Strange/Funny Algorithm in CMX2

[Nemesis]

I have noticed that CMX2 will often kill machine gunners (MG42, BAR, Bren) in a squad before it kills other men. There is very clearly an algorithm here.

This is realistic when it comes to small arms fire. The enemy would target machine gunners because they have the biggest, baddest weapons in the squad and they would clearly stand out on the battlefield. I can totally see why Battlefront put it in.

What is funny though is that this extends to all types of combat.

So:

- If someone in a squad steps on a mine, it is most likely the machine gunner.

- If a squad gets hit by a mortar round / artillery shell, the machine gunner is the most likely to get hit.

-If a squad in a halftrack gets hit by an AT shell, the machine gunner is the most likely to perish.

CMX2 machine gunners are the equivalent of the Star Trek Red Shirted Ensigns!

 

[Bulletpoint]

I think it's just a combination of confirmation bias and the fact that machinegunners are actively firing for longer.

Since troops light up when firing, they also get more fire in return.

 

[A Canadian Cat]

Yep, @Bulletpoint is spot on. The more common complaint is that BFC purposefully kills squad leaders over other troops. Neither of these is true at all.

 

[MeatEtr]

I have noticed that CMX2 will often kill machine gunners (MG42, BAR, Bren) in a squad before it kills other men. There is very clearly an algorithm here.

This is realistic when it comes to small arms fire. The enemy would target machine gunners because they have the biggest, baddest weapons in the squad and they would clearly stand out on the battlefield. I can totally see why Battlefront put it in.

What is funny though is that this extends to all types of combat.

So:

- If someone in a squad steps on a mine, it is most likely the machine gunner.

- If a squad gets hit by a mortar round / artillery shell, the machine gunner is the most likely to get hit.

-If a squad in a halftrack gets hit by an AT shell, the machine gunner is the most likely to perish.

CMX2 machine gunners are the equivalent of the Star Trek Red Shirted Ensigns!

Yeah pretty sure this is incorrect. Sounds like you have had a run of bad luck lately. However, if my fuzzy memory serves, the AI does indeed get a slight bias (small percentage) that leaders don’t get killed. This is only for the AI though, not for players. Think this was even confirmed by BF/Steve.

 

[Hardradi]

Yep, @Bulletpoint is spot on. The more common complaint is that BFC purposefully kills squad leaders over other troops. Neither of these is true at all.

With respect CC, this is not my experience. My "leaders" appear to become casualties at an alarmingly high rate to the point where they must be subject to a target algorithm. MG gunners are the same. I have at least 100 games in the bag over the last three years.

I agree with Nemesis. If is is correct it would be nice if it wasn't so prevalent with indirect fire. Also, game wise I do not have a big problem with it because both MG gunners of all types were/would have been targeted by direct fire.

 

[Bulletpoint]

I definitely felt the same many times. In a recent game, I had lots of infantry in a couple of big buildings. A few 81mm mortars came in, and two guys went down. Both were squad leaders. One was even killed at very long range from the point of impact, and he was facing the other direction so he had the entire building between himself and the explosion.

But I still believe it's just selective memory. I remember that situation because it was so annoying. I forgot hundreds of similar situations where there were no casualties or where it was a guy with a rifle.

 

[Hardradi]

It could be Bulletpoint but as you watch Arty rain down from a distance and the markers start flashing, you just think there goes the Leader or Gunner. I try to keep a track of casualties in battles so I might start keeping a track of this.

With the Leaders I might start watching where they move compared to the rest of the squad. I have a distinct memory of one running into a building first and getting shot down straight away. I use split squads quite a lot, maybe I am assaulting too much with the SMG team and its increasing the casualty rate.

 

[Bulletpoint]

I think there might also be something about the MG-gunner being a bit slower, so when moving, he'll be the last to reach cover. It takes some time for the enemy to spot and get the range, so it could be that the last man to cross the street is in more danger. Of course that wouldn't explain the artillery bias, if there is one.

 

[Bulletpoint]

Here's another theory: If we assume leaders generally have better stats (experience, leadership, motivation..) than other troops, then that would make them more likely to stay up for a longer time than to take cover. That might make them statisticaly more vulnerable.

 

[Hardradi]

Putting my money where my mouth is I have done my first set of stats from a battle. These are the casualty statistics from the first battle from the Scottish Corridor campaign. No distinction between deaths and casualties. Minor injuries not counted. I have forgotten how to do statistical analysis, deviations and such. So I have kept it pretty simple which is usually better and subject to less manipulation.

