StuGs and bazookas and AI trigger happiness

[Guest R Cunningham]

Others have already commented on this, but it seems that the Tac AI favors long-range bazooka shots over short range ones. In games where I am playing the Germans in the Last Defense scenario, I get bazooka attacks at ranges of 170+ meters. All of these are pentrating hits on the StuGs. I ran some tests today using the hotseat option against myself to set up specific situations.

I had a bazooka shooting at a StuG 33m away and missed with 7 of the 8 shots he had. The other one was a non penetrating hit on the upper hull. The bazooka was shooting at the side of the StuG (with schürzen).

I also had another bazooka shooting at the same StuG earlier and at closer range. It hit on the front upper hull and did no damage. The next round missed and then the bazooka panicked and ran away getting cut down in the process.

After those two tests. I drove the StuG over to the third bazooka ambush position where it was engaged from the flank at 39m. Again several misses and a non penetrating hit before the final kill.

As a side note I had ordered the StuG to area fire at a location to ensure that it did not try to shoot at the bazooka. The Tac AI did not change my targeting order.

Anyway what this all leads up to is a question about the penetration model and the ballistic flight of the bazooka rockets. I was wondering if the reason the AI prefers to take long range shots is because it knows that the round falling from higher arcs have better penetration chnces because of a lower angle of incidence with the StuG’s armor? Players, it seems, want short range ambushes to ensure hits, but the AI wants to take long range shots for both self preservation and greater kill chances when hits are scored. Prior to today’s tests, I never had a StuG survive a bazooka hit. During the tests, my StuG took four hits to kill.

Comments?

[Fionn]

Well all I can say is you were amazingly lucky. I've NEVER had a vehicle survive more than two zook hits.

------------------

___________

Fionn Kelly

Manager of Historical Research,

The Gamers Net - Gaming for Gamers

[Guest R Cunningham]

Fionn,

Was it luck or is there something here? I lose StuGs to bazooka hits all the time, but the AI always opens fire at a range that no human player would. I haven't experimented with it yet, but I think a comparison of the feedback using the LOS tool would shed some light. If I'm never gooing to get a very high hit percentage (30-40% tops) AND I get some sort of better indication for a kill at long range then I can see why the AI does this bazooka sniping. These survived hits I've got all happened at close range. I think this might be another of those "unlearning" things. Soldiers are trained, and wargamers have been trained to hold fire until the range is just right to maximize success. The AI was never trained this way. It just compares courses of action and chooses the one that meets its criteria based on some algorithm.

So I think that there has to be something that makes the AI bazooka (and maybe panzerschreck) WANT to shoot at long range and I'm thinking it has to do with the ballistic and armor modelling. But it could be something as simple as the bazooka morale wavering when the target vehicle is close. But to date, I've never had a StuG survive a hit from a bazooka shot from over 100m (mostly front but at least one flank shot in there)and yesterday I had one StuG take 4 hits to KO and the longest shot was 39m (killed on flank, 1 hit on front, other two on side upper hull).

[Fionn]

Well, a long range shot would probably have a much greater chance of passing OVER the schuerzen whereas a short range shot would hit them.

Schuerzen were designed to make tanks survive zook hits.

Perhaps that's why the AI does this..

------------------

___________

Fionn Kelly

Manager of Historical Research,

The Gamers Net - Gaming for Gamers

[Guest R Cunningham]

I thought schuerzen might have someting to do with this but most of these kills were frontal shots. There was one time when the AI attacked my StuG with a bazooka at a range of about 160m. I noticed this near the end of the replay. It was a miss. I ordered the StuG to reverse to get out of the range of the bazooka. It got about another 5 meters before another hail mary hit it. Saturday I had a StuG moving fast hit at a range of 173m from a bazooka. It was knocked out. The Stug was on the hill beside the wheatfield while the bazooka was in the scattered trees between the road and the wheatfield. No human player would have taken that shot.

As has been pointed out, the AI sometimes makes better decisions on targeting than the human player. I would like to see the "to hit" and "to kill" tables that the AI is basing its decisions on. I think this might be a consequence of the modelling that has gotten the AI to develop its own tactics that conflict with what the real world troopers would have done. Perhaps it really was a better idea to take long range shots, but I think most players would want to get a higher hit chance knowing that HEAT penetration is not range dependent. I'm thinking of some simple formula like:

To Kill = (To Hit)x(penetration).

