other technical details of rifle and LMG as found in the infantry section/squad.

What about SLA Marshall's assertion that infantrymen often didn't do much in a firefight? he revised his estimates by Korea, and SLAM was slammed in recent years for his faulty methodology. Other writes such as Galloway (and Grossman?) have echoed some of his writings, certainly with respect for Second World War infantry.

We see that infantry in CM:SF get Rattled, Panicked etc. Is it too frequent? Not frequent enough? Are the squads too homogenous with regards to this treatment? Or has BF.C gotten this right?

What kind of marksmanship do the individual riflemen have? Should this matter, or has this been abstracted in CM? Should the player know, or care, how well PFC Jones can shoot? I suppose if it was possible to detach PFC Jones as a scout, it might matter, but given our inability to do this, it probably does not.

What other aspects of training should be simulated? What about familiarity with captured weapons? It has been suggested that soldiers should be able to pick up dropped weapons, either enemy or friendly. Personally, I wouldn't pick up an enemy rifle that may very well be booby trapped and most certainly had not been zeroed, but should this be included in a 1:1 rep?

What about deeper issues? Stacking up to clear a house - the U.S. Army has certain drills. Should we be able to design our own drills, and implement them? Not in-game, but perhaps before the game begins, by pre-programming certain squad routines or formations we want them to use during play of a scenario? Say we want them to ASSAULT by always using an inverted-V formation.

Or would it be more realistic to have some squads unable to perform these drills to perfection? Particularly, say, a squad that has recently absorbed new replacements, and doesn't know his place in the stack. Would this be too much realism and simply detract from the game?

Another issue that Andreas reminds us of in the trench thread is that of command delays - a device used in CMX1 to simulate training states. So what exactly do we do in CMX2 to represent the differences between "good training" and "bad training" aside from firepower factors/marksmanship and perhaps poorer pathing representing a conscript who hasn't rehearsed battle drills.

Are the differences in "training" really that tangible at the battlefield level?

Or would they be more morale related - i.e. the poorer trained soldiers would actually be more likely to do nothing than the wrong thing?

These are just some issues; the 1:1 rep opens a lot of cans of worms - what are some others?

[ August 14, 2007, 08:34 AM: Message edited by: Michael Dorosh ]

Link to comment
Share on other sites

Interesting post. If 1:1 is the way to go to achieve greater complexity and this problematic phrase 'realism', then as well as the above I'd have thought that factors such as general intelligence and temprament would have to be modelled to a fairly complex degree for individuals. At the most basic level it would have to impact oh how a soldier recated under certain conditons and affected his ability to carry out, or give orders succesfully

Link to comment
Share on other sites

I can add that modelling of Collective Training is basically what the player brings to the game - though stuff like command delays can certainly be used to model stuff like unit cohesion. In multi-multi-player, you can have actual collective training simply by getting together and doing real rehearsals and battle drills. As it stands with one player commanding a force (be it against the AI or another human), the in-game variances in training methods seem to be sparse, but I wonder if by necessity this isn't so. One can see the need for lesser variety between two first world nations, as we had in say, Normandy in 1944 (though more anon) as opposed to the US Army vs. Syria.

But looking at "training" in the German Army in 1944, simply because I am most familiar with that era, we see that basic training fell from I believe 16 weeks to less than 8 by that time, and that advanced (trade) training was also restricted. The question becomes, then, how do you reflect this in CM terms? Simple firepower and morale hits? This works for the 1944 German, but how then do you model a unit which was highly motivated but incompletely trained? Especially if firepower is no longer just a sum total of the squad's weapons, but individually modelled?

Doesn't firepower involve not just the rifle calibre but the willingness of the operator to use it? And moreover, the ability to maneuver himself into a firing position? This brings us back to the basic "battle drills". In the Commonwealth, the action on taking effective enemy fire was DOWN CRAWL OBSERVE FIRE. An untrained troop would be inclined to carry out this drill at the first sight of muzzle flashes, being unable to distinguish "effective" fire from "ineffective" fire. I think John D Salt has had some amusing things to say about the difference between the two.

I have no idea what the contact drills for the Syrians in 2007 or the Russians in 1941 were, if indeed, the latter had any at all. So how do you model a rabble in 1:1?