Only C Company was heavily engaged in combat in the battle. The feeling I get when I look at my forces is that the C Company Leadership has been devastated with a total of 9 casualties across Commanders, Leaders and Assistants. When you look at it percentage wise its about what it should be, there are no significant deviations. The Gunners fared very well in this battle with casualties on par. Antitank soldier casualties stand out as the highest.

The results above are from conclusive but I feel it does point to the influence of "selective memory" or selective view of the casualty results. I have another battle in progress. Its a lot bigger with lots of indirect enemy fire so it should provide a better sample size.

 

[Bulletpoint]

Putting my money where my mouth is I have done my first set of stats from a battle.

Well done. I respect people who are able to test their own assumptions and then admit in public that they have updated their view based on the findings.

 

[A Canadian Cat]

Nice work. After a 100 battles you could probably start doing the statistical analysis .

Another thing that should get added for a fuller accounting is position when casulties are caused and how. What I mean by that is if you have a squad advancing I would suspect that the men at the front of the group will more likely be casulties if they come under fire from the front. If you are getting shelled by mortars that are falling behind your advancing squad then the men at the rear would be more likely casualties. And if the SOP of the soldiers (by that I mean the Tac AI of the game) is to put leaders in front when advancing then this will have an effect. Not an inappropriate effect but an effect.

There is a ton to keep in mind when looking at actual data and methodology plus the take away might be something like "should team member X be at the front when advancing" instead of "the game inappropriately targets team member X". It is just one of the things that doing this type of analysis leads to - learning that your first question or firs hypothesis is not correct but something else is going on.

Side note fun bug story: In the early days some people were asking about bogging with the concern that the amount of bogging and immobilization was too high in the game. Some people, including me, did some testing of various ground type and a few vehicles. In the end I do think that there were tweaks made (long time ago and I do remember that part) but what happened was I had included a road in my test terrain and discovered that vehicles were getting bogged at the road / underlying tile interface much more likely than in the open ground of the same type. That may be OK given ditches etc but it was way more frequent so there was a problem. Also given the pathing issues with roads driving down roads over questionable terrain could turn out to be problematic too. So, in the end we found a different problem than what we were originally investigating.

 

[mcmortison]

How do u generate such a statistic in CM? I didnt know you can do this so detailed. Is there a way to Show detailed damage to a vehicle?

 

[A Canadian Cat]

How do u generate such a statistic in CM? I didnt know you can do this so detailed. Is there a way to Show detailed damage to a vehicle?

I'm not sure what you are asking specifically. One way to gather the information that @Hardradi has is to watch as each person falls. Or if no buddy aid is done you can look at the end of the battle. Perhaps even whith buddy aid you can still tell a squad is missing its leader or assistant leader. As for vehicles there is a system status pane that shows you the condition of things in general terms like wheels, engine, optics etc.

Did that cover it?

 

[mcmortison]

Yeah pretty much. I was hoping for some sort of tool that extracts the exact casualties out of the final save file.

I am a sucker for this kind of stuff. I loved to play the two tank games from gravity team and at the end of the game look over the battlefield inspecting the damage on the guns and tanks (there were small arrows indicating vector and penetration)

 

[Hardradi]

Well done. I respect people who are able to test their own assumptions and then admit in public that they have updated their view based on the findings.

Thanks. I only seek the truth.

Nice work. After a 100 battles you could probably start doing the statistical analysis .

Another thing that should get added for a fuller accounting is position when casulties are caused and how. What I mean by that is if you have a squad advancing I would suspect that the men at the front of the group will more likely be casulties if they come under fire from the front. If you are getting shelled by mortars that are falling behind your advancing squad then the men at the rear would be more likely casualties. And if the SOP of the soldiers (by that I mean the Tac AI of the game) is to put leaders in front when advancing then this will have an effect. Not an inappropriate effect but an effect.

There is a ton to keep in mind when looking at actual data and methodology plus the take away might be something like "should team member X be at the front when advancing" instead of "the game inappropriately targets team member X". It is just one of the things that doing this type of analysis leads to - learning that your first question or firs hypothesis is not correct but something else is going on.

Side note fun bug story: In the early days some people were asking about bogging with the concern that the amount of bogging and immobilization was too high in the game. Some people, including me, did some testing of various ground type and a few vehicles. In the end I do think that there were tweaks made (long time ago and I do remember that part) but what happened was I had included a road in my test terrain and discovered that vehicles were getting bogged at the road / underlying tile interface much more likely than in the open ground of the same type. That may be OK given ditches etc but it was way more frequent so there was a problem. Also given the pathing issues with roads driving down roads over questionable terrain could turn out to be problematic too. So, in the end we found a different problem than what we were originally investigating.