So if penetration is constant, then only the to hit factor affects the decision. One would think that the to hit chance would improve with shorter range. Either the threshold value for a bazooka attack is set low or there is some other factor like team survival or increased chance of penetration based on impact angle that influences the outcome.

[Guest Scott Clinton]

I have also seen the AI take some LONG shots...when I played the US and checked out these same shoots the game told me my chances to hit were anywhere from 2% to 5%.

Not worth the ammo IMO...but your AI seems to be actually HITTING these shots. Mine is not, go figure?

------------------

The Grumbling Grognard

[Herr Oberst]

Some time ago, Fionn posted a good set of words on (essentially) angles of incidence and survivability. Maybe the long range targeting does improve the angle of impact, I'll need to hotseat it to check this out.

For the long shot mentioned earlier, from woods to the Stug, what was the angle like on that one? If the Stug is coming downhill, or is canted sideways, it could be presenting a broad flat surface of the upper hull for attack.

In general, I try to keep the bazookas away unless it is a side or rear shot against a buttoned AFV. Otherwise, those US bazooka teams tend to wet their pants and (unwisely) flee. This just makes them arcade ducks when I am playing the Germans. Anyone available will be targeted at the bazooka team, at least for 1 turn.

[This message has been edited by Herr Oberst (edited 11-08-99).]

[Guest Mikeman]

I too have noticed that long range bazooka shots are more effective than short range shots against the StuGs. I thought it a bit strange, but just figured that was the way it was. (Complete trust in BTS modeling)

Mr. Cunningham brings up an interesting thought. Could the TacAI actually be making targeting decisions that are better than a good bazooka man would make based on it's intimate 'knowledge' of ballistics and armor thickness, etc.? No soldier on the battlefield would be able to call up this information. A soldier would only have training and practical experience to go by when deciding whether to engage a StuG at 150 meters or let it come in to 30 meters. What's the real life truth on this?

Did the good bazooka man wait for close range shots, or take the long ones against StuGs? Which type of shot did he think was most effective?

Mr Cunningham's test clearly shows that long range bazooka shots (at least against StuGs) are preferable in CM. The TacAI 'agrees' with this principle. The big question for me is: What did the real life WWII bazooka man believe to be true? I'd bet they preferred close range shots (as far as being effective). That preference was probably just based on their chance to hit.

Has CM shed new light on how the bazooka could have most effectively been employed against a StuG? A fact unknown to GIs at the time?

Mikeman out.

[Ari Maenpaa]

To me the bazookas seem to be a bit overpowered. Last night my bazooka man killed a Tiger with his first shot at range of 150m and from the front side. A top hull penetration was reported.

Is this even remotely realistic or was that an extremely lucky shot?

Needless to say that my hot-seat opponent was floored, but I was also puzzled by myself.

[Richard]

Yes can we have a discussion about the Bazookas please. I haven't done research but they also seem overpowered to me, killing tanks at ranges they should not be able to.

Or so it just "seems" to me.

------------------

Richard Arnesen

The Wargamer

http://www.wargamer.com

[Paladin]

If it were YOU manning the bazooka, what would you do?

I've never been under fire, but if I were sitting in some trees and saw an enemy tank coming toward me, I'd probably hide and wait for it to pass, then fire at its rear or flank at close range as many times as possible until it turned to fire at me. (Two or three shots?)

However, if I saw enemy infantry supporting the tank, I might very well just take a long-range potshot and get my butt out of there. If I waited for a great shot in this circumstance, I'd doubtless get killed, right? So, better to take a slim long-range chance and live to fight again another day. (Or am I just a lousy soldier?)

What would you do?

[Guest Mikeman]

I like the "What would you do?" approach to these questions. If I was well hidden from an approaching tank and no enemy infantry was in the area I would wait for him to pass and give him the rear shot. I might pee my pants, but I'd wait. Why? I'm sure to have a better chance of killing the tank with a close range rear shot. I wouldn't reason it out like this: "Well now, he's 150 meters out which is a little far, but he's canted at a somewhat steep compound angle towards me so his thinner top armor is somewhat more exposed, so an accurate shot....

I may be wrong, but I wouldn't reason like this in the heat of battle. I'd want to plug him at 20 meters right in the butt if possible (no enemy infantry).

Mikeman out.

[Guest Charles]

I have also observed this long-range shooting (pot-shots) behavior from the zooks.

