Ratio KIA/WIA still dubious

[Vanir Ausf B]

Playing less aggressive and pushing less hard will reduce your total number of casualties, but it will have little to no affect on the KIA/WIA ratio.

[Fizou]

Playing less aggressive and pushing less hard will reduce your total number of casualties, but it will have little to no affect on the KIA/WIA ratio.

Any tests to back that up VaB

I would think that small arms from longer range etc would result in less dead and more wounds. Also getting caught in barrages in ways that did not happen as frequent in RL might also cause higher dead to wound ratios.

The way a battle unfolds will effect the dead/wound ratio. It can be a set ratio that is not effected by things happening in said battle.

[wadepm]

But advancing more slowly and through cover would allow more wounded to receive medic aid. Would that alter the KIA/WIA ratio?

[Vanir Ausf B]

No tests, just my opinion. I think that whether a casualty is initially WIA or KIA is at least mostly random, with buddy aid being the only real control the player has over the final ratio.

Does anyone consistently get a lot more WIA than KIA? Surely there is some ultra-cautious player out there belly crawling his men around for hours. Erwin?

[Childress]

Adding in the lightly wounded/yellow icons, which Steve acknowledges as an omission, won't alter the ratio much. My personal hunch is that the discrepancy is rooted in inflated morale and motivational factors. And/or the accuracy of small arms fire. But dialing down the former will render the game more of a slog.

[sfhand]

Every time this topic comes up I think of the Japanese in WWII. I don't have any numbers but I am assuming many more died than were wounded based on depictions of them in film (including documentaries). It is my understanding that they did not surrender and they also didn't retreat very often (kind of hard to retreat when you're fighting on an island anyway). I have yet to play against an opponent who surrenders because he has suffered casualties and most maps don't allow for retreat exit zones (which would forfeit the battle). I think we, mostly, fight like the Japanese - to the last man: no retreat, no surrender - when we play this game.

[Freyberg]

In a recent QB, I had a 3:8 -ish ratio of dead to wounded (just over 1/4 of total casualties dead), but I was applying buddy-aid as quickly as I could. It seemed realistic, and had a lower fatality ratio than CMBN, it seemed.

[Michael Emrys]

Playing less aggressive and pushing less hard will reduce your total number of casualties, but it will have little to no affect on the KIA/WIA ratio.

And that's the reason for this discussion. I notice that playing as conservatively as I do, my own casualties are usually very low. The AI opponent on the other hand has usually been nearly wiped out. The thing is, in both cases the proportion of killed to wounded is high, in the AI's case very high. This can probably be explained by a number of factors, including that lightly wounded (yellow) cases are not counted. Also, in my own case, once a team or squad has suffered a number of casualties, I will usually hold it off the front line and use it for other duties such as occupying an already captured objective site, or maintaining a watch over an avenue where the opponent might travel.

Michael

[Apocal]

Part of it is that the AI is, as yet, not good enough to recognize its in an impossible position. Isolate a trench network, cut off any escape by fire, roll up a tank and the occupants will standby for the coup-de-grace.

In the real deal, those guys would usually surrender, except in the case of the Japanese.

[Childress]

We're all Japanese now.

[Kommissar]

Agreed with OP and Vanir Ausf B. Using less aggressive tactics reduces your total casualties, but KIA/WIA ratio is still way too high.

I also agree that troops do not surrender nearly as often as I expect they would IRL. While the surrendering issue may have good gameplay and coding reasons behind it, I don't get why they don't decrease the seemingly random chance that a given wound is fatal? Seems like a very simple fix. Also, it would be nice if they would add more stats in the AAR - total casualties (for those of us too lazy to add WIA and KIA), percentage of troops that were casualties (including preferably a breakdown by role - support, infantry, vehicle crew etc.). This would be especially helpful in campaigns.

On a final note, one thing I do like is how wounded troops can become KIA if subjected to further indirect fire. I would roll my eyes in CMBB when I would obliterate an enemy platoon with very heavy repeated artillery and direct fire such that few would have survived, only to see a KIA/WIA ration of 1/5.

[Childress]

While the surrendering issue may have good gameplay and coding reasons behind it, I don't get why they don't decrease the seemingly random chance that a given wound is fatal? Seems like a very simple fix.

Great minds think alike. At end of battle the game operates on the numbers and spews out plausible stats. The retreating side shows a higher K to W ratio; the losers retrieving fewer of their wounded. The Dead? In an actual war their precise numbers remain unknown for many months. But it's hard to imagine how this workaround integrates with the Buddy Aid routine.

