Infantry Problem: Running Out In The Open While Under Fire

[Vanir Ausf B]

Originally posted by leakyD:

Mebbe the rewrite can model *variable* behavior for routed troops. Some squads run, some freeze (and take additional casualites due to men "breaking"), some do a little of both.

This could effectively be done right now in the 1.02 patch. In fact, I've been thinking about suggesting something like this for a while now. All that would have to be done is move the threshold of the "move to friendly map edge no matter what" behavior to when units reach routed status instead of panicked. I think this result in considerably more realistic behavior without messing with the basic modeling to any significant degree. I suspect it would be pretty trivial to code.

[ November 30, 2002, 05:51 AM: Message edited by: Vanir Ausf B ]

[Col Deadmarsh]

Steve, are you available to comment on this?

[Suicides-by-Steve]

Originally posted by Colonel_Deadmarsh:

Steve, are you available to comment on this?

I'm getting the problem of the AI crawling towards/perpendicular to enemy fire, despite my orders, and while they are out of cover. This is after the patch. Can anybody tell me why this is?

[Battlefront.com]

Just got back after T-Day traveling.

I am not really sure there is anything to change here. The TacAI is certainly not perfect, nor is the abstraction of unit behavior based on individual soldier "attitudes" (for lack of a better term ). There are certainly problems that can be pointed to with how Panicked units behave under fire, but like almost all TacAI behaviors it is easier to do this than to aknowledge the times when things work just as they should.

There is a behavioral difference between Panicked, Broken, and Routed units. Panicked and Broken units do not specifically attempt to move off the map. Only Routed units purposefully do this. All three units can recover, of course, but Routed units are the most likely to do the worst possible behaviors.

The question about how a unit would react under fire does depend greatly on the specific circumstances. I generally agree with the things John D. Salt stated on the previous page, but with one exception.

Recon by Fire is designed to flush the quarry out into the open by beating on the bush with a stick, using birding terms. In war terms you shoot into a suspected hiding place and see what runs out. Ask yourself... how could Recon by Fire work if the target just hunkered down? Answer... it wouldn't work.

The subtle point here is how much fire and under what circumstances would a unit run from such a situation rather than stay put? This is not an easy question to answer. Some people in this thread, and in previous discussions, think it is.

Again, I am not saying the system is perfect. There are some circumstances that the TacAI is completely unable to take into account because of the way the game is coded. Some of you have seen my previous postings about unit "memory" in the long Sneak thread. This is one of the primary problems.

Another problem is that the TacAI is not able to distinguish between good/bad cover depending upon the specific type of incomming fire. This means a unit in a Trench is effectively thinking no differently than a unit in Open Ground. Of course the unit in the Trench is less likely to move because the cover is much better, but it is unable to modify that behavior for an even greater desire to stay in the Trench during an artillery strike for example.

As others have said, units picking up and running out of good cover happened in real war. Sometimes, often times, because they only knew what was shooting at them and not what could potentially shoot at them. Therefore, the unit made a bad tactical error because it lacked a better understanding of the Big Picture. You the player have no such lack of information and therefore are inherently judging the unit's behavior based on way more information that that unit would really have. At least most of the times I am sure this is the case.

Because even small changes to the TacAI are painfull to implement and test, we are not going to do anything to this behavior. It is also interesting to note that if anything we have toned it down since CMBO. However, the new engine will deliver much more in this area than was possible for us to do when the original engine was written.

Finally, I suspect that some of the people having problems with this behavior are some of the same ones complaining about the TacAI's use of Sneak, advancing under fire, and a few other issues. They are all interrelated, so that comes as no surprise to me.

That's about all I can think to say about this.

Steve

[John D Salt]

Originally posted by Battlefront.com:

[snips]

The question about how a unit would react under fire does depend greatly on the specific circumstances. I generally agree with the things John D. Salt stated on the previous page, but with one exception.