My question: should/does the guy manning the zook consider if he fires from long range he is giving away his position? If he is about to give away his position, thus drawing enemy fire, I would think he would decide to wait for a better shot before exposing himself. If I knew the enemy considered me (a bazooka team) to be a high priority target, I wouldn't want to draw any unnecessary attention to me unless I was sure of a good chance of a kill.

Great demo, BTW! Very impressive. Really looking forward to best wargame of the year!

[Guest Scott Clinton]

I just don't see the bazooka as being to 'powerful'. Maybe it is just my experience so far but they just are not very accurate in my games at all.

I think perhaps you guys are looking at individual HITS and the kills that result from these hits and not considering the number of shots that it took to actually achieve these hits. Like the example above by Ari Maenpaa: I don't know, but how many shots did it take for the bazooka team(s) to get that hit?

As for why the AI is actually taking these long range shots (when I don't because they never hit) I think it may be looking at its chance to KILL more than its chance to hit.

Also remember that the actual trajectory of the round is tracked by the game. So a side shot on a Stug at long range may actually have a better chance overall to stop that tank because it can 'arch over' the schuzern, whereas a shorter ranged shot would have too flat a trajectory to do this. Perhaps this seems archaic to us gamers but it would be fairly obvious to anyone that fired the weapons and watched it arch through the air like a duck…

Once a bazooka hits its target the range to the target is pretty much irrelevant in determining penetration for a bazooka and any HEAT warhead. A hit on the TOP of a Tigers armor (for example) would be an easy penetration of a bazooka at 20m or at 150m. It is actually HITTING the Tiger (top armor) that is the hard part.

I guess, until I actually see these weapons HITTING (not penetrating) their targets more I think they are pretty much in line with what I would expect.

I think I will try a little hot seat tonight...

------------------

The Grumbling Grognard

[Guest R Cunningham]

That Tiger top hull kill is precisely the kind of thing I think the AI is looking for. A short range shot would have a lower arc and a more direct trajectory and would be more likely to hit the hull side. The long range shot means a higher arc and a highly parabolic trajectory that would allow for top armor hits (at screwy angles). This is what I think is driving the AIs calculation to engage at longer ranges.

On the what would you do angle, if I were a bazooka man I would try to ensure that when I exposed myself I was doing it when I though I had the maximum chance of success which I have, traditionally at least, associated with the short-range flank or rear ambush shot. In the case I use in an earlier post where my StuG was killed at 173m while moving fast, that same bazooka team was killed by one main gun round from the 88 that followed the StuG.

But I don't think the bazooka is represented too powerfully directly, though there may be a flaw in the ballistics somehow. As a cheap comparison, bazookas in CC could never hit at these ranges and when it did it hit at closer ranges couldn't penetrate the StuG from the front. But CC didn't track ballistic flight paths. So do these trajectories make that much of a difference?

[Guest Scott Clinton]

R Cunningham:

I see we think alike on this issue

I think perhaps the AI needs to be tweaked a little for bazookas to look more at its chance to hit and less at its chance to kill once a hit is obtained.

R Cunningham: "So do these trajectories make that much of a difference?"

Well, we will know for sure when they release the full version with the PIAT!

------------------

The Grumbling Grognard

[Fionn]

War is hell AND a lot of weird things can happen.

If you have a 3% chance of killing a Tiger at 200 metres and pop off ten shots a game at that range over three games well, then you'd expect to kill a Tiger once every three games.

Also, people playing the game are forgetting that the units they are playing with are generally green or regular. there's a few vet units in there but even those can be rattled by close-by fire..

Zooks could hit out to 200 metres even though it is rare. I would caution players not to see a hit at 200 metres in one game and simply forget that it CAN happen but is rare.

Also, I presume we all know that HEAT warheads are chemical energy penetrators and NOT kinetic energy penetrators and as such their penetration is independent of range to target.

So remember that a zook hit IS a kill 90% of the time and as such the AI is looking at tanks which it feels aren't going to get any closer and decides to take a chance.

Remember that the proximity of infantry also plays a role as does the experience and whether or not the zook unit is rattled.

And lastly, the beta demo you got is old.. I got a new build this morning and in a US infantry assault on a platoon-sized position backed by two of my Pz IVs the AI maneuvred its zooks in close.

Two zook teams were present. The AI fired one speculative shot at 190 metres, then it rushed the zook teams up towards the ridge.. One stopped about 120 metres from my position in open terrain and killed a Pz IV while the other zook team was run up through covering terrain into some scattered trees from which it took a pot shot at the other Pz IV from about 60 metres. This team did NOT fire from long range.