This topic has surfaced in all three of Battlefront's WW2 forums. It resonates.

[Freyberg]

If any sort of change to the KIA/WIA process were being looked at, I'd love to be able to find out how many were killed by artillery, including friendly fire.

[BLSTK]

Excellent suggestion, Freyberg. Especially considering the presence of aerial support and its perils.

[Collingwood]

One way of thinking of the KIA numbers: so many times I read something like "gefreiter schmidt was wounded and evacuated however he later died of his wounds". I agree with other posters about the intensity of our pixel battles - our games are meat grinders. I'm reading the history of a unit on the eastern front at the moment, and if a company lost 2 killed and 7 wounded in a day they felt severely dealt with.

[BletchleyGeek]

I'm reading the history of a unit on the eastern front at the moment, and if a company lost 2 killed and 7 wounded in a day they felt severely dealt with.

That is giving us some leverage to assess to what degree player or AI "aggressiveness" or "casualty tolerance" are not in line with historical records. As Apocal says, scenario designers don't have the tools to influence how the AI reacts to mounting casualties or situations perceived as hopeless (well, with triggers this is not so much true anymore). In the real world, I can't see a Soviet Rifle Coy standing up to the last man while defending from a German attack on a small village on a secondary road somewhere south of Minsk. Most likely, the Coy commander would have pulled back with the expectation of coming back, along the rest of the Regiment and some armoured support. Not every corner of the Soviet Union was as hardly fought for as Stalingrad was.

In the case of human players, I think we're somewhat under the assumption that, whatever the odds, we're expected to be "successful". Actually, I would say that most scenarios featuring "balanced" attacking and defending forces with equivalent firepower are an invitation to reproduce situations where casualty rates are far from what we can find on the usual action or engagement of World War 2, and would have probably raised strong objections (in Western armies) amongst the officers of the attacking force.

Other forms of balancing would include more "sophisticated" victory conditions - following the guidelines given by JonS on his Scenario DAR, or on his little scenario for CMRT - that disallow binary outcomes, or making the maps so that they offered the defender with more "depth" so that falling back and organizing a counterattack with reinforcing units becomes more feasible (and I can see quite a few scenarios in CMRT where care has been taken to offer the defender more "depth"). It's also important in this latter case, that the reinforcements received are substantial enough so as to become a credible threat to a weakened attacking force (one doesn't counterattack a Battalion with a platoon and two Tigers, while such a force may be sufficient to stall any further exploitation by the attacking force).

The "theories" behind attrition that I am most familiar with are the studies by Trevor N. Dupuy in the early 1990s' "Attrition: Forecasting Battle Casualties and Equipment Losses in Modern War", a follow-up to his 1970s "Numbers, Prediction & War". Both works conclusions are to a great extent derived from hard data on the US Army performance in Italy and Northwestern Europe, so one needs to extrapolate that to the Eastern Front with a grain of salt - or two. He gives some quite interesting statistics in "Attrition", that I think are worthwhile sharing with those interested and are relevant to some of the issues being discussed in this thread. I'll keep posting these, as I have more time.

1. KIA vs. Serious WIA looking random

Dupuy collected and analyzed a statistical dataset covering all wars fought by the US from the Mexican War to the Vietnam War. For World War 2, he makes the following observations:

• Battle casualties are divided into three broad categories: killed in action, seriously wounded and lightly wounded.
  
• Killed in action casualties hover - for all the conflicts covered, interestingly - to be about a 20% of the overall casualties.
  
• Seriously wounded casualties for WW2 tend to be a 15% of the overall battle casualties.
  
• For WW2, he gives the figure of 57% as the chances of seriously wounded personnel surviving his wounds
  

In terms of how this relates to Combat Mission, we can see that the proportion of KIA (downed personnel marked with the skull and crossbones) and serious WIA (downed personnel marked with a pulsating red cross) are very similar. Indeed, one could say that the chances of personnel being hit and becoming incapacitated - either by death or severe trauma - are 50-50. Or in other words, it is almost as if the God of War in the was tossing a coin each time a pixeltruppen gets critically hit. So this perception of KIA vs. WIA being almost perfectly random is actually reflecting a "verity" in Dupuy's data.