Link to comment
Share on other sites

All of the thing you mentioned can be programmed into an AI, be parametrised by a certain set of skills (not necessarily independent), be parametrised by player commands issued runtime etc.

It's a lot of work, and I don't have a clue where are the magic boundary required by the 1:1 representation to offer a good playing experience.

Human intelligence can be never achieved, or even approximated, so your troops probably never employ feints in close combat or use the latest karate kicks. The point is to get a good playing experience in spite of visible 'abstractions', shortcuts, oddities from your pixel individuals.

They are just tanks in different skins and with a different tacai, maybe we need more time to accept this.

To get an understanding what can an "1:1 AI" perform check out an FPS game, I think the Battlefield 2 demo will be good for that.

Link to comment
Share on other sites

In my opinion training is introduced primarily as a method of a) countering instinct and B) instilling effective principles of moment.

There is, however, almost always a synergy between these two concepts. Training takes and bends instinct toward best principles. Without training -- or experience, which may teach different lessons -- a person will likely react with a basic instinct that is remarkably similar to the trained instinct, just less effective.

In my opinion rabble should be modeled as being able to perform the same basic computations as trained soldiers, albeit more slowly and with a marked amount of randomness.

To use your example, my first instinct on hearing gunfire would be to get the hell out of the way and try to figure out who was shooting from where so that I could avoid getting shot.

A trained soldier would be able to perform the task without thinking and at a rapid pace. I, on the other hand, would need more time to effectively figure out and complete the task, but our tasks would be practically the same.

Also, while the soldier would likely be able to pinpoint a) the best nearby cover and B) the exact location of the shooter, I would most likely be incapable of doing so. I would probably randomly pick nearby cover and assume the shooter was in a location indicated by the sound of the gunfire.

The point being -- use the same calculations for untrained and trained people, but introduce a larger amount of randomness and variability in the untrained person's, and have them "think" more slowly. This should do an effective job of modelling the lack of training while still allowing for basic human instinct.

Link to comment
Share on other sites

Originally posted by Kineas:

All of the thing you mentioned can be programmed into an AI, be parametrised by a certain set of skills (not necessarily independent), be parametrised by player commands issued runtime etc.

I asked very specifically if training is better represented by specific drills and the way the troops carry them out, or better by morale considerations, i.e. how quickly they panic and do nothing at all (or even run away/disintegrate) butyou haven't addressed this question in favour of speaking in generalities. I'm not sure if that is, perhaps, my answer or not.

For example, should an "elite" squad ever "disintegrate" (i.e. have individuals rendered hors de combat without being actually physically wounded? Is this a training issue at all? For that matter, what does "Elite" really mean - I think in CM it has always meant a combination of training and morale, but do the lines need to be less blurry in the new engine?

I'm thinking very specifically of what a 1:1 Russian infantry squad is going to look like in 1941 when we get to that point. Assume rudimentary training and low morale. Assume the player orders them to attack a building. We see how the Stryker brigade does it in CM:SF.

Given our 1:1 representation, what should we expect to see happen when a squad of Frontoviki are told to storm a house? And for that matter, what would we have seen in real life? I'm hoping it looks radically different from Fallujah, as I'm sure it must have been in the real thing. We may yet see some national characteristics come into CM yet.

Link to comment
Share on other sites

But Phillip, it goes deeper than that. How close is the enemy? If you're being ambushed, you immediately go into your ambush drill, which is much different than a simple contact drill.

And once you're on the ground, you don't just start shooting back blindly until the firefight is over; at some point you listen to your squad leader who is assessing the situation along with the platoon commander, and somewhere down the line someone is telling your fireteam to try a left flanking or maybe throwing smoke and pulling you back to that gully 50 yards behind you, or whatever - but the training is more than just a simple reaction to fire, it is everything that happens after that, from threat assessment to neutralization of the threat/avoidance of the threat. The rabble doesn't do that - doesn't know how to do that, wouldn't even think to do any of that, and why would they? It is alien to them, just the same as walking off the street into the offensive line of the Dallas Cowboys and hearing the play called and not having a clue what he's saying and just standing there as the ball is snapped. You have a sense of what is about to take place, but to get from A to B is somewhat murky...