Recon by Fire is designed to flush the quarry out into the open by beating on the bush with a stick, using birding terms. In war terms you shoot into a suspected hiding place and see what runs out. Ask yourself... how could Recon by Fire work if the target just hunkered down? Answer... it wouldn't work.

[snips]

That is rather different from my understanding of how reconaissance by fire is supposed to work. The current US Army doctrine states that reconnaissance by fire is intended to make the enemy disclose his position by firing or moving. I have always imagined that the most likely response it would provoke would be return fire.

http://call.army.mil/products/ctc_bull/96-10/chp3pt2.htm

points out that the effectiveness of the technique depends on how well-trained the enemy are. If they have the discipline to sit tight as the bullets whizz over, then indeed it will be an ineffective technique, at least as far as its reconnaissance function goes.

It is hard for me to draw a clear distinction between reconnaissance by fire and what in WW2 was termed "prophylactic fire". Evidently the intent of the former is to find the enemy and the intent of the latter is to discourage any enemy who might be around from interfering with you, but both of them would look pretty much like shooting up the countryside on mere suspicion. If the target did not often "just hunker down", then prophylactic fire would not work.

Whatever you call it, I cannot recall ever having heard of fire being applied to act as a sort of grouse-beater to beat the "birds" out of cover and on to the guns. "Grouse-beating" tactics are often used -- "hammer and anvil" is a more popular term in the US as it sounds a good deal steelier -- but the "hammer" is as far as I'm aware always a line of troops physically occupying the ground rather than just brassing it up. I have played this sort of game in woods-clearing exercises in the TA, and a similar technique is advised for clearing villages. It needs the "grouse-beaters" to walk through the area to be cleared. Real grouses (grice?) are not daft enough to break cover until the beaters are nearly on them, and I doubt that real people are appreciably dafter.

There are, of course, actual accounts of soldiers running for it. I suspect that a lot of these are withdrawals that were originally supposed to be organised but in the event were not. In CM:BB terms, this would be a spot of bad luck with the "W" command.

I have already mentioned that Hal Moore observed that the Viet Namese he was fighting tended to lie down under HE but get up and run around under napalm or WP. I suggest, then, that it might be reasonable to make the specific effect of an adverse morale result dependent on the event that caused it (the WRG Napoleonics minatures rules show how this idea can use lots of very simple reaction tests to create a very rich variety of reaction behaviour). Broadly, one might say that a failure of courage in the face of bullet or fragmentation fire makes people get their heads down in cover, and in the face of specified "terror weapons" makes them run away, like Bold Sir Robin. What counts as a "terror weapon" probably ought to depend on training, but flame weapons would be especially likely to count (though documents I've seen on flamethrowers point out that they are much more effective against troops not trained to face them), and so I think would the threat of close combat. Tanks might count, especially at night. I remark here that of the two cases of soldiers leaving cover and running mentioned in the account of the Yazov Stariy battle posted by Mike Dorosh, one was apparently after the tanks had been heard but before they had been sighted, and the other was after the man had been wounded (in CM:BB terms simply a reduction of 1 in section strength). The account also specifically mentions that these troops had not been trained to fight tanks. I am sure all the old SL fans remember the "tank terror" rule in "Cross of Iron", too.

So, to conclude -- the behaviour I think I would find most convincing would be for troops panicked by fire (as distinct from a fubbed withdrawal) normally to get their heads down, but if attacked by a nominated terror weapon (flame, bayonets, possibly tanks, Stukas, MRLs) to run away from the threat (no, I don't know how you run away from a dive-bomber, but as Rincewind points out you always run away from not towards ).

It would also be fun to have something like the "chain panic" rule in Advance Tobruk, so that when one section is seen to be legging it to the rear, a lot of other people witnessing the occurrence might feel obliged to join it.

I expect this is mostly or all beyond the scope of any imaginable patch, but these are things to mull over as the new engine starts to gestate.

There are plenty of other things to be said about the modelling of infantry, but I've blethered enough for one day.

All the best,

John.