Regarding the general proportion of wounded personnel to fatal battle casualties: the proportion is more like 4 or 5 to 1 in general. The thing is that, besides lightly wounded personnel not being accounted in the final tally (but probably they're accounted for in the "Force Condition" victory conditions), what the engine lefts out - and is reasonable that it is the case - minor wounds which wouldn't impair soldiers performance (much) but would require further treatment after the action (and hence registered as "wounded in action"), such as grazing bullet wounds. I wouldn't be surprised that a vast majority of the WIA figures we can find published actually consist of such wounds.

[JonS]

I wouldn't be surprised that a vast majority of the WIA figures we can find published actually consist of such [very minor] wounds.

According to the figures you quoted

• Battle casualties are divided into three broad categories: killed in action, seriously wounded and lightly wounded.
  
• Killed in action casualties hover - for all the conflicts covered, interestingly - to be about a 20% of the overall casualties.
  
• Seriously wounded casualties for WW2 tend to be a 15% of the overall battle casualties.
  

100 - 20 (KIA) - 15 (Serious) = 65% of total battle cas in the Lightly Wounded category (or 81% of the wounded battle cas are Lightly Wounded). I'm not keen to guess what proportion of that Lightly Wounded would be considered Very Lightly Wounded - to coin a phrase - but it's easy to make reasonable assumptions that would validate your guess.

That is giving us some leverage to assess to what degree player or AI "aggressiveness" or "casualty tolerance" are not in line with historical records.

Oh, it's reasonably clear to me that total cas within any CM battle are vastly higher than any reasonable WWII-era battle, in absolute numbers and as a %age of the forces involved. But, it's equally clear to me that this is the way we all (well ... almost all) want it. No one (well ...) wants an interesting battle halted mid-flight because some arbitrary cas threshold has been breached, and the tail end of a hard fought battle can be chocka block with loads of interesting little fights.

This came to me while composing the Bois de Baugin AAR, way back when. By any reasonable real-world or wider-world measure either Elvis or I should have broken that battle off much earlier than it ended. However, the last ~20 minutes were full of tension, creativity, interest, and fun, and all those individual little fights were realistic in themselves.

It's only when placed in the context of the larger scenario, and then in the context of the Normandy Campaign and WWII as a whole that the existence of those little interesting battles became unrealistic. But all of them would have gone by the board if an arbitrary hard casualty limit had been in place. I'll take those 'unrealistic' but fun fights over someone else's definition of WWII Was Really Like every time, thankyouverymuch.

After all, I'm playing a game here, not refighting WWII. The component pieces and interactions in CM are very realistic, and that realism only really slips when one attempts to place them in a context that simply doesn't exist in the game.

[Fizou]

Good post BG. Very interesting.

[slysniper]

I think the stats used by most are not what should be considered.

You should use dead to wounded in combat from units which are fighting in the lead units of bloody battles which are not one sided to do a similar comparison to what we see and play within the game.

We are always playing heated battles where the offensive side generally does not have a massive firepower advantage. That right there eliminates about 95% of any real war type actions and how it plays out and how real casualities are adcheived.

So, if you just look at the leading units in some of the major fights you can think off. Then I see nothing wrong with how the game represents it.

Then take into account that in the real world when soilders find themselves in that type of battle, most of the time the side that starts losing immediately retreats, then that is another factor that does not happen in the game.

Famous fire fights are the battles where both sides stick it out and we might read about them in history books , the battles where hero's are made.

But every CM battle is played that way and our troops are asked to do task all the time that would never happen or be followed by real troops.

So the topic comes up again and again and sad to say it will come again in the future

[c3k]

My men know that there will be no coin toss regarding their fate: they all rush forward to eagerly join death's cold embrace. Those that survive are forlorn, and are filled with hope that they may get another opportunity to enter the halls of Valhalla!

Which is another way of saying that our CM battles are always going to be faster and bloodier than the real thing.

[BletchleyGeek]

According to the figures you quoted

100 - 20 (KIA) - 15 (Serious) = 65% of total battle cas in the Lightly Wounded category (or 81% of the wounded battle cas are Lightly Wounded). I'm not keen to guess what proportion of that Lightly Wounded would be considered Very Lightly Wounded - to coin a phrase - but it's easy to make reasonable assumptions that would validate your guess.

I wouldn't be too keen to make a guess on the exact proportions, to be honest. It all depends on what was the method used to record those "numbers", how systematic were those surveys, etc. Probably we could get good data on that from Afghanistan or Iraq, and then try to extrapolate it back, making "educated" guesses along the way. On the other hand, Dupuy also considers such conflicts to be in an entirely different league, statistically speaking.