Link to comment
Share on other sites

As for the effects of command, I would consider that a separate issue. I assume that having a place in a wider command structure confers benefits of additional intelligence and organization in determining responses to situations. Would an individual soldier perform the same drills if not with his squad-mates?

A good example would be the fighting in Mogadishu. Well-trained Rangers performed a variety of similar but practically divergent actions while outside of their structured chain of command. They fought to survive, which is precisely what a rabble will do, albeit more quickly, precisely, and in a more deadly manner.

I do tend to be a structuralist when it comes to AI (I'm a particular advocate of blackboard systems, which I wrote an article about for GameDev a few years ago), so I may be the wrong person to ask if we're going to get into a discussion of high-level functions outside of a command chain.

Good thread.

Link to comment
Share on other sites

What a great post! Between this and "Left Turn into the Uncanny Valley", Dorosh is doing some heavy lifting in terms of bringing thoughtful, unique, and interesting discussions to the board.

Reading this made me think of the simple ways that ASL attempted to impart the "flavor" of various nationalities: Americans with a morale of 6, but 8 when broken so they were easier to break and easier to rally compared to German squads with a morale of 7 in both good order and broken states; British troops' immunity from "cowering"; Finnish self-rally (!); etc... As crude as these rules (necessarily) were, I think they did a good job influencing the "story" of the engagements without dominating it. As Dorosh points out, should we worry about training the virtual troops or training the players?

Assuming identical morale, armament, and fitness levels, would we expect squads from different training backgrounds (nationalities) to perform the same types of actions differently? Surely a conscript squad with minimal training would perform differently under a range of combat conditions than, say, Delta Force operators. This seems intuitive, again, even assuming equal morale, armament, and fitness. What about more subtle cases such as average first line British squads compared to average first line German squads in 1944?

It seems the 1 to 1 representation is an awkward spot. If the soldiers were controlled individually, well, exactly how a squad crossed the street would be up to us. WE would control the strings of the battle drill puppet. In CMX1, with abstract squads, any differences were limited to the various ratings, armaments, and our imaginations. However, with 1 to 1 representation and squad level control we might expect to see different battle drills (as distinct from ratings and armaments) played out before our eyes.

Link to comment
Share on other sites

Originally posted by Michael Dorosh:

I asked very specifically if training is better represented by specific drills and the way the troops carry them out, or better by morale considerations, i.e. how quickly they panic and do nothing at all (or even run away/disintegrate) butyou haven't addressed this question in favour of speaking in generalities. I'm not sure if that is, perhaps, my answer or not.

When I wrote my answer your 2nd post wasn't available yet. Apologies in advance if I miss the point again.

My answer is that it looks hopeless to compact a whole "training" into a single quantity (firepower). If I had to design a similar system, I'd use skills (maybe a dozen) and a rule based system to capture the training, probably this is what you mean by "drills".

In that way you could capture simle rules (down-crawl-observe-fire etc.) You don't even need difficult rules, because the drills have to be simple even for humans, for well known reasons.

If you want to simulate a training within the degrees of freedom provided by the CM:SF engine, then I don't really have tips.

Link to comment
Share on other sites

Keep in mind one of the insights we had about the mechancis of the game in your other thread.

1:1 rep. does NOT have to be equal to 1:1 computation. What do I mean by this? The same way we had abstractions for this sort of things in CMx1, there will be some undergoing abstractions in CMx2. A lot of these things may not be yet included in the engine, abstracted or not, but eventually we will see them evolve and will be included either in the form of abstractions - via somekind of bonuses, penalizations or whatever - (IMO, the least, the better) or as calculations.

Ie. the same now type of terrain does not affect much cover and concealment (trees or bushes are not much different than open ground), at some point it will be improved and will make a difference (if I'm wrong on this, I mean, if it really makes a difference allready, someone point out, but my conclussion so far is that it does not to a great degree), will it be adding somekind of undergoing rules/bonuses/abstractions as CMx1 or will the physical objects do a difference? I can't answer, but at soem point it will do a difference (if we want to play in other set up that is not urban/desert based to work out at least haha).

Don't want to knock out the discussion though, keep it going.