[Andreas]

Originally posted by John D Salt:

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

What is the difference in overall game/simulation outcome terms between:

1) a unit routing, running away, and being killed in the process

2) a unit routing, cowering, and surrendering to the nearest enemy?

The biggest difference I would expect is, as I mentioned in passing, that you could not clear any but the smallest area of cover just by shooting blindly into it and then chopping down the infantrymen occupying it as they run out (in a known direction, towards their own baseline).

Lots of sensible stuff snipped...</font>

[The_Capt]

I STAND ALONGSIDE MAJ KONG!!!

[kevsharr]

Take the story about a Panther commander[i believe it was Barkman]who came upon a column of parked Sherman's.At the first sight of the Panther the crew's dismounted and ran,leaving seven or eight Sherman's sitting there idling.As far as human nature is concerned anything is possible.

[Battlefront.com]

John,

So, to conclude -- the behaviour I think I would find most convincing would be for troops panicked by fire (as distinct from a fubbed withdrawal) normally to get their heads down, but if attacked by a nominated terror weapon (flame, bayonets, possibly tanks, Stukas, MRLs) to run away from the threat

I see your point and I somewhat agree with it. I think the critical factors here are"

1. Training

2. Willingness to hold ground

3. Perception of enemy intent

4. Perception of enemy fire

5. Perception of overall situation

If a unit has mediocre training, already skittish about current position, perceives the enemy to be capable of taking them out right where they are, are completely unhappy with the volume and/or nature of the incomming fire (i.e. feels staying put is not safe option), and gets the sense that other neighboring units have the same feeling... yeah, I would imagine the unit would break cover and relocate.

While I do not pretend that CMBB is sophisticated enough to know all of these things, I think it does have a pretty good capacity for it. Here are some of the shortcomings:

1. Lack of standing orders for a unit. In CM there is no way for a unit to know "I am in this position through no reason other than luck" and "I was ordered to sit right here and not move until relieved". If I am going out into a field, and get shot at, I have no particular desire to stay in that spot per se. But if I am in a trench and know that my CO is in the trench behind me, I might be more likely to stay put all things being equal.

2. No unit memory. A unit has to pretty much make up its mind about what is happening based on the "now". It can not look back and can not project forward. A real life unit can do this and therefore better evaluate the situation. CM's units basically have to make snap decisions.

3. Absolute Spotting. This means that defending units can be subjected to unimaginable "attention" that would make any sane unit break OR logically determine that staying put is more risky than moving. This makes open ground battles in CM a bit easier than they would be in real life. City and forest battles are less prone to this problem because LOS is generally restricted.

And to disagree with John a bit about how the game works, I just played a brilliant PBEM game as a Green/Conscript Soviet force. Here is what I feel is applicable...

I was defending from good positions against a clearly superior force. My troops generally held their positions to the death (I think some were Fanatics). I had one situation where a platoon in Tall Pines, mostly Conscripts, took out the bulk of a Reg/Vet German platoon AFTER being whacked from the sides by supporting troops and tank fire. One Squad remained pinned in its foxhole for almost the entire battle. All were wiped out to the man where they were, including the remains of a squad having to be cleared out by hand to hand fighting.

In other section of the map I had a platoon (Green/Reg IIRC) defending a small village. It was flattened by heavy artillery. Not a single unit broke, although one Squad was eliminated and reduced to 3 men. Others suffered some casualties. When the enemy assaulted they resisted. They racked up an impressive kill count, including my #1 high scoring Squad of the game (8 men took out 15). They were mostly eliminated only after getting into grenade range. The light mortar tried to flee but was cut down. Staying in place would have certainly resulted in the same thing.

Then, soon after, my Global Morale dipped below 25%. My guys that were still alive tended to pick themselves up and move towards the friendly map edge after taking serious fire. Some stayed and fought to death. The ones that ran survived. A few remained in their fall back positions covering their fleeing comrades even when the game ended.