Counting these is pretty mentally intensive so I do not know if I have the energy to break it down by the cause of wound particularly on a big battle. Also, best to keep it simple and let the data talk for itself, to much interpretation and analysis brings bias.

Yes, its a lot like problem solving on any sort of job. Even fixing the toilet, its usually not one issue but two and then the third one you create while fixing the other two.

Yeah pretty much. I was hoping for some sort of tool that extracts the exact casualties out of the final save file.

I am a sucker for this kind of stuff. I loved to play the two tank games from gravity team and at the end of the game look over the battlefield inspecting the damage on the guns and tanks (there were small arrows indicating vector and penetration)

Yeah, we can wish. It was all hard work. From memory, Close Combat used to have a nice campaign feature that showed casualties at the end of the battle and gave out medals,etc.

 

[Ithikial]

Yeah pretty much. I was hoping for some sort of tool that extracts the exact casualties out of the final save file.

Not that I'm aware of. The closest is probably my little post battle recorder but that is manual input from details on the AAR screen but this is not down to the level of detail this thread is talking about. One thing I've picked up recording the details from my battles is that the casualty rates are enourmous and would likely to have been considered massacres and far higher than what would be considered acceptable for a single enagement. I'm on the record for saying I miss the older CM1 era force wide morale factor that helped control this to an extent.

As for the script that prioritises LMG's and leaders over other squad members I can see both sides of the arguement. It would be interesting to learn if the script prioritising these high value individuals within enemy squads is continous or only occurs when the shooter is in an unsurpressd themelves. I can understand a squad in ambush or lucky to get the first spot/shot off taking an extra moment to take out an LMG guy over a rifleman but if the lead if flying in both directions you'd it would degenerate pretty quickly into anyone who is an enemy that you can see. Thankfully I've never been in combat in real life so this is just educated guesswork on my part.

 

[Bulletpoint]

One thing I've picked up recording the details from my battles is that the casualty rates are enourmous and would likely to have been considered massacres and far higher than what would be considered acceptable for a single enagement.

It's a point that is often brought up and discussed. Just for the sake of musing, here's what I would change if I were BF:


Broken troops can still be ordered forward.
Solution: Broken troops can only be given orders if in voice range of at least +1 leader. If not, will automatically stay put and occasionally try to fall back towards friendly map edge.

Troops out of C2 can still be ordered to attack.
Solution: You need C2 to give movement orders longer than, say, 50m.

Too hard to make enemies surrender.
Solution: Lower the general surrender threshold. Puts more emphasis on the importance of leadership and motivation.

Troops bunch up too much.
Solution: Make each team spread out over three squares like squads currently do.

Soldiers often can't use dips and bumps in the terrain to find cover. They either stay in bottom of ditches or sit on the edge in full view.
Solution: Make troops seek "hulldown" positions based on the elevation of the actual square the player selects with the facing command.

Hunt command is inflexible. When troops move forward carefully, they often take extremely long to react to contact, and only get a tiny spotting bonus, if any...
Solution: Boost hunt reaction times and spotting bonus slightly, especially against enemies very close (as in forest terrain).

Peeking around corners rarely works, and gives spotter no real defensive bonus to represent only peeking rather than placing whole body around corner.
Solution: Rework/improve.

No proper houseclearing tactics. No throw grenade option either.
Solution: Special houseclearing option for crack/elite/special forces only. Throws grenades before entering and gives large bonus to survivability when entering enemy-held building, even if enemy hides behind the actual house.

And finally, the biggest reason:
Scenario designers often design death traps. Ordering the player to clear out big, dense forests or towns with little time on the clock is recipe for massive casualties. I've been guilty of this too.
Solution: Design what you think is a well-balanced scenario. Then remove half the enemy troops and double the time on the clock.

 

[Hardradi]

Another battle analysed. This one is a lot bigger than the last one, four companies in all. Its full of action being the Gold Beach scenario from the For King and Country campaign (highly recommend this first battle if you want a D-Day fix). I have just made it off the beaches and started pushing inland so I have not finished the battle yet.

I split out the SMG's in these stats. Once again the results seem to indicate no significant bias towards certain units taking high casualties. Overall leader casualties are 2% (27% casualties versus 25% of the numbers) higher when compared to their total numbers. Individual leader categories provided mixed results. Gunners were favourable. Antitank took higher casualties than their overall numbers (possibly a trend forming).

Results of two tests for hypothesis that certain categories of soldiers are taking higher casualties:

Leaders: 1 for, 1 against.​

Gunners: 1 neutral, 1 against.​

Antitank: 2 for.​

 

[WinOrLose]

And fix trees....