Oh, it's reasonably clear to me that total cas within any CM battle are vastly higher than any reasonable WWII-era battle, in absolute numbers and as a %age of the forces involved. But, it's equally clear to me that this is the way we all (well ... almost all) want it. No one (well ...) wants an interesting battle halted mid-flight because some arbitrary cas threshold has been breached, and the tail end of a hard fought battle can be chocka block with loads of interesting little fights.

This came to me while composing the Bois de Baugin AAR, way back when. By any reasonable real-world or wider-world measure either Elvis or I should have broken that battle off much earlier than it ended. However, the last ~20 minutes were full of tension, creativity, interest, and fun, and all those individual little fights were realistic in themselves.

It's only when placed in the context of the larger scenario, and then in the context of the Normandy Campaign and WWII as a whole that the existence of those little interesting battles became unrealistic. But all of them would have gone by the board if an arbitrary hard casualty limit had been in place. I'll take those 'unrealistic' but fun fights over someone else's definition of WWII Was Really Like every time, thankyouverymuch.

After all, I'm playing a game here, not refighting WWII. The component pieces and interactions in CM are very realistic, and that realism only really slips when one attempts to place them in a context that simply doesn't exist in the game.

I wasn't really advocating to make "sweeping" adjustments moving things into the league of the "trivial". I reckon that even a battle where the attacker has realistic firepower and numbers ratios, for instance, can end up as being a truly interesting game. Indeed, there's very little context to the battles we play - with the exception of campaigns which are there to provide to some extent that. It's interesting though, by recalling your scenarios I've played, that you state the above and at the same time, you also make quite an effort to provide that context (within the limitations of the game and reasonable assumptions on the level of masochism of players)

The interesting thing Jon, is that I do think that casualty returns for WW2 were far more varied than what published statistical studies seem to suggest that very bloody engagements were far from uncommon: but they didn't either happen every other day. Case in point are Dupuy's own numbers on average casualty rates for units depending on their size. I think that JasonC quoted them directly or indirectly at some point at the threadnaught on the CMFI forums.

So let's look at the "baseline" more closely, which kind of "proves" that something is inherently wrong with CMx2

2. Averages alone aren't good generative/predictive/forecasting models

Dupuy's tabulated average casualty rates per day depending on the unit size as follows:

Size---Strength---Avg. Daily Casualty Rate

Company---200---21% (aprox.)

Battalion---800---9.5%

Brigade/Regt---3,000---2.6%

Division---15,000---1.0%

Corps (3 divisions)---65,000---0.6%

Corps (4 divisions)---90,000---0.4%

Army---250,000---0.3%

Here we can see an "apparent anomaly", and these are Dupuy's words not mine. The dataset considered included 400 WW2 engagements according to Dupuy of "moderate to high intensity when one or both sides was aggressively seeking to accomplish its (or their) mission(s) during a period of a few hours of days". The anomaly is that a given unit is reported to be suffering double or more casualties than its organic unit.

According to Dupuy there's a small influence in this due to presence of staffs and support units in the larger forces. Dupuy justifies this anomaly with the following observations

• Small units will be engaged more intensively, but for briefer periods of time, than will the larger forces they belong to (note casualty rates are measured per day)
  
• As forces become larger, there are increasing delays in the performance of missions and compliance with orders on both sides of interactions between opposing forces.
  

I'll add a third to this: for the same time period, the smaller the force, the larger is the proportion of personnel directly exposed to enemy firepower.

Note that the decrease in the casualty rates goes down with strength, but that also means that the larger the force, the larger will be the space that it occupies and where it maneuvers. For a company one could define to occupy an area of about 400 meters wide and 200 meters deep, and a reasonable area of operations for a company would probably be about 2 square kilometers. What does "reasonable" mean? Well, that depends on the time frame of the scenario: the larger the space, the longer it will take, generally speaking, to march, to bring heavy weapons to bear, to find the enemy, etc. You can't really expect a foot infantry company to accomplish a great deal when time is limited to 2 hours and the area of operations is 16 square kilometers (and it is operating as a unit, maintaining its cohesion).

These are I think, the source of most of the numbers being quoted on these boards to put a lid on CMx2 as "the game where anything goes" with respect to casualties. I reckon that these figures are also very rough: they're hiding from us a quite broad "diversity" in outcomes.

That's my main gripe with Dupuy. He (and the rest of the HERO members) were amazing data collectors and analysts, but very poor statisticians. They just took their data, measured it in many clever and interesting ways, computed an average, and quickly got a predictive model running which was made to fit their data by the mathematical equivalent of "hammering an sphere into a cube-shaped hole". He takes these averages and runs with the ball all the way to the touch down line, forgetting about the inseparable companion of averages: the notion of variance.