I guess what I am trying to say here is that the TacAI does have a healthy range of reactions to fire. It has not been my experience that one can just toss about some area fire into a defensive position (a good one) and make the defender just get up and trott off the map. The most likely times for this to happen is when a unit is exposed and is getting hammered from all angles by nasty levels of incomming fire. Sometimes running is smart, sometimes it is stupid. No one thing fits all situations

Steve

[Battlefront.com]

kevsharr,

Quite right. There was one case in front of Aachen where a handful of snipers held up a divisional? level attack because they sniped a bunch of tank and HT commanders. Logic would dictate that this wouldn't be enough to hold up the whole attack, but it was. These guys, who were prepared to go into combat a couple minutes before, no longer did because of a couple of well aimed shots. And these were not Greenhorns here. They were seasoned vets.

War is a strange thing that refuses to be shoehorned. Makes our jobs both interesting and quite difficult

Steve

[kevsharr]

Or take the famous[maybe mythical] incident of the german's tying a rope across a road and no one was willing to risk crossing or cutting it.I believe it was Patton who cut it and told his men to get their a***s in gear

[CK]

I do agree that weird things happen in extreme situations, but troops freezing constantly under fire to places with no cover is bit frustrating.

Soldiers in panic or beyond seem to do more rational decicion than pinned ones in CMBB, they RUN.

Units seeking cover from the street, road or pavement when there is a good shelter within few meters is not realistic at all?

When you are shot at open place you want to run to cover, because you get killed anyway in the open?

When you are in at least somekind of cover you don't want to go to open place to get killed?

Would it be a good (or impossible) idea to make infantry to run until they reach cover, exhaust or die if they take considerable fire in areas with poor cover e.g. crossing a street?

And soldiers would refuse to move if fired at when in cover?

How about something like this.

When in poor cover:

1) Run acconding to the given waypoints

- Cautious, shaken units

2) Run to the nearest cover that is not in enemy direction.

- Pinned, panic units

3) Run straight back to home

- Broken, Routed units

When in decent cover:

1) Move acconding to the given waypoints with considerable delay especially towards enemy

- Cautious, shaken units

2) High propability to refuse movement orders and if unit moves it runs to the nearest cover that is not in enemy direction.

- Pinned, panic units

3) Run straight back to home or stay until dead/captured or pressure dimishes considerably.

- Broken, Routed units

[Terrapin]

I think CK's recommendations would cause a situation where shooting at troops in an open field would cause everyone to immediately run adjacent to the incoming fire, and would look almost exactly like grouses trying to take flight after a hunter shoots a shotgun in the air. It wouldn't look realistic, that's for sure.

In city combat, it would mean that the streets would immediately vacate and be ghostly empty, as soon as someone shot down the street.

[CK]

Originally posted by CK:

if they take considerable fire in areas with poor cover

I meant incoming fire that exceeds the protective value of unit's current cover. Fire against the unit, not just a lonely ricochet.

How many shots do you think must be shot before an unit seeks for cover when walking on a paved road? One, two, a clip?

I bet for one shot, except for early russian troops which walked head high no matter what

[Coffin 'Enry]

Originally posted by CK:

I bet for one shot,

I think its a lot more complex than that. G.M. FRASER's recollections in "Quartered Safe Out Here" contain some very interesting examples of infantry behaviour under fire. OK its a different campaign and a different army but human nature does not change that much.

Cheers

[CK]

So, the current behavior of infantry in CMBB is very close to what we would typically see in a real combat?

I still agree that all kinds of things happen in combat, but the issue here is (?) the average behavior of troops. Which is in my opinion very unrealistic especially in case of very poor cover and enemy beeing close.

My quess for the infantry's "stupidity" is that the soldiers do not "see" the surroundings.

When they get shot at enough they are programmed to search for cover no matter where they are or what is near them?

Then we end up with a veteran unit walking in a paved street..it gets shot at from a second floor distance>150..soldiers start looking for cover that does not exist until half of them get slaughtered..they panic..and finally they run five meters to heavy building to get cover.

This happens every time.