So, remember, you're discussing the AI as it exists in the beta demo, not as it exists now..

From what I see its already been tweaked.

------------------

___________

Fionn Kelly

Manager of Historical Research,

The Gamers Net - Gaming for Gamers

[Richard]

Okay stupid me forgot about the range not being a factor thingy . You'll have to excuse me but I am used to games where bazookas are useless and if the enemy has a tank and you don't you are DEAD (and Fionn you KNOW what game I am talking about ).

That makes more sense now and will DRAMATICALLY affect my tactics (hehe poor PBEM opponents ).

------------------

Richard Arnesen

The Wargamer

http://www.wargamer.com

[Fionn]

Chuckle... Oh yeah the infamous "uberpanzer" phenomenon

Actually just to show up the moronic AI in the Eastern Front version of that game I played the demo of the 3rd iteration of that game as the Soviets and bought ONLY flamethrower infantry squads. No tanks, no MGs, no conventional infantry.

I toasted a total of 7 German tanks and all the German infantry and managed to take all the Victory Locations.

Being able to do that with 15 flamethrower units (which have some serious tactical flaws which means they MUST be part of a balanced force to be effective) says something BAD about that AI.

------------------

___________

Fionn Kelly

Manager of Historical Research,

The Gamers Net - Gaming for Gamers

[tss]

Actually just to show up the moronic AI in the Eastern Front version of that game [snip]

I guess it is the same game in which I was playing on defensive. I set up my troops, started game, went to shower, came back, and saw that the enemy had routed. Then I deployed again for the next battle, started game, went to eat breakfast, came back, and I was victorious once again.

-Tommi

[Guest JonS]

Umm, Fionn, of COURSE we're talking about the 'old' Beta Demo - what else could it be? In general the points people are bringing up are what concern them about the B-D, and are concerned will be the same in the final release. In the absence of any other info what else are we to think? You have access to later versions of the game, and probably a lot more inside info than the rest of us, so seem to take for granted changes that the rest of us can only hope for have been, or will be, included. Telling us, the ignorant masses, which changes have already been included is exactly the kind of info I'm (we're?) after.

Oh, and it makes me/us terrifically jealous too...

JonS

------------------

Quo Fas et Vino du Femme

[Bullethead]

I too have seen bazookas open fire at VERY long range, 150m or more. The zookmen seem to open fire whenever they see a target, regardless of range. I have yet to see a hit at any range other than point blank, however.

In a recent run of Last Defense as the US, the 3 zookdudes survived long enough to 21 rockets. Of the 21 shots, 6 were at less than 50m and the other 15 were at 150m or more. The only reason any were fired at short range was due targets suddenly appearing around house corners or over crests.

Of these 21 shots, only 1 hit. This was a zookdude's 2nd shot at 40m at at StuG. The other short-range shots were 4 misses in a row at a track at 30-35m.

-Bullethead

[Fionn]

Well, I've played a few more games and zooks DO seem to open fire at closer ranges...

Will have to play more to be sure though

Fortunately there's enough scenarios to "test"

------------------

___________

Fionn Kelly

Manager of Historical Research,

The Gamers Net - Gaming for Gamers

[Guest L Tankersley]

Ah, this looks like the proper place for this question - I was wondering about the effects of schuerzen on HEAT weapons like bazookas as modeled in CM. Since hit location isn't tracked to particular polygons, is there a "percentage of coverage" figure or something similar used to determine if a round impacts on the skirts?

Also, what is the game (or real-life, for that matter) effect of an impact on the skirt? How effective were schuerzen at defeating bazooka rounds?

Leland J. Tankersley

[Guest R Cunningham]

Not too sure about how Schürzen are handled in CM. I did a few tests and didn't get any changes in the quantized kill probabilities when the target was without Schürzen vs one that had them. (this was done with the StuGs in last defense) There is a lot of room for error in this, because the game is not offering "32.5% kill" but only Excellent, OK, low etc.

As regards the skirts themselves, their fucntion was merely to detonate the HEAT warhead further away from the main armor. Most HEAT warheads have some sort of cone that is designed to detonate the round at the optimum distance from the armor for the best jet creation and thus best armor penetration. Skirts weren't armor though, and the Pz IV J went with a wire mesh skirt that did just as well and saved weight. I've seen a picture of a skirt after a bazooka hit and it looks pretty bad.