In neither of the two books I mentioned there is a straight statement that gives one some leverage on what kind of deviations we can expect from the averages above. In "Attrition" there's something to that effect (that book was written almost 20 years after the more famous "Numbers, Predictions & War"), where a quite interesting list of engagements is given, featuring casualty rates which do not quite look like the averages:

Santa Maria Infante, 12-13 May 1944

E Coy/351st Inf, starting strength: 170, loss: 80, daily casualty rate: 24%

F Coy/351st Inf, start. str.: 170, loss: 170, daily casualty rate: 50%

G Coy/351st Inf, start. str: 170, loss: 84, daily casualty rate: 28%

K Coy/351st Inf, start. str: 170, loss: 102, daily casualty rate: 30%

Rapido, 20th Jan 1944

1st Bn/141st Inf, start: 684, loss: 248, cas rate: 36%

3rd Bn/141st Inf, start: 663, loss: 107, cas rate: 16%

1st Bn/143rd Inf, start: 760, loss: 71, cas rate: 9%

Rapido, 21st Jan 1944

2nd Bn/141st Inf, start: 696, loss: 370, cas rate: 53%

3rd Bn/141st Inf, start: 553, loss: 254, cas rate: 46%

1st Bn/143rd Inf, start: 689, loss: 138, cas rate: 20%

2nd Bn/143rd Inf, start: 673, loss: 249, cas rate: 37%

3rd Bn/143rd Inf, start: 561, loss: 108, cas rate: 19%

I've picked up where we get to see how well fared units of sizes covered by Combat Mission, which were being used in the same "engagement" or "battle" with common or related missions. You'll see that the casualty rates can go anything from double to half of these quite widely quoted statistics collected by Dupuy and his team.

[BletchleyGeek]

Good post BG. Very interesting.

Thanks Fiz

@slysniper and @Ken: I hadn't finished what I wanted to say with the previous post. Indeed, battles will be somewhat bloodier than the average historical engagement, but the thing is that the actual variance regarding casualties can be very, very wide. And also, it doesn't need to be that way.

[Childress]

The choice of the ill-conceived Rapido River battle as an example with its horrendous and atypical US casualties numbers strikes me as a gross departure from the mean, BG. Those responsable should have been lined up against a wall and shot.

[db_zero]

I'm not a stats person so this sort of stuff doesn't freak me out.

As I said elsewhere-as much as we'd like to think of ourselves as great commanders most play like Romans and gladiator matches. 2 opponents enter the ring and get what they want. Lots of action, blood and guts.

Its been said before. Most players are blood thirsty.

[c3k]

It is my men who are bloodthirsty, not me!

I do not see any anomaly in the data set. It follows a very predictable path: as the unit gets bigger (meaning the number of riflemen make up a smaller percentage of the total force) the casualty rate decreases. That makes sense. Not too many officers got injured clustered around the corps hq map table.

A rifle company is about 99% riflemen. (You can dither around the edges with the accounting for radiomen, attached FO's, corpsmen, etc.) Let's just say a rifle company of 250 men is "pure" riflemen. They're all in the front line.

A division (US, WWII, Infantry) had 27 line companies. (Ignoring the weapons companies.) That's 6,750 men in the front line. The division had something on the order of 15,000 men in the TO&E. (Going from memory; forgive my gross approximations.) Of any division's casualties, the great portion (90%+?) will be borne by the 6,750 front line soldiers.

If the companies take 30% losses, (2,025 men), the division only takes 13.5%.

All the above is commonly known. I only post my (gross) example to ensure we're all on the same page.

Another note on casualties: not too many of my men shirk away from fighting. No one sits there waiting for the other guy to go over the top first. They all go. Every time. (If their morale is poor, they'll go eventually.) No one lags back with a suddenly lame leg. There are many survival strategies of which my pixeltroops do not avail themselves. They also never get trenchfoot, etc.

Wound mechanisms: I -think- the game differentiates between rifle and handgun rounds. I do not know if the energy level is taken into account. A 100 yard rifle shot to the torso is different than a 1,000 yard hit. Yeah, both guys are out of the battle, but one is probably dead or dying, the other is just really hurt. (Ignoring heart/aorta, CNS hits.)

I don't think any of my guys get credit for their bibles, multiple layers, etc.

Just tossing ideas into the pot...