Link to comment
Share on other sites

Originally posted by Migo441:

What a great post! Between this and "Left Turn into the Uncanny Valley", Dorosh is doing some heavy lifting in terms of bringing thoughtful, unique, and interesting discussions to the board.

Reading this made me think of the simple ways that ASL attempted to impart the "flavor" of various nationalities: Americans with a morale of 6, but 8 when broken so they were easier to break and easier to rally compared to German squads with a morale of 7 in both good order and broken states; British troops' immunity from "cowering"; Finnish self-rally (!); etc... As crude as these rules (necessarily) were, I think they did a good job influencing the "story" of the engagements without dominating it. As Dorosh points out, should we worry about training the virtual troops or training the players?

Assuming identical morale, armament, and fitness levels, would we expect squads from different training backgrounds (nationalities) to perform the same types of actions differently? Surely a conscript squad with minimal training would perform differently under a range of combat conditions than, say, Delta Force operators. This seems intuitive, again, even assuming equal morale, armament, and fitness. What about more subtle cases such as average first line British squads compared to average first line German squads in 1944?

It seems the 1 to 1 representation is an awkward spot. If the soldiers were controlled individually, well, exactly how a squad crossed the street would be up to us. WE would control the strings of the battle drill puppet. In CMX1, with abstract squads, any differences were limited to the various ratings, armaments, and our imaginations. However, with 1 to 1 representation and squad level control we might expect to see different battle drills (as distinct from ratings and armaments) played out before our eyes.

Migo441 - yes, these are exactly the kinds of issues I am wondering about.
Link to comment
Share on other sites

Michael Dorosh said

DOWN CRAWL OBSERVE FIRE
I think the Canadian reaction to effective enemy fire IA does not apply to the entire commonwealth as ours is slightly different - with a quick dash to change location so the enemy will be less likely to have a fix on you and a sights check to make sure you have the right range.

In Britain it is:-

dash

down

crawl

sights

observe

fire

Cohesion

Is it worth considering cohesion? Poor cohesion after conducting an assault for example can disrupt well trained, motivated and experienced soldiers and piss poor ones alike.

Well trained troops recover more quickly, but without cohesion being modelled you end up with super soldiers who can perform tasks that are impossible in the real world. For example, you may well have an excellent platoon commander, and fit, experienced soldiers with great morale, but get them to fight through and clear a trench system for 50 meters and see how much confusion, disruption and lack of control there is upon completion of the task. They’re still fit, still well led, and still got high morale - but try coordinating them to attack a second strong point immediately and you’ll find in reality the lack of cohesion makes this quite difficult - even for seasoned troops.

Cohesion is distinct from other qualities as it is a temporary effect brought on by the rigours of battle and combines many of the factors already modelled plus more obscure factors that you cannot really lump together other than under cohesion - e.g. a member of a fireteam loses his comrades temporarily in a building. Is that morale, training, fitness, or command? Probably a bit of all plus the X-factor Clausewitz calls ‘friction’ that means you need another model in place - cohesion.

Lack of cohesion is caused by the collective effects of tactical activity - partially tiredness, partially command, control, communications and intelligence (c3i), partially morale and partially administrative-organisational, partially experience related, and partially due to fitness. To this mix you must add luck/chance/the fog of war/friction.

If you model this omnipresent factor that affects combat outcomes you start to see activities in the game that model real life such as the regroup/reorg following any assault to restore cohesion. Without doing this a unit should suffer from inertia and a weakened command rating, being less aware, less responsive and less capable.

In the CMSF MOUT environment modelling cohesion becomes important. Once a platoon starts clearing houses, unless they regroup/reorganise periodically it can go to rat sh!t pretty quickly.

Link to comment
Share on other sites

  • 2 weeks later...

I added this to another discussion on special forces, but it seems germaine here.

They were discussing SF:

quote:Originally posted by Martyr:

</font><blockquote>quote:</font><hr /> quote:Originally posted by flamingknives:

How would one model specops in a tactical wargame? People keep on calling for it but I don't see how it can realistically be accomplished.

Nothing too elaborate; we already see some of this in the ability of Syrian spies and unconventionals to evade sighting, for instance. I can imagine western SpecOps units that receive similar stealth characteristics--not because they blend into the neighborhood, but because of, well, stealth. This would add some spice to the usual run of missions.

It's not nearly as important as adding to the broader base of vehicles and national armies, however. </font>

Link to comment
Share on other sites

Reading this made me think of the simple ways that ASL attempted to impart the "flavor" of various nationalities: Americans with a morale of 6, but 8 when broken so they were easier to break and easier to rally compared to German squads with a morale of 7 in both good order and broken states; British troops' immunity from "cowering"; Finnish self-rally (!); etc... As crude as these rules (necessarily) were, I think they did a good job influencing the "story" of the engagements without dominating it.
-Just wanted to throw in that ASL also layers an Experience Level Rating (ELR) on top of a unit's base morale. This allows the scenario designer to depict troops that may "officially" elite, 1st line, etc., but who may actually have more or less functional experience and/or unit cohesion when the bullets start flying. So, a squad that is trained to "elite" status, but with little or no real combat experience can be depicted, as well as a squad that may not be "elite," but that is comprised of grizzled veterans who will hold up under fire very well. This abstract system functionally depicts whatever subtle shadings of morale/force cohesion might be needed for a scenario and is a beautiful system IMO.

The design for effect here really seems to work well. So, for example (using our imaginations, the game doesn't go into this kind of detail), in late war, a German squad may have a good inherent NCO who keeps the green horns in line - thus presenting a 1st line squad on its surface. However, if that NCO is incapacitated, the squad may fall to pieces rather quickly. This is is represented by the ELR. An early/mid war force of 1st line Germans, for example, may have a high ELR and thus not lose much unit cohesion over the course of a battle as they break and rally during the fight. However, a late war 1st line force (or a force from a different period that has little combat experience) with a low ELR may very quickly begin to drop in effectiveness/cohesion. This is represented in ASL by a drop in unit class (elite to 1st line, 1st line to 2nd line, 2nd line to conscript, etc.).

CMx1 may have used the ASL system as a jumping off point for the design of its morale system. I'm not sure how it is handled in CMx2. For the first time, my rig is too long in the tooth for CM! No luck running the CMSF demo.

Anyhoo. Interesting discussion (as was the Valley Thread).

Thanks, Michael!

Macisle

Link to comment
Share on other sites

Originally posted by Michael Dorosh:

The use of single sentries, at the very least, would be a convenient way to trigger action in a scenario without having to deploy entire squads to outpost stuff in order to ensure the game begins where/when the designer wants it to.

Nicely bumped on your own thread ;) .

I wonder about things like small patrols, two man scout teams, single sentries and the like too.

Especially for WW2.

I am sure CM will never be a commando simulator, and I think there was a thread about this a long time ago.

But it does beg the question in a game that now simulates individuals whether things like capturing a prisoner or a clandestine raid can or should be simulated.

I agree that there is a fine line between a spy with an IED trigger and a Commando setting demo charges on a fuel dump.

But I am sure BFC won't touch this with a ten foot barge pole so I guess it is all academic.

Link to comment
Share on other sites

If I may,

1. Soldier do not do what they are Trained to do, once it seems irrelevant. Soldiers do not Dasch Down Crawl Observe Fire, for example, in that order if at all. They do what they think works

2. We should be able to control squads, fireteams, and/or 3-5 man groups by gving them simple orders related to 4 basic actions - Fire, move, observe, communicate.

3. DO NOT confuse what is written in training manuals with what the actual application of actions in combat. IMO, Stacking is utter BS. - THE GAME should be able to allow the player to carry out the four basic functions = Fire, move, observe and communicate, - without being handicapped by poorly written training manuals of all the utter BS SLA Marshall came out with, - lying stupid old SOB!

4. CMAK actually shows why a lot of WW2 training manuals were BS. - and why I love it.

The most important things to get right are the modelling of the human will to fight, and the equipment with which they fight. DO NOT MODEL training!!

Link to comment
Share on other sites

As anyone who has ever done it or trained for it will tell you, MOUT is F*ing hard. Once you're inside those buildings, just clearing them is a chore. Once the bullets start flying(real, or simunition), it becames downright insane. It's not easy, at all. It's chaotic. It's bloody. And I'm sure it would be murderous to code with absolute realism.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Unfortunately, your content contains terms that we do not allow. Please edit your content to remove the highlighted words below.
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...